JP2016538931A - Classification method and apparatus for seizure type and severity using electromyography - Google Patents

Abstract

A method and apparatus for monitoring a patient for seizure activity comprising collecting and processing EMG signal data and categorizing the detected data to perform risk stratification. An individually adapted transmission protocol for the detected event can be selected and executed.

Description

  The present invention relates to a method and apparatus for classifying seizure types and severity using electromyography.

  Seizures can be characterized as abnormal or excessive synchronous activity in the brain. At the start of a seizure, neurons in the brain may begin to fire in certain locations. As the seizure progresses, the firing of this neuron spreads throughout the brain, and in some cases, many areas of the brain can be involved in this activity. Seizure activity in the brain can cause the brain to send electrical signals to various muscles through the peripheral nervous system, activating the muscles and causing the redistribution of ions within the muscle fibers. Electromyography (EMG) can be configured to measure potential changes resulting from ion flow during muscle activity by placing electrodes on or near the skin.

  Techniques designed to investigate and monitor seizures are usually electroencephalography (EEG) characterized by electrical signals that use electrodes attached to the scalp or head of individuals who are prone to seizures or who have seizures. : Electroencephalography). Detection of epileptic seizures using electroencephalography (EEG) typically requires the use of an amplifier with a number of electrodes and associated wires attached to the head and monitoring brain wave activity. Multiple EEG electrodes are very cumbersome and usually require some technical expertise for application and monitoring. Seizure confirmation usually requires observation in an environment equipped with a video monitor and video recording equipment. Moreover, when measuring brain activity with EEG, not all measured seizure activity or activities related to seizures actually show symptoms as high-risk events. In addition, EEG data without video validation is suitable for grading or differentiating certain types of seizures, including those with minor or minimal caution, from other higher-risk seizures. There may not be.

  Unless used in a staffed clinical environment, EEG devices are often not only intended to determine whether an episode is ongoing, but only provide a post-occurrence episode history. In addition, these devices are typically designed for hospital-like environments where seizures can be confirmed by video camera recording or caregiver observation, and are typically intensive, such as when hospitalized for patients who have had multiple seizures. Used as part of a treatment regime. Hospitalization may be required for diagnosis or to stabilize the patient until appropriate medication is given. At discharge, the patient is returned home with little subsequent monitoring. However, at any time after returning home, the patient may again be struck by a fatal seizure.

  In some cases, the patient should be monitored at home for some time in preparation for another seizure. Seizures with motor symptoms may have a pattern of muscle activity that includes rhythmic contractions of some, most, or all of the muscles of the human body. Seizures can lead to, for example, sudden death of epilepsy (SUDEP). The root cause of SUDEP is still not well understood, but severe central nervous system depressant effects may occur following seizures. A central nervous system inhibitory action is followed by an increase in respiratory rate and a decrease in the respiratory cycle, which can lead to arrhythmias and death. However, not all seizures cause or are associated with SUDEP with the same likelihood, and some patients may not have a significant risk of SUDEP even if some seizure activity is present. It can also be difficult to selectively identify seizure activity that may be most at risk unless it is differentiated by type, severity, or more detailed classification.

  Currently, there are portable devices for seizure diagnosis, but they are EEG-based and are usually not designed or suitable for long-term home use or everyday wear. Other seizure alert systems can operate by detecting movement of the human body, usually the extremities. Typically, such a system can operate on the assumption that a person moves violently during a seizure. However, depending on the type of seizure, this assumption may or may not apply. Electrical signals sent from the brain during a seizure are often transmitted to many muscles at the same time, and the muscles can compete with each other to effectively cancel out intense exercise. In other words, the muscles may function to stiffen the person rather than cause actual intense exercise. As such, seizures may not be consistently detected by accelerometer-based detectors.

  Therefore, it can be used in non-hospital or hospitalized environment without using many electrodes on complicated head or limbs, and accurately detect seizure events with motor symptoms, but to intense movement What is needed is a method and apparatus for detecting epileptic seizures that is not limited to the response of. Furthermore, there remains a need for caregivers to detect epileptic seizures that differentiate mild motor symptoms that do not require emergency response from other types of seizures, including seizures that require emergency intervention. Yes.

  A method of monitoring a patient for seizure activity includes the steps of collecting EMG signal data, processing the collected EMG signal data to determine whether a detected event exists, and risk stratification And categorizing the detected events, including, for example, selecting a transmission protocol included in the group of selectable transmission protocols based on categorization and risk assessment. Once a transmission protocol is selected, an appropriate transmission, such as an alarm or warning message, can be sent to the caregiver and / or designated person.

1 illustrates one embodiment of a seizure detection system. 1 illustrates one embodiment of a detection unit for a seizure detection system. 1 illustrates one embodiment of a base station. FIG. 6 illustrates one embodiment of a method for monitoring a patient for seizure related activity and selecting a protocol for alarm transmission. FIG. 4 illustrates another embodiment of a method for monitoring a patient for seizure related activity. FIG. 4 illustrates one embodiment of a routine for analyzing EMG signals for seizure activity. FIG. 6 illustrates one embodiment of a routine for analyzing EMG signals for seizure activity based on detection of signal bursts. FIG. 6 illustrates another embodiment of a routine for analyzing EMG signals for seizure activity based on detection of signal bursts. FIG. 4 illustrates one embodiment of a routine for analyzing an EMG signal for seizure activity based on detection of a burst train. FIG. 6 illustrates one embodiment of a routine for analyzing EMG signals for seizure activity based on detection of burst periodicity. FIG. 6 illustrates another embodiment of a method for monitoring a patient for seizure related activity and selecting a protocol for alarm transmission. The EMG signal data is illustrated.

  The devices and methods described herein can be used to detect seizures and to alert caregivers in a timely manner about events related to seizures. The device can include a sensor attached to the patient or the patient's clothing. The device can also be configured to measure the electrical activity of the muscle using electromyography (EMG). Detection of seizures using EMG electrodes is described, for example, in Applicant's US patent applications 13 / 275,309 and 13 / 542,596, and Applicant's US provisional patent application. 61 / 875,429, 61 / 894,793, 61 / 969,660, and 61 / 979,225, and these The disclosure of each of which is fully incorporated herein by reference. The devices and methods described herein use EMG to monitor patients for muscle electrical activity, detect possible seizure events, and detect detected events as risks, types, and / or severity. May be used to stratify based on For example, if a detected event is considered to be present, including a seizure of a given type or severity, the surveillance system selects a specific transmission protocol to alert one or more caregivers can do. For example, the transmission protocol can include sending either alarm messages and / or EMG signal data to one or more caregivers or other designated individuals over the network.

  In some embodiments, the transmitted EMG data may be organized to facilitate confirmation or review of detected events. For example, in response to detecting an event, information may be sent to the caregiver to facilitate scanning and reviewing time domain and / or frequency domain EMG data. In addition to the raw signal data, other information related to the analyzed EMG signal data may be transmitted. For example, in some embodiments, one or more patterns typical of abnormal muscle movement may be identified from noisy data and the particular pattern identified may be communicated to the caregiver. For example, other statistical data related to the detected event may be transmitted to the caregiver, including statistical data related to the detection of qualified peak data. In some embodiments, such information may be transmitted and / or presented to the user based on preferences set by the system or user.

  In some embodiments, risk stratification can facilitate transmission of either or both data richer information, such as alarm messages and / or time domain and / or frequency domain EMG data. . Furthermore, stratification may facilitate selection of a transmission protocol that minimizes power consumption. Also, in some embodiments, some detected events may be considered reasonable to convey to the remote user as safely ignored or requiring only a warning alarm condition. . For example, it can be considered that the risk of SUDEP resulting from a detected event, the risk of injury due to falls, and / or the risk of other things is very low. At this time, the caregiver receives, for example, a message that an event has been detected but does not require an emergency response and / or that the event may be logged into a searchable database for later review. be able to.

  Data transmitted from the monitoring system may be customized for a particular individual or recipient group in some embodiments. For example, the transmitted data may include alarm messages, a subset of statistical information related to algorithm detection, time domain or frequency domain EMG data, other data, and / or combinations thereof. Also, such information may be useful for a particular subset of data recipients, but sending such information to other recipients may not be useful (or rather harmful). For example, patient related alarm information, including some sensor data, may be sent to an emergency medical technician (EMT), who has received appropriate training to interpret all EMG data. There may not be. Also, sending such data can be a burden on the caregiver and / or confuse the caregiver during an emergency response. However, other caregivers, such as the epileptic patient's physician, or a remote individual trained to more fully interpret EMG signal data may, for example, reconstruct the time dependence of the collected EMG signal. The range of available data may be sent in a wider range, such as information suitable for, or information suitable for evaluating the output of the algorithm used to identify one or more patterns of muscle activity.

  The data sent from the monitoring system can be related to the status of the detected event, for example, whether the detected event was classified as an emergency event or a warning event. In addition to caregivers, further organize the data for transmission to any group of selected or designated individuals, including any number of other individuals such as selected friends and family Also good.

  In some embodiments, the monitoring system can send data to a remote database or server. The designated individual may have access to the data contained therein, or specific or limited portions of the data stored therein. In some embodiments, an individual can log on to a remote database to view the data. Information can also be sent to individuals from these databases. Thus, in some embodiments, data can be presented directly to an individual from a patient device (eg, a detection unit or base station), from a remote database, or both sources.

  In some embodiments, the methods described herein can be identified, but can detect and classify mild seizures or other events that do not require an emergency response. Such classification may be determined, for example, by detecting seizure features that selectively appear in either the tonic and / or clonic phase of the seizure. Further, in some embodiments, the classification includes analysis of temporal relationships between seizure phases and / or attributes of detected EMG signals collected during an intermediate period between detected phases. You may go out. For example, in some embodiments, the classification may include whether an tonic phase and clonic phase of seizures are detected, respectively, and whether those phases are present continuously, for example, activity between those phases. A determination of whether there is a decline period may be included. Also, in some embodiments, risk stratification may further include additional sensor data and / or other data, as further described herein. For example, in some embodiments, risk stratification includes analysis of additional data, such as may be added from one or more orientation sensors, position sensors, oxygen saturation sensors, or pulse oximeter sensors. May be.

  The system described herein is suitable for patient monitoring in an outpatient environment and includes one or more EMG sensors that can be coupled to the skin on or near the muscle of one or more of the patient's muscles. May be included. Although the EMG signal may be collected substantially continuously, it may be desirable to send only a subset of the collected EMG signal or only alert the caregiver. For example, particularly for portable detection devices, the power consumption for sending signal data is quite large, so it may be desirable to limit the amount of collected signal transmitted over the network. To achieve these goals, the risk stratification of detected events described herein, a feature clearly lacking in other monitoring systems, can be used.

  In some embodiments, a threshold suitable for detecting mild seizure events can be set. Also, these settings can be used without the risk of the surveillance system detecting an inordinate number of false positives. For example, thresholds suitable for identifying events related to mild seizures can be set to identify those events, but the detected events are automatically classified so only a subset of detected events is selected The appropriate transmission protocol can automatically initiate an emergency response. Thus, the system can still alert the caregiver of the presence of those events and / or link those events to a searchable database, but limit the detection of inappropriate emergency responses or false positives It is possible.

  EMG may be ideally suited for this purpose for a number of reasons. For example, a considerable amount of information is available from data collected by EEG, but electrical signals in the brain are not necessarily correlated with true seizures or seizures of a given type or risk Not exclusively. In addition, by looking at the muscular motor symptoms of brain activity, a more accurate route can be provided for classification of seizures by severity and type. Also, the EMG described herein can be used to selectively identify different parts of seizure activity. For example, the increase in EMG signal may be either temporary or persistent. It is also described, for example, in Applicant's US Provisional Patent Application No. 61 / 969,660, by selecting specific detection settings based on, for example, the width of the detected signal or other factors. As such, the detection routine can be configured to be selective for a particular portion of seizure activity. Importantly, because various types of seizures can be detected, seizure data may be risk stratified based on whether the portion most likely to require a particular response has been detected.

  Individually adapted responses can be performed on detected events based on the characteristics of the detected events. For example, by understanding whether sensor signals can be related to tonic-clonic seizures, tonic seizures only, clonic seizures only, or other types of seizures, caregivers can better detect detected events. You will be able to evaluate and plan an appropriate response. In addition, certain seizures may be short-lived and / or lack of characteristic signs of more severe seizures, such as repetitive movements that may occur during the seizure phase. is there. At least some of these seizures are detected as an increase in the strength of the EMG signal, or can be detected using other more sophisticated algorithms or devices, but such seizures are detected, Although alarms can be triggered, they may only indicate a limited or minor risk of injury for some patients. For example, some detected events typically do not pose a significant risk of adverse effects, including SUDEP, when a seizure occurs. Signaling an unnecessary and costly alarm in response to an imminent event, where such detection takes place without further classification, is the only way to respond to a potentially dangerous event Can be. The methods described herein process data and adapt it individually to patient monitoring, for example by estimating whether each detected event poses a significant risk of an adverse seizure. Facilitating strategies that improve efficiency can alleviate such concerns.

  In some embodiments, the methods described herein can classify detected seizures based on the patient's seizure profile or patient demographic seizure profile. For example, as a non-limiting example, classifying detected seizures based on various metrics such as type, intensity, seizure duration, seizure phase duration, other metrics, and combinations thereof. it can. Classification of seizure severity may include, for example, normalizing the seizure metric to a typical value for a single patient or patient demographic. For example, for a patient, the feature quantity detected is a factor of a previously measured value for the feature (eg, during another episode of the patient), or the feature These factors can be used to grade the severity of seizures if the mean value of is a factor. For example, a particular seizure may be detected and the size of the detected feature may be only 50% (or some other factor) of other seizures detected for the patient. Such information can be sent, for example, to a caregiver and / or otherwise used to determine an appropriate response. For example, it can be seen that the patient typically has several mild seizures and the risk of adverse effects of those seizures may be low for the patient. Also, at least some of the detected events can be safely ignored or may be ignored in some circumstances. For example, if the patient has only a mild seizure and the patient is known to be at bed then, both the risk of SUDEP and the risk of falling may be low. Also, in some embodiments, at least some of the detected events may be ignored or logged as events that may not simply be an emergency response.

  Along with or in addition to the initiation of an alarm, the devices and methods described herein can be used to create a seizure event log to help medically or surgically manage a patient. Also good. Events can be categorized to help organize events related to detected or possible seizures. For example, seizure events can be used with automatic classification (eg, based on type and / or severity), especially if the event cannot be confirmed on video or by a trained specialist such as a physician If an individual review of a set of data is inconvenient or would be prohibitively expensive, an organized database of data related to seizures can be created.

  In some embodiments, the methods described herein may include identification of EMG signal regions, including signals processed with increased amplitude, and regions that are peaks (eg, processed signals A region where the signal amplitude including the amplitude changes from rising to falling may be further identified. Peaks that go from rising to falling can be identified because they include regions where the signal amplitude that exists only for a limited time increases. The identified peak is qualified for one or more characteristics typically present in the seizure phase as described further herein, and detects the seizure phase. Qualification can be made to increase selectivity. For example, peaks can be compared against one or more characteristics of EMG signal data from one or more patients experiencing seizure phase. Alternatively, it is qualified as being similar to the characteristics and / or changes of the aforementioned mesophase, compared to the changes that occur during the physiological transformation to the mesophase. In this disclosure, such qualified peaks may be referred to as “clonic-phase bursts”. The presence of a critical level of clonic burst activity can be used, for example, to detect the presence of clonic secular activity.

  In some embodiments, the methods described herein may include identification of EMG signal regions, including signals processed with increased amplitude, and an initial group that is peaks (eg, processed A region where the signal amplitude including the signal amplitude changes from rising to falling may be further identified. The identified set of peaks is then qualified as one that can be used to determine the presence of clonic bursts. Amplitude may refer to either the strength of the signal or the absolute value of the strength as may be appropriate for a given calculation and / or signal format. The collected signal can be rectified, for example, and the amplitude of the EMG signal can refer to the strength of the rectified signal from the EMG sensor. In some embodiments, the EMG signal can be processed to separate one or more frequency bands, so that the amplitude of the signal refers to the strength of the separated signal for one or more frequency bands. Or the magnitude of a statistic related to the momentum level, which may refer to the magnitude of the statistic processed from a signal separated in yet another frequency band. For example, in some embodiments, the statistic may be a T-square statistic related to the momentum level.

  The procedure for determining the group of peaks is described herein, but for example, the applicant's US Patent Application No. 61 / 875,429, which claims priority. It is also described in the specification of 13 / 275,309. In summary, in some embodiments, a peak detection program can be executed to identify portions of EMG signal data that include one or more peaks. Peak identification may include procedures that may include, for example, peak fall and / or rise detection, i.e., searching for portions of EMG data or portions of smoothed EMG data where the curvature of the data is changing. Good. For example, inflection points or other critical points in the data set can be identified and used to identify the presence of one or more peaks.

  Qualification may then include identification or selection of peaks that meet one or more criteria. For example, it is possible to select a peak that satisfies a criterion for enhancing the reliability of appropriately having a pattern in which the peak indicates the activity of the clown phase. For example, the peak can be qualified as a clonic burst. At a high level, the procedure for qualifying as having peak data containing one or more mesomorphic bursts is to compare various peak characteristics to one or more qualification thresholds. May be included. For example, for one peak, if one or more peak characteristic values related to clonic activity meet one or more eligibility thresholds, eligibility criteria are met. This peak can then be referred to as a clonic burst.

  Several qualification procedures can be run for individual peaks. That is, certain characteristics of the peak, such as peak height, area, or time width, can be defined without including data from other peaks. Thus, individual values for that characteristic can be calculated for each peak in the group. Other characteristics can be calculated for more than one peak, as described below. The characteristics of individual peaks include, for example, peak height, peak area, signal-to-noise ratio (SNR) (eg, non-intensity in peak amplitude as can be measured or evaluated from the background region). The ratio of the peak amplitude to the assessment of certainty), the duration, the duration of the intervening period during which the signal is reduced on both sides of the peak, other characteristics of the individual peaks, and combinations thereof may be included.

  In some embodiments of the methods described herein, the peaks identified in the initial group of peaks (eg, the set prior to qualification) are respectively the minimum duration, maximum duration, minimum Signal to noise ratio (SNR), one or more quiescent periods on either side of the peak, or minimum duration of the intervening period, maximum duration of one or more intervening periods on either side of the peak, and / or combinations thereof Can be compared against a qualification threshold selected from a group of qualification thresholds including. The intervening period can be defined by the length of the duration of the region where the signal is stable or low amplitude (eg, low signal variability, RMS noise or signal strength), for example , Which can be indicated by the distance between the peak edge and a nearby area where the signal strength increases or the stability of the signal decreases. In some embodiments, the peak signal-to-noise ratio may be calculated using the peak amplitude data and signal noise estimation or calculation. Noise can be, for example, in measurements of amplitude, height, or area of a signal that can result from variations or uncertainties in the baseline signal (eg, variations in EMG data for regions not related to peak activity of interest). The level of uncertainty) can be determined by calculating or estimating this, for example, from data collected on both sides of the peak or separately from an EMG signal, such as the portion where the patient is resting The determined part can be determined. For example, to calculate noise, signals may be collected and signal variability measured directly. Alternatively, noise can be, for example, an estimate of variability inferred from the amount of change typical of a signal of that strength, as predicted by, for example, the strength of the signal and one or more model functions, such as a normal distribution model function. Can be estimated from In some embodiments, an estimate of the amount of change or uncertainty in the baseline signal or noise can be selected or calculated during one or more system calibration routines.

  In some embodiments of peak qualification, the peak is met by meeting a threshold SNR, by meeting a minimum threshold of a peak activity duration of about 25 to about 75 milliseconds, and from about 250 milliseconds to about By satisfying the maximum threshold of 500 ms peak activity duration, it can be qualified as a clonic burst. In some embodiments, detecting that a sequence of substantially stationary signals from about 50 milliseconds to about 300 milliseconds intervenes to qualify the peak as a clonic burst. it can.

  Depending on the characteristics of the peak data, two or more peaks can be calculated. Also, in some embodiments described herein, a procedure for qualifying a clonic burst includes comparing a plurality of peaks with one or more qualification thresholds. Also good. That is, multiple peaks may be selected, an aggregate characteristic value determined for the multiple peaks, and the aggregate characteristic value compared to one or more associated threshold values.

  The qualification threshold associated with the characteristics of the peak group may be referred to as the aggregate qualification threshold. For example, aggregate qualification thresholds that can be used to qualify multiple peaks include peak repeat minimum and / or maximum rate and / or amount of time duration change between peaks There is a threshold value for.

  In some embodiments, multiple peaks are eligible for about 1 peak per second that is the threshold for the peak repeat minimum rate and about 7 peaks per second that is the threshold for the peak maximum repeat rate. Gender can be certified. In some embodiments, for example, a greater or lesser number of peaks than the boundary defined by the above threshold may be used to measure a suitable interval (e.g., multiple peaks as one peak rate). If present over the appropriate interval, it can be assumed that those peaks may not have been properly qualified.

  Included in the various metrics for characterizing the change in duration of time between peaks is also described in Applicant's related application US patent application Ser. No. 13 / 275,309. There is an average deviation rate. However, other metrics for characterizing peak timing variability, such as standard deviation values, average deviation values, or percent deviation values, are also described therein. Any of the aforementioned metrics for multiple peaks can be calculated and used as an aggregate characteristic value that can be compared to the aggregate characteristic thresholds described herein. In some embodiments, multiple peaks are qualified if the minimum average deviation rate value for time between peaks is greater than about 1% or about 5%. That is, in some embodiments, the aggregate characteristic threshold for the minimum average deviation rate may be between about 1% and about 5%. In some embodiments, if the maximum mean deviation rate value for time between peaks is less than about 40% or about 50%, multiple peaks are qualified as multiple mesomorphic bursts. can do. Routines for determining the amount of change in duration between peaks are described in even more detail in the applicant's various other co-pending applications, which are incorporated herein by reference.

  In some embodiments, the procedure for qualifying a peak includes an initial qualification step based on one or more of the above criteria (eg, criteria based on individual peaks) and the initial qualification. Removal of peaks that did not pass the qualification, and another qualification based on the calculation of one or more aggregate characteristic values for the remaining peaks (eg, all peaks that meet the initial qualification) A certification step. For example, identify peaks and remove some peaks from the overall qualification (eg, remove peaks because they are too narrow or too wide) and then remain If the peak data meets one or more aggregate threshold criteria as a whole, the remaining peaks can be qualified.

  Various systems are suitable for collecting large amounts of EMG and other patient related data, organizing such data for system optimization, and initiating alarms in response to suspicious seizures. FIG. 1 illustrates an exemplary embodiment of the system. In the embodiment of FIG. 1, the seizure detection system 10 can include a detection unit 12. The detection unit can be configured as a portable and wearable device placed (or worn) on or near any suitable muscle or group of muscles that are susceptible to motor symptoms during stroke. . In some embodiments, the system 10 may also include any of a variety of wireless local area network technologies. For example, the detection unit 12 may communicate wirelessly to the Internet using WiFi®, Bluetooth®, or through another local network. Also, in some embodiments, using a local network, the detection unit 12 can send data directly or via the intermediate base station 14 over the Internet. In some embodiments, the caregiver may be contacted directly through a local network such as WiFi. Base station 14 may be connected to the Internet wirelessly (eg, through a local network) or may be linked to the Internet through a hard connection. Also, in some embodiments, in addition to detection unit 12 or in addition to detection unit 12 and base station 14, system 10 may include, for example, acoustic sensor 8, video camera 9, alarm transceiver 16 or these elements. Any of the combinations may be included. The detection unit can include one or more EMG electrodes that can detect electrical signals from muscles at or near the patient's skin surface and deliver the electrical EMG signals to the processor for processing. . The EMG electrode can be coupled or attached to the patient, and in some embodiments may be implanted in the tissue in the vicinity of the patient's muscle that can be activated during the stroke. Implantable devices may be particularly suitable for some patients where the EMG signal can usually be weak, for example, a patient with a significant amount of adipose tissue. The base station receives and processes the EMG signal from the detection unit, the acoustic data from the acoustic sensor, and / or the data from other sensors and determines whether a seizure may have occurred from the processed signal. A computer can be provided that can determine and send an alert to the caregiver. The alarm transceiver 16 can be carried by the caregiver or placed in the vicinity of the caregiver to receive the alarm transmitted by the base station and relay it to the Internet. Included in system 10, including, for example, wireless devices 17, 18, stored database 19, electronics for detecting changes in integrity of the electrode-skin interface, and one or more environmental transceivers Other components that may be used include Applicant's U.S. Patent Application Nos. 13 / 275,309 and 13 / 542,596, and Applicant's U.S. Provisional Patent Application No. 61/894, 793 and 61 / 875,429.

  When using the device of FIG. 1, for example, a person 11 who is prone to having epileptic seizures can rest in bed or be in other places that can be involved in daily life, physically with the body It may have a detection unit 12 that contacts or is close to the body. The detection unit 12 can be a wireless device that can stand up and walk around without being constrained by a fixed power source or a bulky base station 14. For example, the detection unit 12 may be woven into a shirt sleeve, mounted on an armband or bracelet, or an implantable device. In other embodiments, one or more detection units 12 or other sensors may be used in beds, chairs, infant car seats, or other suitable clothing, furniture, equipment, and people prone to seizures. Can be placed in or incorporated into accessories. The detection unit 12 can comprise a simple sensor such as an electrode that can send signals to the base station for processing and analysis, or comprises a “smart” sensor with some data processing and storage capabilities. be able to. The detection unit 12 can include one or more smart client applications. In some embodiments, a simple sensor can be wired or wirelessly connected to a battery powered transceiver that is mounted on a belt worn by a person.

  The system can monitor the patient at rest, for example in the evening and at night. If the patient detection unit 12 detects a seizure, the detection unit 12 may be wired or wireless, eg far away from the base station 14 via a communication network or wireless link and via Bluetooth (Bluetooth®). Mobile phone or other portable or desktop device, or to the base station and a remote cellular phone or other device at the same time. The detection unit 12 can transmit some signals to the base station apparatus for a more thorough analysis. For example, the detection unit 12 may process and use EMG signals (and optionally, or in some embodiments, ECG, temperature and orientation sensors, saturated oxygen, and / or audio sensor signals) for seizures. An initial assessment of the likelihood of occurrence can be made and those signals and their assessment can be sent to the base station 14 for separate processing and confirmation. Upon confirming that a seizure may have occurred, the base station 14 initiates an alarm and designates an individual designated on the network 15 by email, text, or any suitable wired or wireless message indicator. Send an alarm to Note that in some embodiments, the detection unit 12 is smaller and more compact than the base station and is convenient for using a power source that has limited strength. Thus, in some embodiments, it is advantageous to control the amount of data transferred between the detection unit 12 and the base station 14. This is because the life of an arbitrary power supply element incorporated in the detection unit 12 is extended by doing so. In some embodiments, when one or more of the detection unit 12, base station 14, or caregiver, eg, a remote caregiver monitoring a signal provided from the base station, determines the occurrence of a seizure. The video monitor 9 can be triggered to collect information.

  A base station 14 powered by a general household power source and containing a backup battery has higher processing, transmission and analysis capabilities available for operation than the detection unit 12, and a greater amount of signal history. And the received signal can be evaluated in light of the large amount of data. The base station 14 communicates with an alarm transceiver 16 located remotely from the base station 14, such as a family bedroom, or wireless devices 17, 18 carried by a caregiver or located at the workplace or clinic. can do. Base station 14 and / or transceiver 16 may send alerts or messages to designated people via mobile phone 17, PDA 18, or other client device via any suitable means such as network 15. Thus, the system 10 can provide an accurate log of seizures to allow the patient's physician to recognize the success or failure of the treatment regime more quickly. Of course, the base station 14 can simply comprise a computer with an installed program that can receive, process, and analyze the signals described herein and transmit alerts. Base station 14 may include one or more smart client applications. In other embodiments, the system 10 is configured to transmit signal data to a smartphone, such as an iPhone (iPhone®), for example, and an EMG signal as described herein using an installed program application. Can be simply provided with an EMG electrode as part of a device configured to receive an EMG signal from the electrode. In a further embodiment, so-called “cloud” computing and storage devices can be used over the network 15 for storage and processing of EMG signals and associated data. In yet another embodiment, the one or more EMG electrodes are collectively as a single unit having a processor capable of processing EMG signals and sending alerts over the network as disclosed herein. It could also be packaged. In other words, the device can comprise a single product that is placed on the patient and does not require a transceiver remote from the base station. Alternatively, the base station may be a smartphone or a tablet.

  In the embodiment of FIG. 1, the signal data can be transmitted to a remote database 19 for storage. In some embodiments, signal data is sent from a plurality of epilepsy patients to the central database 19 to provide a basis for establishing and improving the general “baseline” sensitivity level and signal characteristics of epileptic seizures. Can be “anonymized” as provided. Database 19 and base station 14 are remotely accessed from one or more remote computers 13 via network 15 to allow detector unit and / or base station software updates and data transmission. Also, in some embodiments, the remote computer 13 or another computer can receive alarm signal and EMG signal data between any number of different devices associated with a designated individual configured to receive signals. The function of monitoring the exchange of data including the exchange of. Base station 14 can generate an audible alarm, such as can be remote transceiver 16 or detection unit 12. All wireless links can be bi-directional for software and data transmission and message delivery confirmation. Base station 14 may also employ one or all of the messaging methods listed above for notifying seizures. The base station 14 or the detection unit 12 can be provided with an “alarm release” button that terminates the incident warning.

  In some embodiments, the transceiver can be additionally mounted in a furniture unit or other structure, such as an environmental unit or object. If the detection unit is sufficiently close to the transceiver, the transceiver can transmit data to the base station. Therefore, the base station can recognize that the information has been received from the transducer and thus from the associated environmental unit. In some embodiments, the base station selects a particular template file, such as thresholds and other data described later herein, depending on whether or not a signal from a particular transceiver is received. Can do. Thus, for example, when receiving information from a detector and from a transducer associated with a bed or crib, the base station may be, for example, clothes worn by an individual during normal exercise, or in a home where a patient brushes their teeth, for example. Data can be handled differently than if received from a transducer associated with another environmental unit, such as an item near the washbasin. More generally, in some embodiments, the surveillance system may be comprised of one or more elements that have global positioning (GPS) capabilities. The location information may also be used to adjust one or more routines available in the detection algorithm. For example, GPS performance may be included with or in one or more microelectromechanical sensor elements included in the detection unit.

  The embodiment of FIG. 1 can be configured to have minimal invasiveness for use during sleep or to inhibit daily activities only minimally, such as one or two minimum Requires only a few electrodes, no head electrodes, can detect seizures with motor symptoms, and alerts to the presence of seizures at one or more local and / or remote sites Can be cheap enough for home use.

  FIG. 2 illustrates one embodiment of the detection unit 12 or detector. The detection unit 12 can include an EMG electrode 20. In some embodiments, an ECG electrode 21 may also be included. The detection unit 12 can further include an amplifier having a lead-off detector 22. In some embodiments, the one or more lead-off detectors indicate whether the electrode is in physical contact with the human body or too far away from the human body to indicate muscle activity, body temperature, brain activity, Alternatively, a signal can be provided that detects other patient events. The detection unit may further include one or more elements 28 configured to detect the position and / or orientation of the detection unit, such as a solid-state microelectromechanical (MEMS) structure. For example, element 28 may include one or more micromechanical inertial sensors, such as one that may include one or more gyroscopes, accelerometers, magnetometers, or combinations thereof.

  The detection unit 12 may further include a temperature sensor 23 that senses the person's body temperature and one or more orientation sensitive or position sensitive elements 28. Other sensors (not shown) such as accelerometers, microphones, oximeters may also be included in the detection unit. Signals from electrodes 20 and 21, temperature sensor 23, orientation and / or position sensor 28, and other sensors can be provided to multiplexer 24. The multiplexer 24 can be part of the detection unit 12. Alternatively, if the detection unit 12 is not a smart sensor, it can be part of the base station 14. The signal can then be communicated from multiplexer 24 to one or more analog-to-digital converters 25. The analog-to-digital converter can be part of the detection unit 12 or can be part of the base station 14. The signal can then be communicated to one or more microprocessors 26 for processing and analysis as disclosed herein. The microprocessor 26 can be part of the detection unit 12 or can be part of the base station 14. The detection unit 12 and / or the base station 14 may further include an appropriate amount of memory. Microprocessor 26 can communicate signal data and other information using transceiver 27. Communication by and between components of the detection unit 12 and / or base station 14 may be wired or wireless communication.

  Of course, the exemplary detection unit of FIG. 2 can be configured differently. Many of the components of the detector of FIG. 2 can be located in the base station 14 rather than in the detection unit 12. For example, the detection unit can simply comprise an EMG electrode 20 in wireless communication with the base station 14. In the above embodiment, A-D conversion and signal processing can be performed by the base station 14. Multiplexing can also be performed at the base station 14 if the ECG electrode 21 is included.

  In another example, the detection unit 12 of FIG. 2 has one or more of an EMG electrode 20, an ECG electrode 21, and a temperature sensor 23 for wired or wireless communication with a small belt-mounted transceiver portion. An electrode portion can be provided. The transceiver portion may include a multiplexer 24, an A-D converter 25, a microprocessor 26, a transceiver 27, and other components such as memory and I / O devices (eg, alarm release buttons and visual displays).

  FIG. 3 illustrates various communication means such as one or more microprocessors 30, a power supply 31, a backup power supply 32, one or more I / O devices 33, and an Ethernet connection 34 and a transceiver 35. 1 illustrates one embodiment of a base station 14 that may include: The base station 14 has a higher processing and storage capability than the detection unit 12 to review the EMU signal in real time as the caregiver is received from the detection unit 12 or to review the historical EMU signal from the memory. Can include a larger electronic display for display of the EMG signal graph. Base station 14 can process the EMG signal and other data received from detection unit 12. If the base station 14 determines that a seizure may have occurred, an alert can be sent to the caregiver via the transceiver 35.

  The various devices of the apparatus of FIGS. 1-3 can communicate with each other via wired or wireless communication. System 10 comprises a client-server or other architecture and can communicate over network 15. Of course, the system 10 may comprise more than one server and / or client. In other embodiments, the system 10 may comprise other types of network architectures, such as a peer-to-peer architecture, or any combination or combination thereof.

  FIG. 4 illustrates an exemplary embodiment of a method 40 for collecting EMG signals, processing the signals to detect seizure events, and performing an alarm transmission protocol based on the detected events. The detected event may include, for example, identification of a part of a seizure or identification of two or more parts of a seizure. For example, the identification of the seizure phase part of a seizure may be treated as an event, or two or more temporally correlated parts of the seizure, such as the identification of the tonic phase part of the seizure and the identification of the clonic phase part. Identification may be treated as an event. That is, the detection described above can be considered as detection of one tonic-clonic seizure event. In some embodiments, two or more routines can be executed in the monitoring protocol, including routines configured to detect different seizure activity portions. For example, in some embodiments, the first routine may be configured to respond to tonic seizure activity and the second routine may selectively respond to clonic seizure activity. Different events can be detected by combining outputs from various routines.

  In step 42, the patient can be monitored for seizure activity by collecting EMG signals using one or more EMG sensors. In some embodiments, additional sensor data can be collected and used to determine the risk of adverse effects of detected seizure activity. The signal data can be collected and analyzed using, for example, any of a variety of routines as further described herein. In step 44, different detection events can be identified based on responses in various performed detection routines. For example, as shown in Table 1, various routines can produce different output responses and identify different events based on those routine responses. In step 46, the detected event can be linked to one of several selectable alarm transmission protocols. That is, a specific alarm transmission protocol can be selected. The particular form of data determined for transmission may depend on, for example, the risk associated with the detected event. The data determined for transmission may in some embodiments include, for example, alarm messages, a subset of statistical information related to algorithm detection, and any combination of time domain and / or frequency domain EMG data, etc. It can take any form among various forms. As shown in step 48, the transmission protocol can then be executed. In addition, appropriate data can be transmitted if it is considered appropriate.

  In some embodiments, one or more EMG sensors can be used to monitor a patient for seizure activity and EMG signals can be collected and analyzed for the presence of one or more characteristics of seizure activity. . For example, the presence of a particular feature of seizure activity can be determined by analyzing the collected EMG data using one or more analysis routines. Also, in some embodiments, at least one routine may be selective for an activity of a particular part of the seizure activity to facilitate risk stratification. The routine may be “selective” for features present in some part of the seizure. This feature can be detected as a routine positive response in patients experiencing (or transitioning to) that part of the seizure, but the feature is virtually absent or not detectable Or may produce substantially different output responses in the absence of these parts of the seizure. Thus, the likelihood that a positive response to such a routine (or combination of routines) is appropriately associated with the presence of a particular part of the seizure can then be determined. Thus, identification of a particular portion of the seizure can be facilitated by executing one or more selective routines. Also, at this time, the classification of the detected seizure may be based on the presence or absence of such part of the seizure. Then, for example, if the presence of such portions of the seizure is related to some extent to the likelihood that the patient is experiencing the adverse effects of the seizure, risk stratification can be performed accordingly.

  For example, in some embodiments, one routine can analyze EMG data for the presence of an increase in EMG amplitude, or a persistent increase in EMG amplitude. Another routine can also analyze EMG data for the presence of clonic burst activity. Two or more detection routines may be executed simultaneously. Also, in some embodiments, the detected event may be that a particular portion of the collected EMG signal data indicates a positive response in one routine, or a particular response in more than one routine. Identification can be based on whether a combination is indicated. Also, in some embodiments, detected events may be accompanied by responses from one or more routines, in which case the routine responses are spaced in time. For example, two or more routine responses from EMG data collected at different times may be temporally correlated and treated as related to a single detected event, as described further below. May be. Or, if properly spaced in time, two or more routine responses can be considered as separate events that are then separated by, for example, separate event responses. Can be linked to.

  A routine for analyzing an EMG signal with increased EMG amplitude includes, for example, collecting the signal over a period of time and the EMG signal amplitude or signal amplitude collected within one or more time windows within that period. Determining whether the integral value has increased beyond a certain threshold. Also, in some embodiments, for example, based on multiple time windows from which a particular threshold amplitude was obtained, it can be determined whether the level of signal increase was sustained for a threshold duration. The threshold level of EMG signal amplitude may be set such that in some embodiments, such routines respond even to mild muscle movement symptoms. For example, in some embodiments, the threshold setting can be determined based on a measurement of the maximum signal amplitude that an individual can provide during medullary muscle contraction. For example, for some patients, any maximum value between about 2% and about 50% can be selected to capture mild motor symptoms. Other thresholds may be used, but if the threshold is set too high, routine responsiveness may be limited to some motor symptoms.

  In some embodiments, the signal amplitude can be measured in units of standard deviation higher than the baseline signal. For example, evaluating whether there is a continuous amplitude increase is to measure the difference between the measured signal amplitude and the baseline signal amplitude in units of standard deviation, and the number of standard deviations is ( It may include an assessment of whether a threshold (or Z factor) has been exceeded (in standard deviation units) and / or whether such factor has been exceeded over a particular time interval. In order to be persistent, a signal value or smoothed signal value must maintain a threshold level at least a critical number of times within a time interval, or maintain a threshold level throughout the time interval. May be necessary. To improve the signal-to-noise ratio of the detected signal, it may usually be desirable to integrate the signal for a specific time interval, such as about 100 milliseconds to about 500 milliseconds. By choosing a long integration interval, the system is less susceptible to random fluctuations and signal noise sources, and overall detection sensitivity can be improved. Thus, for some routines, including those that respond to mild motor symptoms and / or certain parts of tonic activity, a relatively long time window, for example by integrating over a time window on the order of a few hundred milliseconds It may be advantageous to integrate over.

  However, the integration of the EMG signal over a considerable duration usually results in a loss of time resolution of the signal data, and a transient EMG typical for clonic bursts (which lasts only for a few hundred milliseconds). Routines that are designed to be selective for activity may typically require split EMG signal amplitude measurements within a shorter window than is used by other routines. For example, to determine whether a signal varies over a time interval on the order of about 100 milliseconds, it is necessary to measure and analyze the signal at least some times over the period of each variation. In some embodiments, routines for detecting clonic bursts may include identifying peaks and qualifying peaks that meet the criteria for being qualified as clonic bursts. . For example, in some embodiments, EMG data may be sampled over various intervals suitable for detecting signal temporal changes during a pattern of clonic bursts. The data is accessed using a peak detection program and then qualified so that any of the identified peaks can be used to determine whether it may be considered a clonic burst. receive. For example, by selectively counting qualified growth based on meeting the minimum and / or maximum width requirements typical of mesomorphic activity, the algorithm can It can be selective for activities. That is, routines suitable for measuring characteristics of clonic burst activity, such as the number of clonic bursts, can be configured to be selective for seizure clonic phases.

  By analyzing the EMG data for signal enhancement that meets the minimum and / or maximum width requirements, as described herein, the routine can be selective to patterns characteristic of clonic activity. Can be. In addition, because the seizure phase is correlated with the risk of seizures, inclusion of such routines, ie routines that are selective to the activity of the clonic phase, facilitates risk stratification. be able to. Further detailed descriptions of some embodiments of routines selective for clonic activity are further included herein in the Examples section at the end of this application. The routines can respond when a seizure phase is present and / or during a period that normally leads to a clonic phase.

  In some embodiments, the method 40 uses a routine configured to measure sustained EMG activity to collect and analyze EMG signals (step 42) for seizure activity, and tonic phase seizures. If activity is present, settings and / or thresholds suitable to provide a positive routine response can be included. Another routine may include settings and / or thresholds that, in some embodiments, are suitable for selectively detecting a clonic phase. That is, if there is a clonic activity, the routine can indicate a selective response. At step 44, the method 40 can then determine whether there is a detected seizure event based on, for example, data from either or both of the aforementioned routines.

  For example, if the first routine is configured to respond to tonic seizure activity and the second routine selectively responds to clonic seizure activity, the EMG signal may be positive during analysis, for example. The first routine response may be initiated, but a positive second routine response may not be initiated. These routine responses can be used to classify EMG signals as corresponding to tonic phase detection events. In some embodiments, the first routine is associated with a threshold setting and corresponds to either a tonic phase seizure or a mild motor symptom including symptoms unrelated to the seizure using a positive routine response. Possible events can be detected. Also, for some patients, such an event can be given a warning event state because the risk of adverse effects of having a seizure is only minimal. For example, in some embodiments, an event can be qualified as requiring a warning condition and only select an associated warning transmission protocol by decoupling detection of mild motor symptoms or tonic phase activity. can do.

  In step 46, it may be determined whether the transmission of EMG data or event messages is reasonable. For example, an event can be assigned a warning state, and in some embodiments, a transmission protocol can be initiated that sends a message that a warning event has been identified. Such a message is advantageous, for example, because it can be transmitted using only minimum power consumption (as shown in step 48). The message is limited in some embodiments to include an indication of the time of the event and can describe that an alert has been reached, or whether the reached amplitude value or how much time amplitude threshold is present, for example. A description that the warning threshold has been reached can be included along with summary data including whether it was maintained. In other embodiments, extended EMG signal data may be transmitted with a warning message, but the range of transmitted data may typically be limited for events that pose only a limited risk.

  The monitoring system may also initiate the transmission of an emergency response, such as one that may include an emergency message. For example, in some embodiments, each of the aforementioned routines (eg, a first routine for increased or sustained activity and a second routine configured for temporary EMG activity) each occurs during a collection period. And the detected EMG signal event can then be classified as being associated with the presence of a seizure phase. Such a classification can, for example, initiate an emergency response message. That is, the detected clonic event can be linked to an alarm protocol that includes an emergency response. For example, a clonic phase event can also be considered to be present only if the EMG signal is in response to the second routine. In contrast, in some embodiments, the first routine can respond to tonic phase activity, but can also respond to clonic activity. However, even if there is no selectivity for the first routine in this way, it cannot prevent selective identification and classification of seizure activity based on the presence of either tonic or clonic phase. This is because such ambiguity can be resolved by considering the other routine. For example, the second routine may not be responsive if only tonic phase activity is present because the second routine may be selective for clonic activity. Thus, in some embodiments, both tonic activity and clonic activity can be selectively detected.

  Based on the detection of clonic activity, the events can be risk stratified, and then an appropriate transmission protocol is selected, as shown in step 46. For example, in some embodiments, a clonic event can be considered an emergency event and a transmission protocol can be selected such as one that can include an alarm message sent directly to an EMT worker. An ambulance can then be sent to the patient's home or another appropriate emergency response can be made. Alternatively or in addition, for example, another caregiver can be contacted and instructed to confirm the patient's location and examine the patient. In some embodiments, detection of a clonic event can initiate a transmission protocol in which time-dependent data and / or frequency-dependent data is transmitted remotely to a system user. The remote user is then given the option to evaluate the data and the user can perform an appropriate response, such as one that may include termination of the emergency response or start of the emergency message. If data is sent to a remote user and it is necessary to actively initiate an emergency message, such data can be considered part of the alert protocol. That is, data or messages that still need to be initiated by the user to raise the event to an emergency state can be considered herein as part of the alert protocol. In some embodiments, data can be sent to the recipient, but after some time the system can automatically deliver an emergency message to the caregiver. Thus, some transmission protocols can be considered to be part of an emergency protocol because they can be preset to initiate contact with a caregiver. Also, as used herein, unless otherwise noted, an “emergency transmission protocol” is an EMT that has received instructions to physically support the patient, for example, actively moving to the patient's location. An emergency message may be sent to a caregiver, including but not limited to a caregiver, and may refer to a transmission protocol that is preset to send. Also, in some embodiments, the emergency protocol can still be interrupted and canceled by the remote user.

  Continuing with some examples with respect to an example of a first routine configured to identify an increase or duration of EMG activity, and a second routine configured to identify temporary EMG activity, As shown in Table 1, classification and risk stratification can be selected for various detected events.

  Table 1 applies to how EMG data collected at approximately the same time, i.e., to each EMG data collected at approximately the same time, how responses to routines are combined and associated with seizure events In one case, one embodiment of how it can be applied is shown. However, the monitoring system may collect and analyze the EMG signal using one or more routines that may respond to the EMG signal data over a period of time. Also, for example, a routine or group of routines can respond to different portions of the EMG signal that are spaced in time. If the routine response is derived from properly separated EMG data, the routine response can be treated as equivalent to a separate event. For example, in some embodiments, if two routine responses are associated with portions of EMG signal data that are spaced in time from each other by about 1 minute to about 10 minutes, the two routine responses Responses can be treated as being from isolated events. Individually detected events can be classified based on the risk and the appropriate transmission protocol linked to that event. Similarly, in some embodiments, if two routine responses are separated by an interval that is less than a specific period, they are considered to be related in time, for example, and both It can be considered an event. That is, a temporal relationship between two or more routine responses associated with different parts of the signal data can be determined, for example, the signal data can be considered to be part of a multi-stage seizure event. For example, routine responses associated with seizure tonic and clonic phases are considered to be temporally related and treated as potentially the result of a single tonic-clonic seizure event. Can do. For example, in some embodiments, the response of one or more routines indicates that a tonic phase is identified, and within about 2 seconds to about 60 seconds after the identification is made, If multiple routines indicate that a clonic phase exists, it can be considered that there is a possibility of tonic-clonic seizures. Second, detection of tonic-clonic seizure events can be treated as a single event for purposes of classification and risk stratification. For example, tonic-clonic events may be considered events that in some embodiments are reasonable to perform an emergency transmission protocol.

  The system described herein can, in some embodiments, analyze collected EMG signals based on a combination of responses obtained from individual routines. Also, in some embodiments, multiple responses to signal data including combinations of multiple routine responses over multiple time intervals may be considered when classifying detected data based on risk.

  In some embodiments, if multiple routine responses are detected over a period of time, the system can perform, for example, one or more additional actions. For example, the system can execute an algorithm suitable for evaluating the intervening period between positive routine responses and examining the intervening period for signs of different activities. In some embodiments, the amplitude of the EMG data in the intervening period between positive routine responses, for example, is increased, but is still below the appropriate threshold level to trigger responses in individual routines. Signals can be analyzed for the presence of EMG signals that can be increased. Also, for example, a series of responses can be handled in different ways based on whether a particular EMG signal level has been reached within one or more intervening periods between responses. For example, responses can be treated as individual events, or different algorithms can consider whether an individual response can be part of one or more multi-stage seizure events. Also, in some embodiments, the multi-stage seizure pattern may be based on data collected for a particular single patient or a particular patient demographic.

  In some embodiments, the amplitude of EMG data in the intervening period between positive routine responses can be analyzed for the presence of an EMG signal that can increase or decrease. For example, some patients may experience a series of seizure events followed by a general state of central nervous system (CNS) inhibitory action. Such an inhibitory effect may be associated with a high risk of SUDEP. It may also be particularly important to detect this when a CNS inhibitory effect follows a series of seizure events. Some multi-stage algorithms can examine the presence of multiple events that pose a SUDEP risk. In some embodiments, the algorithms can investigate whether signals between routine responses meet any of a variety of profiles that may reflect CNS suppression effects. . Such a pattern may include the presence of a trend in the relative signal during the baseline period between responses. For example, variability and / or trends in EMG data can be considered.

  In some embodiments, data from either saturated oxygen levels or heart rate can also be taken into account. Changes in heart rate or oxygen level may be a particularly important consideration in determining a patient's condition. However, for some patients, there may be a fairly long period of time during which heart rate suppression is negligible before the sudden change in heart rate during SUDEP. Similarly, the oxygen saturation level may change only gradually during the progress of SUDEP. Since the measurement of carbon dioxide levels can change more rapidly in the early stages of SUDEP progression, the development of SUDEP can be more reliably diagnosed. However, it may be difficult to measure the carbon dioxide level in an outpatient environment. Also, analysis of the sensor elements described above along with EMG data can improve detection of initial signs of SUDEP. In some embodiments, EMG data can be used to trigger or adjust the execution of one or more other sensor routines. For example, one or more responses can be used to trigger data collection using a pulse oximeter or to adjust thresholds and response sensitivity based on heart rate and / or saturated oxygen levels. . Also, the trend or variability of the pulse oximeter data can then be taken into account when determining the overall response to the available data set. For example, in some embodiments, one emergency event or some suitable warning event related by some intervening time interval is used to configure one or more additional sensors including a pulse oximeter. The collection of the data used can be triggered.

  In some embodiments, for the presence of one or more responses over a particular period of time, eg, when one or more responses are detected that include positive responses that can be considered individually as warning events. An appropriate message to notify the user can be sent to the remote user. The remote user then requests, for example, transmission of additional EMG information, eg transmission of data that can be used to display the time and / or frequency dependence of the relevant EMG signal. Can do. In addition, the user can use any of a variety of calibration procedures or other procedures that may be useful to further evaluate the collected EMG signal data and / or the condition of the sensor during a specified period of time. Can start. Other appropriate measures can be taken, including calling the patient or another caregiver. In some embodiments, the monitoring system can handle a series of responses in a manner that can vary depending on whether the system is in one of several different selectable states. For example, a patient may have one or more different monitoring settings based on, for example, whether the patient is sleeping in a bed, whether the patient is alone at home, or whether the patient is at home with another person. Choices to select may be given. Also, depending on whether the patient is in any of those states, for a particular response or group of responses, the monitoring system selects, for example, a transmission protocol associated with any of the emergency alerts be able to.

  The method 40 may be useful in organizing whether it is desirable to transmit detected data and / or what form the transmitted data can take for a particular event. For example, a transmission protocol that sends only patient status messages (rather than bandwidth-intensive time and / or frequency domain EMG data) can be selected, and the remote user can further analyze and analyze the EMG data. An option may be given to consider and not count the data or take one or more other actions described herein. Importantly, for at least some of the detected events, such analysis may be performed before the emergency message is transmitted. For example, in some embodiments, further analysis or review may be required or may be possible before an emergency message is sent. This is because automatic alarm messages are queued for transmission but not sent immediately. Some epilepsy patients tend to trigger responses from non-seizure events or mild events somewhat frequently, and the cost and / or inconvenience of initiating an inappropriate response to a single event is quite large In some cases, risk stratification and consideration of events, including those that are difficult to access, may be highly desirable. In addition, in some embodiments, data analysis and review can be performed without manual intervention by a remote user.

  For example, in some embodiments, in addition to or instead of performing a warning transmission protocol, detection of a warning event may include, for example, for proper calibration and / or complete surface contact with the skin. Further analysis and review of the collected EMG data can be automatically initiated based on the application of one or more algorithms, including algorithms that survey the sensor for gender. Also, such analysis and review can be initiated automatically by some remote users or detection systems in some embodiments, or by both remote users and detection systems. In addition to EMG data associated with one or more detected events, such further analysis includes processing of EMG data collected prior to and / or after the detected event. May be.

  In some embodiments, a period for further EMG collection can be triggered in response to the detected event. One or more algorithms may be executed at the completion of the period, or at some other time during the period, again taking into account the patient's condition. For example, in response to some events, a further analysis warning period can be triggered. At the end of the warning period, the selected algorithm can direct the system to start the transmission protocol, or clear the warning condition and return monitoring to the standard measurement protocol. Importantly, upon completion of the warning period, the system has access to both the event that triggered the warning period and also the collected EMG data, and all the data available to determine the appropriate response. Can be considered. Execution of the alert period may be triggered in response to an alert event in some embodiments, in place of or in addition to executing the alert transmission protocol as described in method 40. be able to.

  For example, because a suitable event can be detected to initiate an alert, the monitoring system can continue to monitor the patient and collect more information to better evaluate the patient's condition. it can. During further evaluation, the routine or routines may further respond to the collected EMG signal and detect another event (eg, an event that changes the patient's condition). Ideally, such events can be detected with greater certainty than initial warning events. Thus, by waiting until a more certain event is detected, the system can determine a more appropriate response for the patient.

  In addition, in some embodiments, other routines can be triggered to individually review data for signs of non-seizure activity, so such data routines can be used to determine the significance of warning events. By not counting sex, false positive detection can be avoided. In some embodiments, a dedicated routine that negatively weights EMG data, including, for example, one or more periodic routines that examine an unnatural periodic signal or a normal signal when initiating a warning period. Can be triggered. Other routines that can negatively weight EMG data may include comparing the data to a non-seizure waveform. For example, this information can be used to compare EMG waveform data with a database of other spectral profiles (eg, model profiles for patient-specific seizure and / or non-seizure events). It is possible not to count warning events. In some embodiments, it may be beneficial to perform at least some routine at the base station. Thus, in some embodiments, the response to the detection of the warning event may be transmitting data to the base station.

  Some embodiments described herein that include instructions for further characterizing the warning event include one or more routines that are also very sensitive to mild motor symptoms. Also, for example, by including those routines and instructions, some patients can be successfully monitored in an outpatient setting, including some patients that are particularly difficult to monitor. For example, for some patients with a significant amount of adipose tissue, it may be difficult to distinguish seizure activity from non-seizure activity. It may also be difficult for such patients to set a threshold based on EMG amplitude or persistent EMG amplitude that completely distinguishes seizure event data from non-seizure events. To alleviate these concerns, it may be desirable to collect further information after a warning event that may be performed to more fully assess the event, and after such further assessment, the dangerous event If it is suspected that it has occurred, an emergency alarm can be initiated.

  In some embodiments, the detection method can include the execution of a routine (or group of routines) that analyzes EMG data for the presence of clonic bursts, an initial motor symptom, or a mild motor symptom. It may further include the execution of another routine suitable for detecting. These embodiments can include instructions that can activate a warning flag to determine a warning period if a routine suitable for detecting mild motor symptoms is detected. Additional EMG can be collected within the warning period. Also, for example, the EMG signal can be analyzed for signs of clonic activity. An alarm can be triggered immediately if clonic activity is detected during the warning period. In some embodiments, in the alert period, when the EMG activity increases over time, or if the alert signal does not decay, whether or not there is clonic activity, the response Can start. For example, in some embodiments, if the motor component of general seizure activity is detected and there is an increase in EMG amplitude for at least about 15 seconds to about 30 seconds, An alarm can be started without detection. In some embodiments, the method may delay triggering an alarm even if seizure or clonic seizure activity is detected during the warning period. For example, the alarm period can be completed and the alarm can be started only after the end of the warning period. For example, the method can increase the confidence that it is appropriate to wait until the end of the warning period to collect further data and send an emergency alarm. In addition, another event can be detected that qualifies the detection of possibly false positives within the warning period. For example, a routine can be triggered that indicates that the sensor may have lost calibration. Alternatively, it can be determined that the signal is unnaturally periodic and possibly due to a non-seizure cause.

  As described above, the EMG data can be further characterized within a warning trigger period. For example, the mesophase EMG data may be detected, or the routine may not specifically count the warning trigger. Decisions based on warning threshold detection may include, in some embodiments, other validation data, including data collected by embedded sensors or other sensors that may provide information about the patient's physiological condition. It can also be linked to detection. Alternatively, after a warning period is triggered, the threshold event that triggers this period may not yet be characterized further. If the first threshold event has not yet been further characterized, the method can initiate a response based on the collected data available. For example, an emergency response can be triggered or some other response can be made. Decisions based on those first threshold detections may be patient specific or condition specific as discussed above. For example, a patient who is considered at high risk for adverse effects from a mild seizure event can be treated differently from another patient who is considered at lower risk. That is, a patient who is alone or sleeping at home can be treated differently from a patient who is known to have another person or caregiver.

  An exemplary embodiment of an embodiment of a method 50 that can determine a warning period is shown in FIG. As shown in step 52, EMG data can be collected and analyzed for the presence of seizure activity characteristics. For example, the routine can be configured to detect persistent EMG signal amplitudes that can be characteristic of seizure activity, and in some embodiments, the routine is sensitive to mild muscle movement symptoms. It may include setting a threshold to be. If a positive response is determined during execution of the routine, a warning flag can be activated. For example, a timer that defines a warning period can be started by operating a warning flag. The warning period can be configured to continue, for example, for an appropriate period to further evaluate the EMG data. The warning period can last, for example, between about 15 seconds and about 30 seconds in some embodiments. As shown in step 56, additional EMG data can be collected during a warning period that can be used to analyze the EMG data collected for signs of additional seizure activity. In some embodiments, at least one routine applied to analyze such additional EMG data can respond to clonic seizure activity. At the end of the warning period, an appropriate response can then be performed.

  In some embodiments, the routine may be selective for clonic seizure activity, and the collection of EMG signals and the EMG signal data for transient EMG signal augmentation or presence of bursts. An evaluation may be included. Methods for identifying bursts, collecting and counting bursts are described in this and Applicant's US Patent Application No. 13 / 275,309 and Applicant's US Provisional Patent Application No. 61 / 969,660. Also, routines for detecting seizures, such as those described in these documents, including routines that determine, for example, burst activity, the number of bursts, or the presence of a burst sequence, are clonic burst activity. Can be applied as a selective routine, and can selectively relate to the activities of the clown phase. In addition, some of the routines described in these references describe monitoring algorithms that can be weighted with different routine inputs and used to determine whether a seizure is present. Also, by appropriately weighting individual routines based on clonic burst activity, the monitoring algorithm can be made larger or smaller selective to the secular phase. Thus, in some embodiments, the output of the monitoring algorithms described in these references may be used as a routine that may be selective for the presence of clonic activity.

  As described in these documents, EMG signal data that meets various requirements or thresholds is qualified as a clonic burst. For example, clonic bursts can be qualified based on minimum and / or maximum width requirements. Processing operations such as signal rectification, filtering, and / or other processing operations may be performed as appropriate to identify data of clonic bursts and determine the number of clonic bursts. For example, the analysis routine can include a peak detection program. The peak detection program can identify and shape the data after, for example, bandpass filtering and rectification. Once processed, generation of burst statistics, comparison with thresholds, and data qualification can be more easily achieved. Any suitable peak detection technique (eg, continuous wavelet transform) may be used. For example, in some embodiments, peak detection may be performed using a data smoothing technique (eg, moving average filter, Savitzky-Golay filter, Gaussian filter, Kaiser window, various wavelet transforms, etc.), baseline correction process (eg, monotonic Minimum, linear interpolation, loss normalization, minimum moving average, etc.) and one or more peak finding criteria (SNR, detection / intensity threshold, peak slope, local maximum, shape ratio, ridge, model-based criteria, peak Application of width etc.).

  In some embodiments, a peak is a characteristic associated with a seizure phase, i.e., a transition to a seizure phase, such as can occur when tonic activity moves to the seizure phase. Can be qualified for relevant characteristics. Peaks can be qualified based on, for example, the associated waveform, regularity, periodicity, width (sometimes referred to as time width), amplitude, or other factors. Qualification makes it easier to distinguish seizures from non-seizure events and can improve the confidence that an increase in detected signal is appropriately associated with seizure activity. Also, because other sources of EMG signal increase (eg background interference, noise and non-seizure behavior) can be distinguished from qualified burst activity, by monitoring the EMG signal for such activity, Sensitive seizure detection is provided.

  In some embodiments, a portion of the EMG signal can be converted to the frequency domain and analyzed for the presence of clonic burst activity. For example, a frequency transform can be used to transform the collected signal data into the frequency domain, and the integrated strength of signals in one or more frequency bands can be calculated for a given time period. Also, by repeating such treatment over adjacent periods, EMG signal data can be analyzed for data features that meet the temporal width requirements associated with clonic burst data. Alternatively, an appropriate bandpass filter can be used to separate activity in one or more bands. For example, in some embodiments, signals in one or more bands ranging from about 2 Hz to about 120 Hz or about 240 Hz can be separated and analyzed for the presence of clonic phase bursts. Also, for at least some patients, a specific frequency band that facilitates differentiation of seizure events from non-seizure causes can be selected. Alternatively, EMG signals associated with all collected frequencies can be analyzed for the presence of clonic phase bursts, and the collected EMG signal data can be analyzed, for example, without performing frequency conversion. it can.

  In some embodiments, the identified peak is qualified for one or more characteristics present in the seizure phase of normal seizures, as further described herein, and during seizures Qualification can be achieved by increasing the selectivity for detecting a proxy. A peak can be qualified based on, for example, the associated waveform, regularity, periodicity, width (sometimes referred to as time width), amplitude, or other factors described herein. Also, by selecting a threshold that excludes activity from other parts of the seizure, the routine can be made selective to, for example, the secular phase of the seizure. Non-seizure causes and other noise sources can also be distinguished from qualified peaks, so peak data qualification not only facilitates selective identification of clonic phases, It is also possible to enhance the distinction from non-seizure causes and can therefore be used to enhance the overall system performance with respect to seizure detection.

  In some embodiments, the region where the EMG signal amplitude is increasing can be qualified as meeting one or more conditions related to the duration of the signal increase. That is, an area where the signal is increasing, or an area that is increasing by at least the SNR above the background, qualifies as maintaining the requirements for the minimum duration and / or the maximum duration Is done.

  For example, in some embodiments, in order to qualify the signal data as suitable for clonic burst identification, burst statistics analysis, and / or burst counting, the signal data may include a minimum burst width and A positive response can be logged if qualified on the basis of satisfaction of the maximum burst width criteria and a certain number of bursts are detected over a period of time. That is, the routine can count clonic bursts or determine the rate of clonic bursts, so if a number or rate exceeds a threshold, log a positive response. Can do. In some embodiments, a burst envelope is generated, which can affect the SNR threshold that can be used to identify the burst. For example, in the case of a simple peak detection method, the mesophase burst meets a threshold SNR of about 1.25 to about 20, and a minimum threshold for a burst width of about 25 to about 75 milliseconds and about 250 milliseconds. Qualify by meeting a maximum burst width threshold of seconds to about 400 milliseconds or less. Next, clonic bursts are counted and multiple bursts or burst rates can be determined. For example, if about 2 to about 6 mesophase bursts are then measured within a time window of about 1 second, or another suitable number of mesophase bursts in some other suitable time window Can be triggered for some patients, a positive routine response can be triggered.

  In some embodiments, a positive routine response can be made if at least about 2 to about 6 mesophase bursts are measured within a time window of about 2 seconds to about 5 seconds. The threshold number of times of mesophase burst for a patient may be the same as or different from the predicted number of physiological events that can cause a clonic burst. For example, depending on the SNR threshold level and / or other thresholds, at least some of those physiological events associated with the burst may not be detected, but the threshold number of clonic bursts is still detected. Can be used to trigger a positive routine response.

  In order to further qualify the signal data, in some embodiments, the evaluation of the EMG data includes, for example, other characteristics of clonic activity in addition to the minimum and / or maximum width of the duration. May be included. For example, in some embodiments, a portion of the EMG signal can be identified to include one or more suspicious bursts, and the portion of the EMG signal can be converted to the frequency domain and The presence of frequency characteristics related to the activity of That is, the frequency data can be compared to a model frequency waveform typical for clonic activity. If this comparison does not support the finding that the frequency data may be related to clonic activity, for example, suspicious bursts may not be included in the burst count feature. That is, the identified activities do not pass qualification and may not be counted. The model waveform is based on EMG data for a particular patient or about patient demographics, and in some embodiments, the waveform data can be updated as historical EMG data is being collected.

  In some embodiments, regions identified as likely to be clonic bursts can be further qualified based on additional criteria. For example, identified data, e.g., data that is likely to be clonic bursts, includes periodicity, amplitude regularity, waveform regularity, other criteria described herein, and combinations thereof. Further analysis can be based on one or more features included. For example, in some embodiments, if the identified data is determined to be unnaturally periodic, the data may not pass qualification and may not be counted. Once properly qualified as meeting the criteria for clonic activity, use, for example, the number of clonic bursts to assess whether a seizure activity or seizure phase exists can do.

  In some embodiments, for example, to further qualify as a clonic burst, the identified data, eg, data that meets the minimum duration and / or maximum duration requirements is substantially Adjacent periods of the stationary signal may include stationary periods that last for a period of about 50 milliseconds to about 300 milliseconds. That is, the presence of a quiescent period of adjacent threshold durations can be used to qualify bursts as related to clonic activity. Once finally qualified as meeting the criteria for clonic activity, use the number of clonic bursts to assess whether seizure activity or a clonic phase exists. be able to. For example, if about 2 to about 6 mesophase bursts are then measured within a time window of about 1 second, or another suitable number of mesophases in some other suitable time window. If bursts are counted, a positive routine response can be triggered for some patients.

  In some embodiments, the data identified as possibly resulting from clonic activity can further be determined to be a burst train. Any burst sequence that is temporally related to each other (exists within a specific time window) and can be pooled together to generate burst statistics and / or examine trends in burst data over time Can include any number of bursts. For example, a burst train can include a plurality of bursts, such as from about 3 to about 30. The burst train can include any portion of the total number of bursts that exist in the seizure phase. For example, a plurality of bursts selected to be included in the burst sequence can be appropriately concentrated on one part of the seizure phase, eg, the beginning part of the seizure phase. In contrast, the burst sequence or the bursts contained therein (as further described in connection with FIG. 10 and routine 130), the distribution of periods between individual bursts in the burst sequence, and the distribution is seizure Qualification can be based on whether typical for activity or non-seizure activity. That is, a metric related to the distribution of periods between bursts (eg, standard deviation or percent average deviation) can be calculated and used to qualify burst sequences, or bursts contained in burst sequences. If too many bursts are included in the burst sequence, the metric is high because the period over which the burst sequence is measured can continue beyond the time frame where the average length of the period between bursts has increased There is a case. Also, in some embodiments, bursts can be collected from each of the anterior, central, and posterior phases of the seizure phase. Also, in some embodiments, a pooled value is then (based on data from each part) to determine such statistics that can be used to identify standard deviation or other burst statistics. Can be calculated.

  In some embodiments, to further qualify as a clonic burst sequence, multiple qualified burst sequences, e.g., a plurality of neighbors that may meet the minimum duration and / or maximum duration requirements above. Individual burst components can meet a threshold level for the percentage burst deviation of the period between individual bursts in the column. The presence of one or more mesophase burst sequences can be used to detect seizure activity and initiate alarm execution or log a positive response to the routine.

  More generally, any number of steps to identify suspicious data in clonic bursts and qualify such data to increase confidence that the data is properly associated with the mesophase Any number of steps can be performed. FIG. 6 illustrates an exemplary embodiment of a routine 60 for monitoring EMG data for seizure activity features including burst identification, which can be used to initiate a response. These responses can be used individually or in combination with other routines as a way to monitor patients to more fully characterize seizures and perform risk stratification. In routine 60, a certain value (such as clown burst number, clown burst number rate, or other clown burst statistics), such as a threshold value (such as number, rate, or other statistical threshold), and Peak data can be identified and qualified in one or more steps that produce values that can be directly compared and used to initiate a positive response in the routine. Also, for example, when used individually as a monitoring method, the positive response in routine 60 can be used to directly trigger one or more transmission protocols.

  In step 62 of method 60, an EMG signal is collected and a portion of the collected signal can be selected for processing. At step 64, one or more routines may be executed to identify whether a peak is present in the selected data. For example, any of routines 72 (FIG. 7), 96 (FIG. 8), and 120 (FIG. 9), or a combination of these routines may be performed. As shown in step 66, in some embodiments, the routine 60 includes a further assessment of whether the identified region meets one or more characteristics associated with a seizure phase. Depending on how many of the identified features are eligible, the number of clonic bursts can be determined. In step 68, the determined burst number or rate can be compared to a threshold. Also, in step 70, a positive response can be logged in a routine or an alarm can be executed if the threshold is met. After a positive response is logged, the system can then collect the next signal sample, for example. More generally, EMG data can satisfy an initial screen to be identified. Also, any number of different characteristics can be compared with this identified data and used to qualify or further qualify that the data is related to clonic activity . During qualification, the number of clonic bursts serves as a very sensitive and selective seizure variable for analyzing EMG data for the presence of seizure activity or for the mesophase portion of seizures. be able to.

  In some embodiments, as described in further detail in routines 72 (FIG. 7) and 96 (FIG. 8), step 64 of method 60 identifies EMG data that includes a period of time during which the EMG amplitude temporarily increases. This may be a candidate for further qualification, eg, further qualification as exhibiting clonic behavior. For example, the data can be qualified as containing an amplified signal that is maintained for a minimum threshold duration, as further described in step 90 of the more detailed routine 72 (FIG. 7). Similarly, as further described in step 110 of routine 96 (FIG. 8), the data can be qualified as containing an amplified signal that is maintained for a duration that is less than or equal to the maximum threshold. Based on the above criteria, the EMG signal can be identified as containing one or more mesophase bursts or a series of mesophase bursts as well. Also, for example, depending on the selection of the minimum and / or maximum threshold levels for duration, the identified signal may be highly selective for clonic activity for at least some patients. For example, the identified EMG signal data properly qualifies that the data may be correlated with clonic activity based on appropriate minimum and / or maximum duration thresholds, for example. Can be incorporated directly into an algorithm for identifying seizure activity. For example, a temporary increase or burst can be counted, and the number of bursts can be compared to a threshold number of times to assess whether a seizure or seizure portion exists.

  FIG. 7 illustrates one embodiment of a clonic burst identification routine 66 that can be used individually or in combination with another routine to identify portions of signal data that may include clonic bursts. Is illustrated. Routine 72 is shown to include a start trigger 74. The start trigger 74 can be set based on, for example, detected signal changes or changes to background noise. Alternatively, the start trigger 74 can be set to start after a predetermined interval, such as one that can be driven by an appropriate clock routine. In step 76, signal portions can be selected for analysis. The signal portion can include any number of data points collected by the EMG sensor. At step 78, the routine 72 can determine whether the signal selected for analysis exceeds a maximum allowable value or threshold setting. For example, at least in some routines, the signal may not be inappropriately large or likely to correspond to actual seizure activity, for example if the sensor or sensor contact is unstable or detected It may be desirable to determine whether a level that can correspond to an artifact has been reached, such as what may be present if the device needs calibration. In step 80, another routine may be instructed that the signal has exceeded the maximum allowable value. This other routine may be, for example, an alarm determination or routine configured to trigger recalibration of the detection unit, or a monitoring algorithm associated with a routine that examines signals that are too regular to be generated by humans. be able to. As shown in step 82, the routine 72 may now include obtaining the next signal sample or waiting for the next trigger and reinitiating the flow shown for the routine 72.

  In step 84 (which may be included in addition to or instead of performing step 78), routine 66 determines whether the SNR of the selected signal sample is greater than the selected threshold. Can be determined. If the SNR of the signal is greater than the background by at least the SNR threshold, the routine 72 first exceeds the SNR threshold in the routine 72 (as shown in step 86) in the selected signal sample SNR. You can decide whether or not. If the SNR of the signal sample is greater than the SNR threshold for the first time in the routine, a burst timer can be started, as shown in step 88. Burst timers can help, for example, prevent temporary spikes in activity (eg, spikes in data that may not reflect actual seizures) from being qualified as clonic bursts. The threshold level for the burst duration can be determined. For example, the burst timer may run for about 25 to about 100 milliseconds, or some other suitable value period shorter than typical for bursts of data obtained from seizure activity. Following the start of the burst timer, the routine 72 can acquire another signal sample and begin other steps of the routine. In addition, it is possible to activate the appropriate flag so that the next signal sample evaluated for signal and / or SNR level is found to contain a repeat of the previous sample selection. For example, a second sample of data can be selected (step 76). Also, the selected signal can go through the routine 72 step again. In step 84, if the selected signal (which may include, for example, data obtained from several acquisitions of signal samples) does not meet the threshold SNR, then step 90 is performed. Can be executed. In step 90, it can be determined whether the duration of the selected signal exceeds a threshold duration level, and if this duration is exceeded, the selected signal is Can be qualified as a clonic burst (step 92). Once the clonic burst is qualified, the routine 72 can be exited.

  Upon exiting routine 72, the data is sent to a routine where it can be further qualified for, for example, clonic burst characteristics and / or the data can be evaluated to determine an appropriate response to the detected burst data. can do. For example, a signal qualified in routine 72 may exit the routine if it exceeds a threshold SNR for a threshold level of duration. In some embodiments, it may also be useful to determine whether the signal has not exceeded the maximum threshold duration. To evaluate whether the outgoing signal does not exceed the maximum threshold duration, the signal exiting routine 72 can be sent to routine 96 of FIG. 8, for example. Again, as an example, burst data is sent to a circular buffer that is periodically evaluated for the detection of a burst sequence, or a decision is made as to whether an alarm should be triggered each time a burst is detected. It can be carried out. In step 90, if the selected signal does not exceed the threshold signal duration, the signal data can be considered too short to qualify as a burst. The routine can also collect the next signal sample-clearing any appropriate flag, such as may have been previously activated. Instead of collecting the next signal sample, EMG data can be collected and the routine 72 can be started again only when the next trigger is established.

  FIG. 8 illustrates another embodiment of a burst detection routine 96 that can also be used individually or in combination with another burst detection routine to identify portions of signal data that may include bursts. . For example, when used in combination with routine 72, routine 96 may facilitate selection of bursts that meet the minimum and maximum width criteria, respectively. In order to individually adapt the burst duration threshold level in routines 72, 96, for example, EMG data increasing above the background level and maintaining the level increased from about 25 to about 400 milliseconds. Selecting a region can be included. In routine 96, the start signal (step 98) may be performed based on, for example, a detected signal change or a signal change relative to a background level. Alternatively, the start trigger can be set to start after a predetermined interval, such as can be determined by an appropriate clock routine. In some embodiments, if a pre-qualified burst is found in another burst detection routine, such as routine 72 as described in FIG. 7, routine 96 is automatically started and is also processed by routine 72. Data can be used to further qualify.

  The routine 96 can include obtaining a signal sample as shown in step 100. In some embodiments, the routine 96 can include step 102. In step 102, the routine can determine whether the sample is greater than the maximum allowable level. If the maximum level is exceeded, the method can direct another routine, as shown in step 104. The method can also acquire the next signal sample at step 106 or wait for the next trigger before starting over. Instead of step 102 or in addition to step 102, method 96 may include step 108. Step 108 can include determining whether the sample is greater than background by at least the SNR. If the sample signal is greater than the background by at least the SNR, method 96 may perform step 110. At step 110, the method 96 may determine whether the burst duration timer is greater than the maximum burst duration level. When executing routine 96 in combination with routine 72, the burst timer may be activated in advance, for example, at step 86 of method 72, as described above. Alternatively, for example, if the routine 96 is executed separately from the routine 72, then a separate burst timer can be triggered in another step (not shown).

  If the maximum burst duration threshold is exceeded, the burst can be considered too long to indicate a seizure, as shown in step 114. With the appropriate flag cleared, the process can be restarted from the beginning by collecting the next signal sample or waiting for the appropriate trigger and starting the process again. If it is determined in step 108 that this sample has not exceeded the background by the SNR threshold level, the burst is considered to have ended, and this burst is qualified as shown in step 112. can do. As further shown in step 112, a signal qualified as a burst, for example, in some embodiments, is characterized by a confidence value, the burst center is determined, and the burst data is: It can be written to a circular buffer for further analysis. Upon exiting routine 96, the data may be sent to a routine that may be further qualified for, for example, burst characteristics and / or evaluate response to detection of burst data. For example, burst data can be sent to a circular buffer that is periodically evaluated for detection of a burst sequence, or a determination can be made as to whether an alarm should be triggered each time a burst is detected.

  FIG. 9 illustrates one embodiment of a burst sequence detection routine 120 suitable for EMG burst sequence data selection and / or processing. Burst train detection routine 120 can be used, for example, in combination with either or both of routines 72 and 96.

  In step 122 of burst sequence detection routine 120, the routine scans a circular buffer of data that can store information related to the burst, such as may have been detected in routines 72 and 96 above. Can do. In some embodiments, the scanning of the circular buffer can be initiated at predetermined intervals, such as those that can be set based on a clock routine. In other embodiments, the scan can be initiated based on a trigger signal. For example, a scan may be initiated each time such data is entered into the buffer, such as can occur whenever one of routines 72 and / or 96 sends qualified data to the buffer. As displayed in step 124, in some embodiments, the method 120 may evaluate whether the data in the circular buffer meets one or more threshold conditions. For example, in step 124, the method 120 may exclude bursts that are too close to the subject of consideration or too far from the subject of consideration. As displayed in step 126, the method 120 may determine whether an appropriate number of bursts has been selected to qualify as a burst sequence. For example, a plurality of bursts suitable for qualification as a burst train can be about 3 to 20 bursts. More generally, an appropriate number of bursts can be selected to generate statistics and / or examine the trend of the signal as the seizure progresses. For example, a plurality of burst sequences can be analyzed, and a change in data between adjacent burst sequences can be calculated. In some embodiments, burst or burst sequence information may be collected even after the alarm is initiated. For example, the length of time between bursts can be used to model or track the progression of seizures and provide useful information to the caregiver. Such information can also be used by other devices or sensors, such as those that can be used to treat seizures or to collect physiological data of patients. In step 128, the characteristics of the detected burst sequence can be analyzed for one or more features that may indicate a seizure or event associated with the seizure and an appropriate response can be performed. For example, step 128 may include an analysis of the characteristics of the burst train, such as may be implemented using the steps or elements in routine 130 as shown in FIG. As a further example, the response may be to enter a value in a monitoring algorithm that can then be used to determine if a seizure has been detected. In addition, routine 120 can be used to qualify a possible clonic burst sequence as a clonic burst sequence, and the presence of the detected clonic burst sequence can then be detected, for example, an alarm Can be used to start.

  In some embodiments, it may be useful to detect one or more burst sequences and determine whether the burst sequence indicates a seizure. For example, the burst sequence can be analyzed using one or more routines that determine, for example, burst periodicity, burst regularity, or both. For example, for some patients, a burst is characterized by some natural variation between individual bursts, and if the data is characterized by an excessively small burst periodicity, the data is It can be considered that there may be no indication of an actual seizure. In addition, for many patients, as the seizure progresses, the rest period between bursts may increase during the seizure. Thus, if a burst is evaluated over one or more periods and the periodicity of the burst increases over a period of time, the confidence that the burst data is a seizure is increased. The burst sequence can be compared to one or more thresholds related to the characteristics of the burst, and if the threshold conditions are met, the burst sequence is considered to indicate an episode Can do. The threshold condition can be based on data obtained from patients experiencing clonic seizures or based on clonic data obtained from patient demographics. Thus, the burst sequence is a qualified burst sequence and the bursts in the sequence can be qualified as clonic bursts. In some embodiments, the burst sequence can be weighted with a value that is not only related to the detection of the burst sequence, but also to the certainty of detection of the burst sequence.

  As described herein, it may be beneficial to evaluate bursts over an extended period of time. This is because using statistics obtained over a long period of time increases the confidence that a seizure is present, as well as the certain type or severity. However, it may be beneficial to detect seizures as quickly as possible, as also described herein. In light of those criteria, in some embodiments, particularly if pre-clonic seizures have been detected very recently, wait for a certain period of time before starting an alarm after the first burst is detected It can be useful. For example, as long as only motor symptoms are present for about 15 seconds to about 30 seconds (or some other suitable period), the method collects further burst data and before the alarm is triggered, The confidence in the mold and / or severity can be increased. Also, in some of those embodiments, periodic algorithms or other algorithms that calculate burst statistics may be particularly useful. This is because, for example, if a large number of bursts are measured and those bursts qualify for periodicity, it can be a major feature of seizures. Furthermore, such data can be collected for some patients and used to diagnose risk assessments, for example, at least in some patients can be associated with a previous CNS suppression effect. Such embodiments can be individually tailored to specific patient needs. Also, for some patients, an emergency alarm can be issued immediately after detecting a seizure or mesophase activity.

  FIG. 10 illustrates one embodiment of the routine 130. Routine 130 is used to characterize the periodicity of the burst to indicate whether the burst is in a seizure or stage of seizure and / or whether the burst was properly counted in the burst detection routine can do. The periodicity routine 130 stores, for example, qualified burst data, or EMG data identified as containing previously qualified burst data, for a selected period or a determined period of time. A circular buffer or register can be examined, which can investigate what the regularity period between bursts was. The periodicity routine scans the different data values from the various time windows that the burst detection algorithm writes to the circular buffer and examines the periodicity of signal features, including those that may not indicate a seizure. Can do.

  In step 132 of the exemplary method 130 of FIG. 10, the average duration between bursts over a period of time can be calculated. In step 134, the individual durations between bursts are subtracted from the average duration, and in step 136 the absolute value of the difference can be used to calculate the mean deviation of the duration. In this example, the average deviation can be converted to a percentage, but other suitable metrics may be used to characterize the variation between periods.

  In step 138, the aforementioned percentage (ie, average deviation percentage) can be compared to a threshold value. Such thresholds can be taught to the operating system and customized for a particular patient.

  For example, in a certain period (measured in seconds), 9 bursts at the following times: 12, 13, 13.75, 14.35, 15, 15.8, 16.2, 16.5, 17.4 If detected, there will be 8 periods between bursts. Therefore, there were 9 bursts with 8 periods between bursts in the whole period including the above 5.4 seconds. The average period can be calculated as 5.4 / 8 = 0.675 seconds per burst. The period between bursts is as follows.

13-12 = 1
13.75-13 = 0.75
14.35-13.75 = 0.6
15-14.35 = 0.65
15.8-15 = 0.8
16.2-15.8 = 0.4
16.5-16.2 = 0.3
17.4-16.5 = 0.9
In this example, the time when there is a burst center around can be used as a time display of such a burst by a simplified method. In other words, each time the burst algorithm qualifies a burst, the time indication can be written to the circular buffer for use by the periodic algorithm. In other embodiments, the actual burst width can be used to calculate the actual duration length between bursts. For example, if a burst occurring at 12 seconds lasts 0.02 seconds, the period between the burst starting at 12 seconds and the burst starting at 13 seconds is 0.98 seconds. The absolute value of the deviation from the average can be calculated as follows:

│1.0-0.675│ = 0.325
│0.75-0.675│ = 0.075
│0.6-0.675│ = 0.075
│0.65-0.675│ = 0.025
│0.8-0.675│ = 0.125
│0.4-0.675│ = 0.275
│0.3-0.675│ = 0.375
│0.9-0.675│ = 0.225
The averaging of absolute values can be achieved as follows.

Sum of all deviations:
0.325 + 0.075 + 0.075 + 0.025 + 0.125 + 0.275 + 0.375 + 0.225 = 1.5
Average deviation: 1.5 / 8 = 0.1875
The average deviation percentage of this average is 0.1875 / 0.675 = 27.8%. This deviates significantly from the average and can be considered unlikely to be artificial. For example, the average deviation percentage threshold can be set, for example, to 15%, and then the periodicity algorithm can declare that the data is likely to be a seizure feature. For example, as shown in steps 138 and 140, the system will not vote against declaring seizures. The result can be placed in a register and voted for alarm declaration, for example, so that the monitoring algorithm can be used. Furthermore, if the method 130 is applied to burst qualification, the method can support the finding that the burst is associated with a seizure, and the burst is then appropriate to be counted. And can be used in a burst detection routine to support a positive response.

In another simplified example, the burst sequence may appear (in seconds) as follows:
17, 17.5, 18.02, 18.51, 19.04, 19.56, 20.1, 20.6, 21.13
Thus, there were 9 bursts with 8 periods between bursts over the periodic time window including the 4.13 second period. The average period can be calculated as 4.13 / 8 = 0.51625 seconds per burst. The individual times between bursts are as follows:

17.5-17 = 0.5
18.02-17.5 = 0.52
18.51-18.02 = 0.49
19.04-18.51 = 0.53
19.56-19.04 = 0.52
20.1-19.56 = 0.45
20.6-20.1 = 0.5
21.13-20.6 = 0.53
The absolute value of the deviation from the average can be determined as follows.

│0.5-0.51625│ = 0.01625
│0.52-0.51625│ = 0.00375
│0.49-0.51625│ = 0.02625
│0.53-0.51625│ = 0.01375
│0.52-0.51625│ = 0.00375
│0.45-0.51625│ = 0.06625
│0.5-0.51625│ = 0.01625
│0.53-0.51625│ = 0.01375
The sum of all deviations can be calculated as follows:

0.01625 + 0.00375 + 0.02625 + 0.01375 + 0.00375 + 0.06625 + 0.01625 + 0.01375 = 1.6
Therefore, the average deviation is 1.6 / 8 = 0.02.

  Thus, the average percent deviation is 0.02 / 0.51625 = 3.87%. Thus, this example shows a very regular pattern. For example, if the average deviation threshold is set to 15%, the algorithm will declare that the true seizure is very unreliable and will vote against declaring the seizure alarm. (Step 142). The result is placed in a register so that the monitoring algorithm can be used. Similarly, if the method 130 is applied to burst qualification, the method can support the finding that the burst is not associated with a seizure, and the burst is now considered uncountable. It is considered appropriate and may not be used in a burst detection routine to support a positive response. Of course, standard deviation calculations or other suitable metrics related to variability can be used in place of the average deviation calculation. The threshold can also be obtained from a particular patient, an average model value, or some other method. These threshold values vary depending on the detection unit, and can be changed as appropriate.

  In some embodiments, the data can be automatically formatted and presented for review in a manner suitable for displaying the determined risk value. For example, in some embodiments, EMG data can be sent and qualified or labeled as merely representing a warning event. That is, the data can be risk stratified and specified in non-emergency or warning conditions. In order to identify the EMG data as being associated with a warning or non-emergency condition, the EMG data can be labeled with an appropriate status indicator, for example. For example, graphical data can be displayed in color, font, or some other indicator that identifies the status of the data as being only relevant to warnings. Similarly, emergency data can be appropriately labeled, for example, in different colors, fonts, or other signs.

  Some caregivers prefer to be notified immediately and send EMG data for warning events (or like some patients), but other caregivers or other patients Some people do not like such notifications. Also, in some embodiments, marking an event as being associated with a warning condition may include instructions to store data in a buffer without actively displaying or notifying the caregiver, for example, the caregiver's Instructions for storing data in the memory of the computer can be included. Also, in some embodiments, the system is a caregiver or other designated personal device that installs a smart client application that facilitates the organization of incoming data, etc. Can include installing. For example, the information can be transmitted to a computer in some embodiments or using some device settings, but is not presented to the user until, for example, confirmation that the event requires an emergency response. If the data is validated, a signal can be sent from the patient detection device (or base station) to update the event state. Also, for example, data stored “hiddenly” in the caregiver's computer can then be immediately displayed based on an update signal with the minimum amount of data containing only instructions for the updated state. Such an embodiment may be particularly useful when, for example, the caregiver is in a location where the connection is intermittent and data transfer may be restricted. In such a case, it is difficult to send a large amount of data promptly, and it may not be desirable or timely to send a large amount of data only after an emergency confirmation.

  The data transmitted from the monitoring system can be customized for a particular individual or recipient group in some embodiments. For example, the transmitted data may include alarm messages, a subset of statistical information related to algorithm detection, time domain or frequency domain EMG data, or a combination thereof. For example, in some embodiments, the detection unit can execute one or more peak detection programs as part of the burst detection routine. The burst detection routine can be configured to identify whether one or more bursts of data are present that can be used to determine that a secular phase portion is present. . Thus, the burst detection routine may be ideally suited for risk stratification and automatic classification of detected event data. Also, as part of the arrangement for transmitting EMG data, a burst statistics window can be provided to present the data to the caregiver. The burst statistics window can include, for example, a summary of statistical information regarding the detected pattern of bursts. By way of example only, the data that may be included in the burst statistics window includes the number of detected bursts, the rate of burst detection, the average signal-to-noise ratio (SNR) of the detected bursts, the SNR spread of the detected bursts, the detection Average width of detected bursts, spread of detected bursts, average length of periods between detected bursts, spread of length of periods between detected bursts, bursts over any number of burst sequences Includes deviation, frequency characteristics of burst data, and combinations thereof.

  More generally, for example, depending on the particular routine or set of routines used to detect seizure events, other processed or unprocessed EMG data (eg, raw EMG data files) Analytical information organized along with data for reconstruction can include a summary of data useful to the caregiver that indicates what triggered the transmission. That is, the organization of the transmitted data can include a summary of information that immediately tells the caregiver why the particular EMG data was selected. In addition, for example, a data portion that is positively responded to the routine can be identified. Also, for example, summaries of the output of individual detection routines can be presented including threshold settings for routines and / or accuracy values to meet a given threshold event. As a non-limiting example, some of the routines described herein may include analysis of EMG data for sustained amplitude increases, transient EMG increases or bursts, and spectral separation of EMG data. it can. In addition, related information can be organized and presented according to what seizure features are considered to exist.

  In some embodiments, the transmitted data may include EMG data and / or one or more other data. For example, in some embodiments, the detection unit can further include one or more microelectromechanical inertial detection elements, such as a gyroscope, magnetometer, and / or accelerometer. Such a detection element can be configured to determine the orientation of the detection unit and thus the orientation of the muscle to which the detection unit is worn. Thus, the system described herein can include, for example, a description of whether the sensor (and thus the muscle to which the sensor can be coupled or worn) was oriented in some direction. For example, whether the detection unit is oriented in a substantially vertical direction, e.g. in a direction parallel to the normal vector from the ground as normally seen in standing patients, or the sensor is horizontal to such normal Indicates whether the patient is oriented perpendicular to such a normal vector, as normally seen when the patient is lying on a bed or the like. Other information that can be organized along with the EMG data can include other sensor data as may be relevant to blood oxygen saturation measurements. In some embodiments, such information can be used to further influence further classification of events, eg, determined risk stratification. Such information can also be used with EMG data to designate an event as either an emergency event or a warning event. In addition, in some embodiments, a patient may be monitored using a pulse oximeter. A pulse oximeter may be included in the detection device. In some embodiments, risk stratification may further comprise separately analyzing patient risk due to falls and / or patient SUDEP risk. They can also be associated with a patient condition, respectively. For example, whether the patient is asleep, alone at home, or alone with another patient at home. By way of example only, if the patient is asleep, the template file automatically assigns an assessment that the risk of falling is low, but may not adjust the risk of SUDEP.

  In some embodiments, organizing information for transmission can include tailoring data transmission individually for various designated individuals, including caregivers, friends, and family members. For example, data suitable for displaying either or both of time-dependent EMG data and / or spectrum-dependent EMG data, as well as transfer of summary of important analytical features can be transmitted to a specific caregiver However, such information may not be necessary or harmful to some friends and family members. Thus, in some embodiments, the organization of information for transmission can be determined by a profile or setting adapted to the recipient. For example, friends and / or family members are notified that an emergency alarm or warning has been sent, but they are time-dependent EMG data and / or spectrum-dependent EMG data playback or other suitable for data analysis In some cases, the raw or processed data of the aspect is not received. Also, in some embodiments, as non-limiting examples, whether specific phases of seizures are present, detailed data on key seizure characteristics such as various aspects of seizure symptomology, and / or microelectronics Information can be sent to a group of trained caregivers accessing from orientation data from mechanical sensor elements. It is also possible to send information to other caregivers, including some caregivers involved in emergency responses. Other caregivers may not have received another specific training in EMG signal interpretation. Such caregivers can be sent, for example, only information regarding whether the emergency response is valid, or other sensor data trained for such caregivers to interpret the data.

  Data selected and organized for transmission can be transmitted and received by one or more caregivers, or other designated individuals. As described above, in some embodiments, EMG sensor data or other sensor data may be sent from either or both of the detection unit 12 or the base station 14, or from the alarm transceiver 16. Also, any of the devices 12, 14, and 16 can communicate information to a designated personal device, as shown in FIG. In some embodiments, the information can communicate directly or alternatively to a caregiver or other designated individual over a local network such as WiFi.

  In some embodiments, the caregiver's device, in some embodiments, includes a set of one or more programs or instructions installed in the hardware or software, so that signal data from the transmission device can be transmitted. Communication and organization can be facilitated. Also, in some embodiments, instructions for executing one or more routines can be performed using one or more smart client applications. The instructions for updating the state of the detected event may be included in the caregiver's device, for example, and updated by computer management based on one or more signals or based on an update signal. Similarly, instructions for performing any of a variety of routines, including routines that may be executed by either detection unit 12 or base station 14, may be included in the caregiver's device application. . Also, the caregiver may change threshold settings and / or manually set threshold values or the like, which may be used to further access EMG data and / or other data supplied from detection unit 12 or base station 14, for example. It may be possible to adjust other routine settings.

  FIG. 11 illustrates another exemplary embodiment of a method 160 for collecting EMG signals and transmitting selected EMG signal data to monitor a patient for seizure activity. In step 162, a portion of the EMG signal, or in some embodiments, an EMG signal and other sensor data may be collected. For example, in some embodiments, EMG signals are collected along with oximeter data and / or data from one or more orientation sensitive devices and / or position sensitive devices, or both. can do.

  At step 164, the collected data can be analyzed using one or more routines configured to determine whether one or more events have been detected. As noted above, at least one routine may be selective for a particular part of the seizure and may facilitate risk stratification if such part is related in part to the adverse effects of the possible seizure. . At step 166, a determination can be made as to whether a seizure event has been detected, for example, based on any number of EMG analysis routines and / or other information. Individual routines can be configured to be selective for a given part of the seizure. For example, in some embodiments, a first routine can analyze the collected data for the presence of persistent EMG amplitude, and a second routine allows the patient to experience a seizure phase. When collected, the collected data can be analyzed for the presence of activity that may be selectively present or a temporary increase in burst activity. In some embodiments, the event can also be determined using a monitoring algorithm. The monitoring algorithm can execute a plurality of subroutines that combine the contributions of various seizure variables in a routine to determine the likelihood that a seizure can occur, such as by weighting appropriately. Also, a routine that includes a monitoring algorithm, for example, by adjusting the contribution from one or more burst detection subroutines can be a selective routine. In some embodiments, other sensor data can be incorporated into the EMG sensor data during the execution of step 164. For example, position data and / or orientation data may be provided in some embodiments by a detection unit or in combination with a detection unit and a base station and / or environmental transceiver.

  Also, in some embodiments, the system may notice that the patient may be in one place, such as a bed, or some other place, such as a bathroom. The system then engages in some other activity such as, for example, applying different routines and / or routine settings so that the patient is having a seizure or a non-seizure move depending on the user's location The likelihood of doing so can be determined. Also, in some embodiments, as described in Applicant's co-pending US Provisional Patent Application No. 61 / 979,225, the setting or threshold is set so that the detection unit is oriented horizontally. Or whether it is oriented vertically. That is, the system can be specifically calibrated as to whether the patient's muscles are in a particular direction. It can also be calibrated specifically based on, for example, whether the patient is lying horizontally, whether the patient is lying horizontally with weight on the sensor, or facing vertically. In some embodiments, the monitoring system can be configured to allow the patient to update their status based on yet another selectable profile. For example, the patient may have one or more different monitoring settings based on, for example, whether the patient is sleeping in bed, whether the patient is home alone, or whether the patient is home with another person. An option to select may be given. Also, depending on whether the patient selects one of those options, the system can select different settings and / or thresholds. Further, for a particular response or group of responses, the monitoring system may select a transmission protocol associated with any of the emergency alerts that depend on the selected option, for example.

  Any of the various routines described in this specification or, for example, applicant's co-pending US Provisional Patent Application No. 61 / 969,660, is whether an event is present in the collected EMG signal. (Step 164) and then can be performed to determine if a seizure has been detected (step 166). For example, a burst is qualified on the basis of meeting minimum burst width and / or maximum burst width criteria, and detects a seizure event or a possible seizure event if a certain number of bursts are detected over a period of time can do. For example, if about 2 to about 6 bursts are measured within a time window of about 1 second, or if another suitable number of bursts are counted in another suitable time window, the burst routine May show a positive response. In some embodiments, the burst routine may show a positive response if at least about 2 to about 6 bursts are measured within a time window of about 2 seconds to about 5 seconds. Also, for example, seizures can be detected based on a positive response in a burst routine.

  As shown in step 168, the data is transferred and stored in one or more buffers of accessible memory and maintained there for a desired duration. The next portion of the EMG signal and / or the next portion of the EMG signal and other signals can then be collected. Upon completion of the desired duration interval of EMG signal collection, the data stored in the buffer can be transferred to another unit of memory, such as to a fixed storage device, as shown in step 170. More generally, events can be configured such that the transfer of data between one buffer and another storage device can configure system 160 to maintain an appropriate portion of the data in accessible memory. If detected, the data can be used for transmission. That is, if it is deemed necessary for seizure analysis and / or event categorization, or if a transmission protocol is later selected and such data is included for subsequent transmission, Can be kept available.

  As shown in step 172, if a seizure event or a possible seizure event is deemed to exist, the method 160 may include categorizing the collected EMG signal based on type and / or intensity. Also, in some embodiments, the categorized EMG data can then be updated with status indicators such as warnings or emergency conditions. As shown by method 160 and arrow 161, in some embodiments, one or more portions of data buffered in memory may be accessed when categorizing seizures. That is, when categorizing seizures, data previously sent to the buffer can be taken into account.

  In some embodiments, seizure categorization can include determining whether one or more routines responded to a portion of the EMG signal. For example, as shown in Table 1, detected events can be categorized based on a particular combination of responses to one or more routines. In some embodiments, seizure categorization can include accessing data from any combination of oximeters, orientation sensors, or heart rate sensors. Also, for example, in some embodiments, the EMG data only indicates the presence of a warning event, but the critical level of saturated oxygen has been determined or the patient orientation has changed from substantially vertical to substantially horizontal. In either case of a change in direction, an emergency condition can be determined instead. For example, risk stratification can be based on either a determination that the patient may be falling or a determination that the patient is showing early signs of SUDEP.

  In some embodiments, as a non-limiting example, detected based on various metrics such as seizure type, intensity, duration, duration of seizure phase, other metrics, and combinations thereof Seizure events or possible seizure events can be classified. Classification of seizure severity can include, for example, normalizing seizure metrics to a single patient or to typical values of patient demographics. For example, for a patient, the amplitude measurement of the detected feature is a factor of a previously measured value for the feature (eg, during another episode of that patient), or If the average feature size is a factor, this factor can be used to grade the severity of the seizure. Such information may be sent, for example, to a caregiver, used to determine an appropriate response, or both.

  As described further with respect to step 172, data including EMG data and / or other sensor data may be selected for transmission. Also, the selected protocol can be executed in step 174. As discussed above, the decision to choose and whether to transmit data may depend on the categorization of the collected data. Also, in some embodiments, the transmission protocol can include an instruction to send information to all or only some of the designated individuals. That is, in some embodiments, an event can be detected and sent to only some caregivers or quickly. For example, a routine may identify that a possible seizure has been detected, at which time data may be sent to one or more remote system users but not to other designated individuals. is there. One or more users of the remote system, for example, are trained to interpret EMG signals and / or other signals and have actual seizures using any of a variety of protocols May include one or more individuals who can confirm

  For example, the remote user can determine the patient's condition based on the video monitor 9. For example, if video data is not available, the remote user can attempt to contact the patient, for example, by calling the patient. However, in some cases, the remote user may not have access to the video data, cannot call the patient to contact the patient, or does not want to contact the patient in this manner. possible. For example, the detected event, i.e., the event detected for a particular patient, is only one of the low risk events, or the event is likely to be a false positive detection. If such an event is common for the patient, it may be burdensome and undesirable to contact the patient repeatedly based on false detections. In some embodiments, the user performs further analysis of the data and attempts to conclude if the event is really urgent, and then if the event is deemed insignificant, the patient Attempt to contact or update the state of the event, or both contact and update. For example, the user can now contact or initiate contact with other designated individuals, including other caregivers, friends and family. Thus, in some embodiments, and in some situations, the caregiver's device, eg, mobile phone 17, PDA 18, or other client device contact is mediated via a remote system user There is also.

  In general, the device of the seizure detection system can be of any suitable type and configuration that achieves one or more of the methods and goals disclosed herein. For example, the server may comprise one or more other computers or programs or one or more computers or programs that respond to commands or requests from clients. The client device may comprise one or more other computers or programs or one or more computers or programs that issue commands or requests for services provided by the server. The various devices of FIG. 1, for example 12, 13, 14, 16, 17, 18, and / or 19, can be servers or clients depending on the function and configuration. Servers and / or clients include, for example, mainframe computers, desktop computers, PDAs, smartphones (Apple iPhone ™), Motorola's Atrix ™ 4G, Motorola ( Motorola's Droid (TM), Samsung's Galaxy S (TM), Samsung's Galaxy Note (TM), and Research In Motion Blackberry ™ devices, etc.), tablets (Sony's Xperia (trade) ), Samsung Galaxy Tab ™, Amazon Kindle ™, etc., netbooks, portable computers, portable media players with network communication capabilities (Microsoft) Zune HD (TM) and Apple's ipod Touch (TM) devices), cameras with network communication functions, smart watches, wearable computers, etc. May be.

  The computer can be any device that can accept input, process the input according to a program, and generate output. The computer can include, for example, a processor, a memory, and a network connection function. Computers can be of various types, such as supercomputers, mainframes, workstations, microcomputers, PDAs, and smartphones, depending on the size, speed, cost, and function of the computer. Computers can be fixed or portable and must be programmed for various functions such as mobile phone, media recording and playback, data transfer, web browsing, data processing, data query, process automation, video conferencing, artificial intelligence, etc. Can do.

  The program can be any sequence of instructions, such as an algorithm, whether it is in a computer-executable format (object code), a human-readable format (source code), or otherwise. Can be included. A program can include or call one or more data structures and variables. The program can be embodied in hardware or software, or a combination thereof. The program can be created using any suitable programming language such as C, C ++, Java (registered trademark), Perl, PHP, Ruby, or SQL. The computer software can include one or more programs and associated data. Computer software includes, for example, system software (such as operating system software, device drivers, and utilities), middleware (such as web servers, data access software, and enterprise messaging software), application software (such as databases, video games, and media players). , Firmware (such as calculators with device-specific software, keyboards, and cell phones), and programming tools (such as debuggers, compilers, and text editors).

  The memory may comprise any computer-readable medium on which information is stored temporarily or permanently. Examples of the memory include various RAMs and ROMs such as SRAM, DRAM, Z-RAM, flash, optical disk, magnetic tape, punch card, and EEPROM. The memory can be provided virtualized, such as with RAID technology, to one or more devices and / or geographical locations or the entirety thereof. An I / O device may comprise any hardware that can be used to provide information to a computer and / or receive information from a computer. Exemplary I / O devices include disk drives, keyboards, video display screens, mouse printers, printers, card readers, scanners (such as barcodes, fingerprints, irises, QR codes, and other types of scanners). RFID devices, tape drives, touch screens, cameras, motion sensors, network cards, storage devices, microphones, audio speakers, touch pens and transducers, and related interfaces and drivers.

  Networks are virtually switched, routed, fully connected cellular networks, the Internet, intranets, local area networks (LANs), wide area networks (WANs), metropolitan area networks (MANs) : Metropolitan Area Network), including other types of area networks, cable TV networks, communication satellite networks, telephone networks, public networks, private networks, wired or wireless networks, and any combination thereof and sub-networks thereof it can. The network can use various network devices such as routers, bridges, switches, hubs, repeaters, converters, receivers, proxies, firewalls, translators and the like. The network connection is wired or wireless, and a multiplexer, a network interface card, a modem, an IDSN terminal adapter, a line driver, or the like can be used. The network can include any suitable topology, such as point-to-point, bus, star, tree, mesh, ring, and any combination or hybrid thereof.

  Wireless technology can be used for ISM band devices, WiFi (registered trademark), Bluetooth (Bluetooth (registered trademark)), mobile phone SMS, cellular (CDMA2000, WCDMA (registered trademark), etc.), WiMAX (registered trademark), WLAN, etc. It can take various forms, such as person-to-person wireless, person-to-fixed receivers, person-to-remote alarm devices, using one or more of various wireless technologies.

  Communication within or between computers, I / O devices and network devices can be accomplished using various protocols. Protocols can include, for example, signaling, error detection and correction, data formatting, and address mapping. For example, the protocol can be provided according to a 7-layer open system interconnection model (OSI model) or a TCP / IP model.

  Additional information related to the methods and apparatus described herein can be understood in connection with the examples described below.

  In this Example 1, multiple patients prone to seizures were monitored for seizure activity using EMG electrodes. A sensor was placed on the patient's biceps, EMG signals were collected, the collected signals were analyzed for the presence of seizure activity, and seizures were detected. The data in this Example 1 herein includes a summary of 20 different seizures measured from a total of 11 separate patients. The seizures and patients described herein include a plurality of patient subsets in studies that evaluate various seizure detection methods. Further, in each of the 20 seizures in Example 1, the seizure phase portion of the seizure was identified. EMG data for one of the recorded seizures is shown in FIG.

  EMG data obtained from various parts of the recorded seizures was processed to determine the average duration between bursts as well as the average duration of burst duration. Burst data was obtained from several parts of the recorded data, respectively. For example, average burst width and average timing between bursts were analyzed during time intervals near the beginning, middle and end of the clonic phase. Table 2 shows the results for various patients and seizures in this study during a period near the beginning of the clonic phase.

  The average duration burst duration in the early part of the mesophase for patients in this study was about 0.12 seconds (ie, 120 milliseconds). Also, bursts can be detected by selecting a threshold for a minimum burst width of about 25 milliseconds to about 75 milliseconds and a maximum burst width threshold of about 250 milliseconds to about 400 milliseconds or less. Further, other signals that may be present during monitoring, eg, non-seizure causes and / or other phases, generally did not follow this pattern with significant frequencies. Also, by selecting an increase that meets the aforementioned width requirement, bursts are counted and can be used to determine if there was clonic activity. By detecting clonic activity, for example, seizures can be distinguished from other seizure events that do not exhibit clonic activity. In addition, seizure activity can be highly detrimental to seizures, so seizures can trigger appropriate responses such as emergency situations.

  Additional data for the middle and end part of the clonic phase is shown in Table 3 (intermediate part) and Table 4 (end part).

  For some of the seizures monitored, the late part of the activity was mild and / or the seizure ended quickly. Thus, for some of the measured seizures, only the early period of activity was measured. That is, depending on the patient and / or seizure, only the initial or intermediate portion was measured.

  This Example 2 describes a specific embodiment of a method for analyzing collected EMG signals. For example, such methods can be used to analyze the collected seizure data for patients in the aforementioned studies. In Example 2, a first routine can be executed to analyze EMG data for the presence of persistent EMG activity. The presence of this persistent EMG activity may indicate early motor symptoms of seizures. A second routine may be executed for the presence of persistent EMG activity, but the second routine may include a higher threshold than the first routine. It is also possible to execute a third routine that is configured to be selective for clonic activity. The response can be performed simultaneously on a given portion of EMG data. Also, in some embodiments, a warning period can be initiated if certain conditions are met.

  Table 5 shows, by way of example, several settings that can be applied to monitor a patient in the above study using the first routine of Example 2.

  Table 6 shows, by way of example, several settings that can be applied to monitor a patient in the above study using the second routine of Example 2.

  Table 7 shows, by way of example, several settings that can be applied to monitor a patient in the above study using the third routine of Example 2.

  Table 8 shows one embodiment of how the detection of different combinations of the first routine, the second routine, and the third routine can be arranged.

  Using the routine settings of Example 2, multiple positive responses can be made, some of which are due to non-seizure causes. However, it is not necessary to initiate an emergency alarm for every combination of responses. For example, detection can only initiate a warning, not an emergency response (this warning may or may not be given to an attendant or remote caregiver). Alternatively, the system can only trigger an analysis warning period for some responses and trigger an emergency response only after completion of such warning period. For example, as shown in Table 8, if either the first routine or the second routine responds, a warning period of about 20 seconds can be started.

  For example, as shown for events A and B in Table 8, if triggered to indicate a warning period, the system continues to capture EMG data and further classify the responses that can be made during the warning period Can do. For example, Table 9 shows one embodiment of how responses to various events (EH) obtained during a warning period can be considered and the possible results for those responses.

  For example, if routine 2 maintains one or more responses or further responses during the warning period, it may be considered that an emergency transmission protocol should be performed. Similarly, if Routine 3 shows a positive response within the warning period, it may be a tonic-clonic seizure and an emergency alarm can be executed. In some embodiments, other sensors can be triggered. For example, as shown in Table 9, a pulse oximeter sensor can be triggered. Also, for example, in some embodiments, an emergency response can be adjusted if the pulse oximeter indicates that the patient is under further stress. For example, an emergency transmission can be transmitted to a caregiver, such as a family member, or other caregivers who may be in another room of the patient's home. However, if the saturated oxygen level and / or patient pulse indicates that the patient is under physiological stress, the EMT worker can be contacted immediately.

  Although the disclosed method and apparatus and their advantages have been described in detail, it will be understood that various changes, substitutions and modifications may be made without departing from the invention as defined by the appended claims. Should. Furthermore, the scope of this application is not intended to be limited to the specific embodiments of the processes, machines, products, compositions, or materials, means, methods, and steps described herein. For example, use of the term “include” should be construed to be the term “comprising”, ie, open-ended. As can be readily appreciated from the present disclosure, existing or later developed processes that achieve substantially the same function or achieve substantially the same results as the corresponding embodiments described herein. , Machines, products, material compositions, means, methods, or steps may be utilized. Accordingly, the appended claims are intended to include within their scope such processes, machines, manufacture, compositions of matter, means, methods, or steps.

Claims (42)

A method of monitoring a patient for seizure activity, the method comprising:
Collecting EMG signal data;
Processing the collected EMG signal data to determine whether a detected event exists;
Categorizing the detected event and selecting a transmission protocol included in a group of selectable transmission protocols based on the categorization;
Transmitting a message to a remote user identifying whether seizure activity may exist;
The method wherein the format of the message depends on the selected transmission protocol.   The method of claim 1, wherein at least one of the group of selectable transmission protocols identifies an emergency event.   The method of claim 1, wherein at least one of the group of selectable transmissions identifies a warning event.   The method of claim 1, wherein at least one of the group of selectable transmission protocols includes an instruction to send the collected EMG signal data to the remote user.   The method of claim 1, wherein at least one of the group of selectable transmission protocols includes an instruction to send a message notifying the detected event to the remote user. The processing of the collected EMG signal data comprises:
Identifying whether an increase in signal amplitude is present in the collected EMG signal;
Qualifying the increase based on a minimum width and a maximum width;
Counting the number of increases;
Determining whether the number of increases is indicative of the presence of a seizure portion of seizures;
The method of claim 1, comprising logging a detection event if a seizure portion of the seizure is present.   The method of claim 1, further comprising displaying the message for the user on a remote device, wherein the message is displayed to inform the user whether a warning or emergency event has been detected. .   The method of claim 1, further comprising displaying the message for the user on a remote device, wherein the message is displayed to inform the user whether a seizure phase has been detected. Method.   The method of claim 1, wherein at least one of the group of selectable transmission protocols identifies an emergency event and another one of the group of selectable transmission protocols identifies a warning event. A method of monitoring a patient for seizure activity, the method comprising:
Monitoring the patient by collecting EMG signals;
Processing the collected EMG signal to identify whether an increase in signal amplitude is present in the portion of the EMG signal;
Qualifying the portion of the EMG signal based on one or more characteristics of seizure phase activity;
Determining whether the portion of the qualified EMG signal is indicative of seizure activity.   The method of claim 10, wherein the step of qualifying a portion of the EMG signal comprises determining whether the increase meets at least one threshold for a maximum time width or a minimum time width.   Qualifying the portion of the EMG signal includes counting the number of increases satisfying the at least one threshold, and the portion of the qualified EMG signal is indicative of seizure activity. The method of claim 11, wherein the step of determining whether or not comprises comparing the number of increases to a threshold number of increases.   The method of claim 10, wherein the minimum time width is about 25 milliseconds to about 100 milliseconds, and the maximum time width is about 250 milliseconds to about 400 milliseconds.   The step of qualifying the portion of the EMG signal is substantially stationary whether the increase has a particular time width and the increase lasts for a period of about 50 milliseconds to about 300 milliseconds. The method of claim 10, comprising determining whether the signal has adjacent periods of time.   Analyzing the collected EMG signal to determine whether a region of persistent EMG amplitude is present in the collected signal; and wherein the persistent EMG amplitude is indicative of seizure activity. 11. The method of claim 10, further comprising the step of determining whether or not. The step of analyzing the collected EMG signal comprises:
Comparing the amplitude of the EMG signal with a threshold amplitude value;
16. The method of claim 15, comprising: performing a response if the collected EMG signal exceeds the threshold amplitude value over a sustained period of about 1 second to about 5 seconds.   16. The method of claim 15, wherein the response is the start of a warning period for further evaluation of EMG signals for seizure activity.   11. The method of claim 10, further comprising sending a message to a caregiver that a seizure portion has been detected if the seizure activity is present.   The method of claim 16, further comprising sending a message to the caregiver that either the tonic or clonic part of the seizure has been identified. A method of monitoring a patient for seizure activity, the method comprising:
Monitoring the patient by collecting EMG signals;
Processing the collected EMG signal to determine if there is an increase in the signal;
The number of bursts if the increase is qualified based on one or more characteristics of seizures and the increase is qualified as satisfying the one or more characteristics A step of determining
Comparing the number of bursts to a threshold;
Performing an alarm response if the number of bursts exceeds the threshold.   21. The characteristic of claim 20, wherein the one or more characteristics are selected from the group of characteristics consisting of a burst duration, a duration of time between bursts, a burst periodicity, a burst regularity, and a burst waveform. Method.   21. The one or more characteristics is a burst duration, and the qualification includes comparing the burst with a minimum value for the burst duration and a maximum value for the burst duration. the method of.   23. The method of claim 22, wherein the minimum burst duration is about 25 to about 100 milliseconds, and the maximum burst duration is about 250 milliseconds to about 400 milliseconds.   24. The method of claim 23, wherein the one or more characteristics further include a duration of a quiesce period adjacent to the burst, wherein the burst is adjacent to a quiesce period of about 50 milliseconds to about 200 milliseconds. .   Analyzing the collected EMG signal to determine whether a region of persistent EMG amplitude is present in the collected signal; and determining whether the persistent EMG amplitude is present 21. The method of claim 20, further comprising:   26. The method of claim 25, further comprising sending a warning alarm if the persistent EMG amplitude is present.   21. The method of claim 20, wherein the alarm response is an emergency alarm.   26. The method of claim 25, further comprising sending a message to the caregiver that either the tonic or clonic part of the seizure has been identified. A method of monitoring a patient, the method comprising:
Monitoring the patient by collecting EMG signals;
Processing the collected EMG signal to determine if there is an increase in the signal;
One or more if the increase is qualified based on one or more characteristics of seizures and if the increase is qualified as meeting the one or more characteristics Determining burst statistics for
Comparing the burst statistics to a threshold;
Performing an alarm response if the burst statistic exceeds the threshold.   30. The individual bursts are qualified based on minimum burst width and maximum burst width that are characteristic of clonic activity, and the threshold is the number of qualified bursts. Method.   32. The method of claim 30, wherein the minimum burst width is about 25 to about 100 milliseconds and the maximum width is about 250 milliseconds to about 400 milliseconds.   Analyzing the collected EMG signal to determine whether a region of persistent EMG amplitude is present in the collected signal; and determining whether the persistent EMG amplitude is present 30. The method of claim 29, further comprising:   30. The method of claim 29, further comprising collecting signal data from one or more orientation sensors.   34. The one or more orientation sensors are part of a detection device, the detection device further comprising one or more EMG electrodes configured to perform the collection of the EMG signal. The method described.   30. The method of claim 29, further comprising collecting signal data from one or more pulse oximeters.   36. The one or more pulse oximeters are part of a detection device, the detection device further comprising one or more EMG electrodes configured to perform collection of the EMG signal. The method described in 1.   36. The method of claim 35, wherein a threshold setting for heart rate is adjusted based on whether the alarm response is made.   30. The method of claim 29, wherein the alarm response includes triggering data collection using a pulse oximeter.   Analyzing the collected EMG signal to determine whether a region of persistent EMG amplitude is present in the collected signal, wherein the alarm response is an emergency response, and the persistent 30. The method of claim 29, wherein another response is initiated if EMG amplitude is present.   40. The method of claim 39, further comprising selecting a protocol for transmission based on whether the alarm response or the another response has been initiated.   40. The method of claim 39, further comprising triggering data collection using a pulse oximeter if any emergency response is made or if the other response is initiated.   40. The method of claim 39, further comprising triggering collection of data using a pulse oximeter if more than one other response is initiated.