DE102019104798A1 - APPARATUS AND METHOD FOR THE PASSIVE ARRHYTHERMETIC DETECTION BASED ON PHOTOPLETHYMOGRAM (PPG) INTERVALS BETWEEN THE HITCH AND MORPHOLOGY - Google Patents

Abstract

A method of event detection in a user wearable device (100) includes: receiving, from a first sensor (152) implemented in the user-portable device (100), photoplethysmogram (PPG) signals; Processing (204), in a processor (130), the PPG signals to obtain PPG signal samples; Detecting (206), in the processor (130), beats in the PPG signal samples; Dividing (208) the PPG signal samples into PPG signal segments; Extracting (210) at least one Inter-Beat Interval (IBI) feature in each PPG signal segment; Classifying (214), in the processor (130), each PPG signal segment using the extracted IBI feature associated with the PPG signal segment and using a machine learning model; in response to the classifying (214), generating, in the processor (130), an event prediction result for the PPG signal segment based on the extracted IBI feature, and displaying the event prediction result on the user wearable device. In another embodiment, the method further includes extracting (212) morphology-based features.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS

This application claims the priority of the provisional U.S. Patent Application No. 62 / 648,821 entitled PASSIVE ARRHYTHMIAS DETECTION BASED ON PHOTOPLETHYSMOGRAM (PPG) INTER-BEAT INTERVALS AND MORPHOLOGY, filed March 27, 2018, which is hereby incorporated by reference for all purposes herein.

This application is a sub-treatment of U.S. Patent Application No. 15/145 . 356 entitled METHOD AND APPARATUS FOR TRIAGE AND SUBSEQUENT ESCALATION BASED ON BIOSIGNALS OR BIOMETRICS, filed May 3, 2016, which is hereby incorporated by reference in its entirety for the purposes hereof.

TERRITORY

The present disclosure relates to a medical monitoring device and a method thereof, and more particularly relates to a system and method for arrhythmia detection.

BACKGROUND

Cardiac arrhythmia, also known as cardiac arrhythmia or irregular heartbeat, is a group of ailments in which the heartbeat is irregular, too fast, or too slow. While most types of arrhythmia are not severe, some predispose individuals to complications such as stroke or heart failure. Others can lead to cardiac arrest. For example, atrial fibrillation (AFib) is one of the most common cardiac arrhythmias and the onset of AFib could potentially lead to significant health risks. Conventionally, cardiac arrhythmias are detected by an electrocardiogram (ECG) or a Hotler monitor.

ECG measurements require sophisticated detection devices with multiple electrodes attached to the patient and require active involvement of humans. Typically, ECG measurements are only made for diagnostic purposes after the patient has shown symptoms. The photoplethysmogram (PPG) has been described as an alternative to ECG in arrhythmia detection. However, some conventional PPG measurement techniques for cardiac arrhythmias rely primarily on heartbeat or heart rate detection using an average heart rate, such as a 5 to 20 second window. The average heart rate variation is not a reliable indication of arrhythmia.

Other PPG-based methods for detecting cardiac arrhythmias lack sensitivity and specificity. In addition, some methods require a large continuous volume of a PPG signal, such as a 30 second continuous PPG signal, to make the measurement.

The exemplary embodiments of the present disclosure may provide a PPG-based arrhythmia detection apparatus and method that enable passive and asymptomatic arrhythmia detection.

SUMMARY

The present disclosure discloses an apparatus and method for arrhythmia detection substantially in the form described below, such as in conjunction with at least one of the figures, as more fully set out in the claims.

These and other advantages, aspects and novel features of the present disclosure, as well as details of an illustrated embodiment thereof, will be better understood from the following description and drawings.

In one embodiment, a method for event detection in a on the body of a user-portable device, receiving, from a first sensor implemented in the user-portable device of a user, photoplethysmogram (PPG) signals; Processing, in a processor, the PPG signals to obtain PPG signal samples; Detecting, in the processor, beats in the PPG signal samples; Dividing the PPG signal samples into PPG signal segments; Extracting at least one Inter-Beat Interval (IBI) feature in each PPG signal segment; Classifying, in the processor, each PPG signal segment using the extracted IBI feature associated with the PPG signal segment and using a machine learning model; in response to the classifying, generating, in the processor, an event prediction result for the PPG signal segment based on the extracted IBI feature; and displaying the event prediction result on the user wearable device.

In another embodiment, a method of event detection in a user wearable device includes: receiving, from a first sensor implemented in the device worn on a user's body, photoplethysmogram (PPG) signals; Processing, in a processor, the PPG signals to obtain PPG signal samples; Detecting, in the processor, beats in the PPG signal samples; Dividing the PPG signal samples into PPG signal segments; Extracting at least one morphology-based feature in each PPG signal segment, the morphology-based feature being associated with statistical properties of the PPG signal samples or waveform characteristics of the PPG signal samples; Classifying, in the processor, each PPG signal segment using the extracted morphology-based feature associated with the PPG signal segment and using a machine learning model; in response to classifying, generating, in the processor, an event prediction result based on the extracted morphology-based feature; and displaying the event prediction result on the user wearable device.

In another embodiment, an apparatus includes a sensor module that includes a first sensor configured to measure photoplethysmogram (PPG) signals; and a processor comprising a data processing module configured to process the PPG signals to obtain PPG signal samples, detect beats in the PPG signal samples, and subdivide the PPG signal samples into PPG signal segments; an inter-beat interval detection module configured to extract at least one inter-beat interval (IBI) feature in each PPG signal segment; a morphology detection module configured to extract at least one morphology-based feature in each PPG signal segment; and a classification module configured to classify each PPG signal segment using the extracted IBI feature and the extracted morphology-based feature associated with the segment and using a machine learning model, the classification module further configured therefor is to generate an event prediction result based on the extracted IBI feature and the extracted morphology-based feature.

THE IMPACT OF THE REVELATION

A PPG-based arrhythmia detection device disclosed herein may be configured to extract accurate arrhythmia information using short PPG signal segments. Thus, according to one embodiment of the present invention, passive and asymptomatic arrhythmia detection is enabled.

list of figures

• 1 veranschaulicht eine elektronische Vorrichtung gemäß Ausführungsformen der vorliegenden Offenbarung.
• 2 veranschaulicht ein Blockschaubild einer am Körper eines Nutzers tragbaren Vorrichtung gemäß Ausführungsformen der vorliegenden Offenbarung.
• 3 ist ein Flussdiagramm, das veranschaulicht ein Verfahren zur Arrhythmiedetektion in einer von einem Nutzer am Körper tragbaren Vorrichtung in Ausführungsformen der vorliegenden Offenbarung.
• 4 veranschaulicht beispielhafte Signalwellenformen eines EKG-Signals und eines PPG-Signals.
• 5 veranschaulicht eine beispielhafte Signalwellenform einer Sequenz von PPG-Signalabtastungen.
• 6 veranschaulicht eine beispielhafte Signalwellenform einer Sequenz von PPG-Signalabtastungen mit Vorhofflimmern.
• 7 ist ein Diagramm, welches das Histogramm des quadratischen Mittelwertes aufeinanderfolgender Differenzen des IBI von PPG-Signalen mit normalem Sinusrhythmus und PPG-Signalen mit Vorhofflimmern veranschaulicht.
• 8 ist ein Diagramm, welches das Histogramm der Multiscale Sample Entropy-Analysen des IBI von PPG-Signalen mit normalem Sinusrhythmus und PPG-Signalen mit Vorhofflimmern veranschaulicht.
• 9 ist ein Diagramm, das die Histogramme der Standardabweichung des Bereichs unter der Kurve für normales Sinus und Vorhofflimmern veranschaulicht.
• 10 veranschaulicht eine beispielhafte Signalwellenform einer Sequenz von PPG-Signalabtastungen, die lange Ausklingmerkmale aufweisen.
• 11 veranschaulicht eine am Körper eines Nutzers tragbare Vorrichtung in Kommunikation mit einer Mobilgerät in Ausführungsformen der vorliegenden Offenbarung.
• 12 veranschaulicht eine am Körper eines Nutzers tragbare Vorrichtung in Kommunikation mit einem Cloud-Server in Ausführungsformen der vorliegenden Offenbarung.

Various embodiments of the present disclosure are disclosed in the following detailed description and the accompanying drawings.

• 1 illustrates an electronic device according to embodiments of the present disclosure.
• 2 FIG. 12 illustrates a block diagram of a user wearable device in accordance with embodiments of the present disclosure. FIG.
• 3 FIG. 10 is a flow chart illustrating a method of arrhythmia detection in a user wearable device in embodiments of the present disclosure. FIG.
• 4 illustrates exemplary signal waveforms of an ECG signal and a PPG signal.
• 5 illustrates an exemplary signal waveform of a sequence of PPG signal samples.
• 6 FIG. 12 illustrates an exemplary signal waveform of a sequence of PPG signal samples with atrial fibrillation.
• 7 Fig. 4 is a graph illustrating the squared-average histogram of successive differences in IBI of PPG normal sinus rhythm signals and PPG atrial fibrillation signals.
• 8th Figure 9 is a graph illustrating the histogram of multiscale sample entropy analyzes of the IBI of PPG signals with normal sinus rhythm and PPG signals with atrial fibrillation.
• 9 Figure 12 is a graph illustrating the histograms of the standard deviation of the area under the sinus and atrial fibrillation curve.
• 10 Figure 12 illustrates an exemplary signal waveform of a sequence of PPG signal samples having long erasure characteristics.
• 11 FIG. 12 illustrates a user wearable device in communication with a mobile device in embodiments of the present disclosure. FIG.
• 12 FIG. 12 illustrates a user wearable device in communication with a cloud server in embodiments of the present disclosure. FIG.

DETAILED DESCRIPTION

The present disclosure may be implemented in a variety of ways, including as a process; a device, a system; a composition of matter; a computer program product embodied on a computer readable storage medium; and / or a processor, such as a hardware processor or processor device, configured to execute instructions stored in memory and / or provided by a memory coupled to the processor. In this specification, these implementations or any other forms that the present disclosure may take may be referred to as techniques. In general, the order of the steps of the disclosed processes may be changed within the scope of the present disclosure. Unless otherwise stated, a component, such as a processor or memory described as being configured to perform a task, may be implemented as a generic component that is temporarily configured to perform the task at a given time, or as a special component made for carrying out the task. As used herein, the term "processor" refers to one or more devices, circuitry, and / or processing cores configured to process data, such as computer program instructions.

The following is a detailed description of one or more embodiments of the present disclosure, along with the accompanying figures which illustrate the principles of the present disclosure. The present disclosure will be described in conjunction with such embodiments, but the present disclosure is not limited to any specific embodiment. The present disclosure includes numerous alternatives, modifications, and equivalents. In the following description, numerous specific details are set forth in order to provide a more thorough understanding of the present disclosure. These details are set forth for exemplary purposes, and the present disclosure may be practiced without some or all of these specific details according to the claims. For the sake of clarity, technical material known in the technical fields involved in the present disclosure has not been described in detail so as not to unnecessarily obscure the essential aspects of the present disclosure.

In embodiments of the present disclosure, a system and method for arrhythmia detection implemented in a user-portable device allows high-precision detection of cardiac arrhythmias using photoplethysmogram (PPG) signals by analyzing inter-beat intervals (FIG. IBI) features (intervals between beats) and / or morphological beat-by-beat characteristics (beat by beat) of the PPG signals. In some embodiments, the system and method for arrhythmia detection process segments of the PPG signals to analyze irregular-irregular inter-beat interval and / or morphological beat-by-beat characteristics of a user's PPG waveform to detect arrhythmias. In one embodiment, the arrhythmia detection system and method monitors inter-beat interval (IBI) information and / or beat-by-beat morphology information in short PPG signal segments. The PPG-based arrhythmia detection system and method is designed to extract accurate arrhythmia information from the short segments of PPG, including PPG signal information, which is typically rejected by traditional PPG heart rate monitoring techniques. In some embodiments, the system and method for PPG-based arrhythmia detection provide improved detection accuracy, sensitivity, and specificity as compared to conventional heart rate-based detection techniques. In particular, the system and method for PPG-based arrhythmia detection enable passive and asymptomatic arrhythmia detection. That is, arrhythmia detection may be performed before a user displays or notices symptoms indicative of arrhythmia.

More specifically, conventional arrhythmia detection techniques rely solely upon the statistical distribution of intervals between interbeat intervals in the heartbeat information and do not rely on morphological information that may be present in the heartbeat signal. In one aspect of the present disclosure, the arrhythmia detection system and method of the present disclosure includes the analysis of morphological features in the PPG signals, and therefore, benefits from the morphological information available in the PPG signal, which also exists in the physiological conditions.

In the present specification, "photoplethysmogram" (PPG) refers to an optically obtained plethysmogram, which is a volumetric measurement of an organ. A PPG is often obtained using a pulse oximeter that illuminates the skin and measures changes in light absorption. A conventional pulse oximeter monitors the perfusion of blood to the dermis and subcutaneous tissue of the skin as a result of the cardiac cycle the heart pumps blood to the periphery, resulting in a pressure pulse on the skin. The volume change caused by the pressure pulse is detected by illuminating the skin with light from a light emitting diode (LED) and then measuring the amount of light that is either transmitted or reflected to a photodiode. The advantage of PPG-based detection is that PPG signals can be readily recorded and monitored by consumer-level wireless devices without the active participation of participants. This advantage, along with affordable wearable devices and smartphones, may allow for passive monitoring and detection of cardiac arrhythmias.

In embodiments of the present disclosure, the PPG-based arrhythmia detection system and method of the present disclosure detects cardiac arrhythmias using short discrete segments of the PPG signals and performing one or more signal analyzes on the short PPG signal segments. In one embodiment, the system analyzes the statistical regularity of the distribution of the inter-beat intervals of the PPG beats contained in each PPG signal segment. In another embodiment, the system analyzes morphology-based features of the PPG beats contained in each PPG signal segment. In some examples, the system may analyze the statistical distribution of morphology properties or the similarity between morphology-based features of adjacent PPG beats extracted from each PPG signal segment. In another embodiment, the system executes both sets of analysis; that is, both the inter-beat interval characteristics and the morphology features are analyzed to detect arrhythmia.

Accordingly, in some embodiments, the system and method for PPG-based arrhythmia detection enables highly accurate detection of cardiac arrhythmias by analyzing only inter-beat interval (IBI) characteristics of the PPG signal segments or by analyzing morphological beat-by-beat-based features of the PPG signal segment. In some cases, information regarding the morphological beat-by-beat characteristics of the PPG signal segments is added to the analysis based on the IBI features to increase the detection accuracy.

In the present specification, the term "feature" in the sense of "IBI features" or "morphology-based features" refers to waveforms, waveform properties, waveform quality, and statistical properties or attributes, measured features or attributes, or derived features or attributes. For example, IBI features may include the statistical distribution of intervals between beats. In another example, morphology features may include waveform similarities between adjacent PPG beats. Morphology traits may also include the statistical distribution of certain morphology characteristics.

In alternative embodiments, the system and method for PPG-based arrhythmia detection of the present disclosure utilize electrocardiogram (ECG) signals to adjust the PPG signal to further enhance detection accuracy. In one embodiment, the ECG signal is used to adaptively adjust the time interval of the detected beats in the IBI feature analysis. In another embodiment, the ECG signal is used to adjust the decision threshold during classification to increase the detection accuracy.

The system and method for PPG-based arrhythmia detection of the present disclosure realize many advantages over conventional arrhythmia detection methods that use a PPG. For example, the PPG-based arrhythmia detection system of the present disclosure is implemented such that it is capable of analyzing many short PPG signal segments and does not require a long continuous PPG segment. Information or features from multiple PPG signal segments, for example 2 to 15 seconds, can be summarized for statistical evaluation and classification. The short PPG signal segments are much more available than the 30 second signal segment needed in some conventional systems. Second, the PPG-based arrhythmia detection system of the present disclosure allows for more accurate detection over conventional systems that use only average heart rate information. The PPG-based arrhythmia detection system of the present disclosure uses inter-beat interval and beat-by-beat morphology features that can better represent the characteristics of arrhythmias and result in a significant improvement in detection performance.

In an exemplary embodiment of the present disclosure, the system and method for PPG-based arrhythmia detection is implemented in a wrist-worn device that includes a PPG sensor. The user or patient is constantly wearing the bracelet device and the system issues notifications of detected arrhythmias. In some Embodiments, the wrist-worn device includes an accelerometer that is used to continually determine whether the wearer is in the same location. In one embodiment, when the carrier is in the same location, the PPG optical sensor is activated to measure the PPG signal. The PPG signal is monitored for a set of primary indicators to determine if the wearer has an arrhythmia. In alternative embodiments, the PPG optical sensor continuously measures and the accelerometer transmits a motion alert signal to the PPG-based arrhythmia detection system. The PPG-based arrhythmia detection system processes the PPG signals and can discard portions of the PPG signals associated with a high degree of motion, which can affect the accuracy of the PPG signal.

1 illustrates an electronic device according to embodiments of the present disclosure. As in 1 to see has an electronic device 100 , which may be a wearable device of a user, a display 160 , a processor 130 , a sensor module 150 , a battery (not shown), a band 140 and a clasp 142 on. The ribbon 140 can be placed around the wrist and the device worn on a user's body 100 can Mitt the clasp 142 to be held on the wrist. The sensor module 150 includes one or more sensors 152 . 156 and a local processor 154 (Not shown). The local processor 154 implements a control function for the sensor module and may also perform processing or preprocessing of the sensed signals. The processor 130 implements control functions for the device wearable on a user's body and may perform other signal processing functions on the sensed signals. The local processor 154 or the processor 130 can also be referred to as a diagnostic processor.

Although the device worn on a user's body 100 can be worn on a wrist, various embodiments of the disclosure need not be limited thereto. The device worn on the body of a user 100 may also be designed to be worn on other parts of the body, such as an arm (around the forearm, elbow, or upper arm), on one leg, on the chest, on the head like a headband, around the neck like a collar, and on the ear. The device worn on the body of a user 100 may be able to communicate with other electronic devices, such as a smartphone, a laptop, or various medical devices in a hospital or doctor's office.

The ad 160 may issue monitored physiological signals from a user's body for the user and / or others to view. The monitored physiological signals are sometimes referred to as biosignals or biometric data. The monitored biosignals may be, for example, heart (pulse) rate, pulse morphology (shape), pulse intervals (intervals between beats), respiratory rate and blood pressure. The ad 160 may also include instructions to the user or others during use of the device worn on a user's body 100 or the use of other measuring devices and, for example, output status and diagnostic results.

The processor 130 receives the monitored or sensed signals from sensors in the sensor module 150 , For example, the sensors detect 152 . 156 Signals from the user's wrist when the device worn on a user's body 100 is worn by a user. In embodiments of the present disclosure, the sensor module includes 150 a sensor 152 which is a biophysiological sensor. In one embodiment, the biophysiological sensor is a photoplethysmogram (PPG) sensor. In other embodiments, the sensor module comprises 150 further a second sensor 156 which is an inertial measurement sensor. In one embodiment, the inertial measurement sensor is an accelerometer. The sensor module 150 can the processor 154 include to the sensors 152 . 156 and also to process the signals sensed by the sensors. For example, the processor 154 through the sensors 152 . 156 disassemble monitored signals and then reassemble the disassembled signals. In various embodiments of the disclosure, the processor 130 also serve the functions of the processor 154 perform. Various embodiments of the disclosure may also include different numbers of sensors.

In some embodiments, the sensor is 152 a PPG sensor used to continually or periodically monitor cardiac-related physiological information, such as cardiac output or cardiac output, of a user. Furthermore, the sensor 156 an accelerometer that is used to continually or periodically monitor user motion information. The sensor module 150 may include other sensors, such as a thermometer for measuring the temperature of the user.

The device worn on the body of a user 100 implements the PPG-based arrhythmia detection system of the present invention Revelation in the processor 130 , In some embodiments, the PPG-based arrhythmia detection system includes a signal processing module for segmenting the PPG signals into short PPG signal segments, and also includes a machine learning network for assessing the PPG signal segments and estimating the likelihood of having cardiac arrhythmias in the monitored signal.

2 FIG. 12 illustrates a block diagram of a user wearable device in accordance with embodiments of the present disclosure. FIG. As in 2 includes a device that is portable on a user's body 100 a sensor module 150 , a processor 130 , an ad 160 and a battery 170 to power other components with power. The processor 130 controls the on the display 160 displayed output. The ad 160 may also include input devices (not shown) such as buttons, rotary switches, a touch screen, and a microphone.

In embodiments of the present disclosure, the sensor module includes 150 a biophysiological sensor 152 for measuring a biological signal of the user. In the present embodiment, the biophysiological sensor is 152 a PPG sensor. The sensor module 150 may further include an inertial measurement sensor 156 for measuring a motion signal of the user. In the present embodiment, the inertial measurement sensor is 156 an accelerometer, such as a three-axis accelerometer. The sensor module 150 can with the local processor 154 Be provided with the sensors 152 . 156 to steer, and also to those through the sensors 152 . 156 sensed biosignals or motion signals to process. In some embodiments, the signal processing operation may be in the local processor 154 and / or in the processor 130 be implemented. Alternatively, the local processor 154 perform some of the signal processing, such as certain signal preprocessing, and the processor 130 implements other signal processing algorithms for biometric determination or other functions. In embodiments of the present disclosure, the particular processor used to carry out the biometric signal processing algorithms is not critical to the practice of the present disclosure.

In embodiments of the present disclosure, the processor is 130 configured to control the fetch operation, the scan regime, the signal processing operation and device communication events, and other device-specific functions in the user wearable device. In the present embodiment, the processor includes 130 a CPU 132 , a store 134 , an input / output (I / O) interface 182 , a communication interface 184 and a detection module 190 , Although the processor 130 is described as including these various elements, other embodiments may use other architectures in which the various functions are grouped differently. For example, the grouping may reside in various integrated circuit chips, or the grouping may include various elements such as the I / O interface 182 and the communication interface 184 together.

The processor 130 includes the detection module 190 for performing arrhythmia detection on the sensed biosignal, such as the PPG signal. In embodiments of the present disclosure, the detection module comprises 190 a data processing module 192 , an IBI detection module 194 , a morphology detection module 196 and a classification module 198 , The signal processing module 192 is configured to perform signal preprocessing on the sensed biosignal. For example, the data processing module 192 perform baseline removal or DC signal level removal on the sensed PPG signals. In other embodiments, the data processing module 192 perform signal segmentation to divide the sensed PPG signals into short PPG signal segments. For example, each PPG signal segment can be between 2 and 15 seconds. Alternatively, each PPG signal segment may comprise a number n of detected beats. In one example, n is between 40 and 70. That is, each PPG signal segment may comprise 40 to 70 strokes.

The IBI detection module 194 implements an analysis of inter-beat intervals in the detected beats of the PPG signal segment. The morphology detection module 196 implements an analysis of morphology-based characteristics of the detected beats of the PPG signal. In some embodiments, the detection module 190 include the IBI detection module or the morphology detection module or both modules.

The classification module 198 implements the classification of the PPG signal segments to detect the presence of arrhythmia in the PPG signal segments. The classification module 198 uses the analysis results from the IBI detection module 194 and / or the analysis results from the morphology detection module 196 , The classification module 198 predicts the likelihood of arrhythmia in the PPG signal segments.

In alternative embodiments, the arrhythmia detection system of the present disclosure may be implemented in an electronic device that communicates with the user wearable device that includes the sensor module. For example, the electronic device may be a mobile device, such as a smartphone or a tablet device. The sensed PPG signals and other accompanying signals, such as motion signals, may be sent to the electronic signal processing and arrhythmia detection device according to the arrhythmia detection method of the present disclosure, as shown in FIG 11 shown to be transmitted. The electronic device may output the detection results to display on the wearable device of a user.

In another embodiment, the arrhythmia detection system of the present disclosure may be implemented in a cloud server located on a data network and communicating with the user-portable device containing the sensor module. The sensed PPG signals and other accompanying signals, such as motion signals, may be transmitted through the data network to the cloud server for signal processing and arrhythmia detection according to the arrhythmia detection method of the present disclosure, as shown in FIG 12 shown to be transmitted. The cloud server may output the detection results to display on the user wearable device. In the present specification, a cloud server refers to a logical server that is built, hosted, and provisioned by a cloud computing platform over a data network, such as the Internet. For example, cloud servers have and have similar capabilities and capabilities as a typical server, but are remotely accessed by a cloud service provider.

3 FIG. 10 is a flowchart illustrating a method of arrhythmia detection in a user wearable device in embodiments of the present disclosure. FIG. In some embodiments, the method 200 in a processor in a wearable device, such as the processor 130 the device worn on a user's body 100 in the 1 and 2 to be implemented. Alternatively, the method 200 in a mobile device that communicates with the wearable device. In a further embodiment, the method 200 be implemented in a cloud server that communicates with the wearable device. As in 3 to see, receives the procedure 200 a channel of biosignal data signals from a first sensor implemented in the device worn on a user's body ( 202 ). For example, the biosignal data signals may be PPG signals. In the present embodiment, the method receives 200 Raw data samples, that is, data samples that have not been processed or that have been minimally processed.

at 204 performs the procedure 200 processing of the PPG signals to obtain PPG signal samples. In some examples, the processing of the PPG signals may include removing the DC baseline signal level. In other examples, the processing may include other signal processing to improve the signal level. As a result of processing, PPG signal samples are generated. In one embodiment, the processing step is by the data processing module 192 in the detection module 190 of the processor 130 the device worn on a user's body 100 executed.

4 illustrates exemplary signal waveforms of an ECG signal and a PPG signal. More specifically, a PPG signal for measuring cardiac rhythm or heartbeat has a particular waveform profile and is different from the waveform profile of an ECG signal. As in 4 To see, the ECG measures the electrical activity of the heart, and the ECG signal (curve 250 ) includes a salient feature known as the QRS complex, which indicates the major pumping contractions of the heart. The peak R in the ECG signal is used by heart rate algorithms to measure the amount of time that passes between each pulsating heartbeat. The length of time between each peak R is referred to as the RR interval.

Furthermore, the PPG measures the pressurized pulse of blood into the arteries of the body, causing the arteries to swell slightly before returning to their former state. The PPG signal is an optical signal, with the amplitude of the optical signal being directly proportional to the pulse pressure. The PPG signal (curve 252 ) includes quasi-periodic pulses, with peaks and valleys, which can be used to indicate the periodicity of the signal waveform, thereby enabling estimation of the heart rate. Specifically, the duration between the peaks of two adjacent pulses, or the valleys of two adjacent pulses, is referred to as the Inter-Beat Interval (IBI), which may be used as an indication of the heart rate. In some cases, the PPG signal shows a dicrotic dip. A dicrotic intrusion is a small, downward deflection that can be observed on the downward deflection of an arterial pressure waveform. He represents the Intersection of superimposed primary and reflected pressure waves in the arterial tree.

We return 3 back. at 206 detects the process 200 Beats in the PPG signal samples. In one embodiment, the method detects 200 the peaks or valleys in the signal waveform of the PPG signal, and uses the detected peaks or valleys to indicate the position of a beat in the PPG signal samples. 5 illustrates an exemplary signal waveform of a sequence of PPG signal samples. As in 5 seeing the process detected 200 in the present embodiment, the valleys in the signal waveform to determine the position of a beat in the PPG signal samples. Accordingly, the method identifies 200 the boundary of each pulse in the PPG signal samples as a beat.

We return 3 back. at 208 subdivides the procedure 200 the PPG signal samples in PPG signal segments. In one embodiment, the method divides 200 the PPG signal samples into PPG signal segments with a certain amount of time, such as a number t of seconds. For example, any PPG signal segment 2 to 15 Seconds. In a further embodiment, the method divides 200 the PPG signal samples into PPG segments of a number n of beats. For example, any PPG signal segment 40 to 70 Include beatings.

In embodiments of the present disclosure, the method detects 200 Beats over a certain period of time to form a PPG signal segment, where the beats in a PPG signal segment may or may not be continuous in time. In other embodiments of the present disclosure, the method detects 200 a certain number of strokes to form a PPG signal segment, where the strokes in a PPG signal segment may or may not be time-continuous. The procedure 200 can detect some beats in the PPG signal samples, then discard some beats, and then proceed to detect some other beats to form the PPG signal segment.

at 210 extracts the procedure 200 Inter-Beat Interval features in each PPG signal segment. More specifically, the procedure is judged 200 the PPG signal segments to analyze the statistical regularity of the distribution of the inter-beat intervals of the PPG beats contained in each PPG signal segment. That way, the process can 200 Extract inter-beat interval properties that are irregular-irregular. In one embodiment, the IBI feature extraction step is performed by the IBI detection module 194 in the detection module 190 of the processor 130 the device worn on a user's body 100 executed.

In normal PPG pulses, the time interval between human heartbeats due to respiration and other longer term sympathetic characteristics varies in a fairly predictable manner. However, if the person has an arrhythmia, the intervals between the beats become very irregular due to the abnormal activation patterns present in the tissue, which makes the intervals much more erratic and statistically less predictable. The erratic intervals between beats can be determined by comparing normal PPG pulses with PPG pulses with arrhythmia.

More specifically, illustrated 5 PPG signal samples collected from patients in a normal sinus rhythm. The IBI time periods vary from 1.04 seconds to 1.12 seconds. As the IBI time periods vary across the PPG pulses, the IBI time periods are consistent and vary in a predictable manner. 6 FIG. 12 illustrates an exemplary signal waveform of a sequence of PPG signal samples with atrial fibrillation. If the PPG signal samples have arrhythmia, the IBI time periods vary widely over the set of PPG pulses. In the present example, the IBI times vary from 0.56 seconds to 1.14 seconds. This irregular irregularity is an indication of arrhythmia or atrial fibrillation.

In embodiments of the present disclosure, the method analyzes 200 each PPG signal segment and extracts the "irregular irregularity" feature of the IBI in the PPG signals. In some embodiments, the irregularity in the IBI of the PPG signals is characterized using one or more statistical measures of the distribution of the Inter Beat Interval (IBI) quality. In some embodiments, the method implements 200 one or more statistical measures to evaluate the IBI time period. In one example, the statistical measures may include standard deviation, skewness, kurtosis, information entropy, root-mean-square of successive differences (RMSSD) of the IBIs, inflection point ratio, and multiscale sample entropy. The statistical measures are used to extract features of the PPG signal segment, where the features may indicate a deviation from the normal sinus rhythm.

7 Fig. 12 is a graph showing the histogram of the root mean square of successive differences of the IBI of PPG signals with normal sinus rhythm and PPG signals with atrial fibrillation illustrated. As in 7 By using the "Root Mean Square of Successive Differences" analysis, PPG signals with normal IBI and PPG signals with abnormal IBI can easily be distinguished.

8th Figure 12 is a graph illustrating the histogram of IBI multiscale sample entropy analyzes of PPG normal sinus rhythm signals and PPG atrial fibrillation signals. As in 8th By using the "Multiscale Sample Entropy" analysis, PPG signals with normal IBI and PPG signals with abnormal IBI can easily be distinguished.

We return 3 back. at 212 extracts the procedure 200 Morphology-based features in each PPG signal segment. More specifically, the procedure is judged 200 the PPG signal segments to analyze morphology-based features of the PPG beats contained in each PPG signal segment. The morphology-based features may include statistical properties, measured features, or derived features of the PPG beats in the PPG signal segment. The morphology-based features may also include waveforms, waveform characteristics, and waveform quality of the PPG beats in the PPG signal segment. In one embodiment, the method analyzes 200 the statistical distribution of morphology properties and the similarity between morphology features of adjacent PPG beats extracted from each PPG signal segment. In one embodiment, the morphology feature extraction step is performed by the morphology detection module 196 in the detection module 190 of the processor 130 the device worn on a user's body 100 executed.

In a normal sinus rhythm, adjacent PPG beats from a patient in normal sinus rhythm have a very similar waveform or morphology, as seen by the PPG pulse waveforms in FIG 5 shown. However, this is not the case with arrhythmias, as by the PPG pulse waveforms in FIG 6 shown. For example, beats that arrive irregularly early do not usually allow the blood of the previous beat to disperse completely in the skin, resulting in an increase in the DC signal, while a similar negative baseline motion is observed in PPG beats that arrive late , Similarly, since the heart muscle has abnormally contracted and the blood is ejected differently, the perfusion detected by the PPG waveform may also have a different shape and pressure reflections from the edge, resulting in the added reflected waveforms that overlay the original pressure pulse. As a result, morphology traits can be used as a good measure to distinguish normal sinus from cardiac arrhythmias.

In embodiments of the present disclosure, the arrhythmia detection method extracts 200 Morphology-based features that include a standard deviation of the areas under the curve (AUC) of PPG knock in a PPG signal segment. For example 9 a graph illustrating the histograms of the standard deviation of the area under the curve for normal sinus and atrial fibrillation. The difference in morphological properties between the histogram of the normal sinus and the histogram of the atrial fibrillation is in 9 to see clearly. The procedure 200 assesses the standard deviation of the areas under the PPG beat curve in a PPG signal segment to detect an indication of possible cardiac arrhythmia.

In other embodiments, the arrhythmia detection method extracts 200 Morphology-based features that include waveform similarity of adjacent PPG beats in a PPG signal segment. The waveform similarity may be judged by cross-correlation or similarity measures. The procedure 200 evaluates the waveform similarity measures in the PPG beats in a PPG signal segment to detect an indication of possible cardiac arrhythmia.

In other embodiments, the arrhythmia detection method extracts 200 Morphology-based features that include the ratio of PPG long beating portion beats in a PPG signal segment. For example, the method detects 200 the percentage of onset of PPG beats with a long fade out section. In the present specification, a PPG blow with a long tail portion refers to a PPG beat waveform having an extended downward slope. That is, the long decay feature refers to a pronounced downward deflection of an arterial pressure waveform. 10 Figure 12 illustrates an exemplary signal waveform of a sequence of PPG signal samples having long erasure characteristics. As in 10 to see, the identifier LT indicates the PPG pulses with long Ausklingabschnitt. The procedure 200 assesses the PPG beats within a PPG signal segment to detect the number of beats with the long escape feature. The percentage of onset of the long decay feature is a morphology-based trait that can be used to indicate possible cardiac arrhythmias. In some embodiments, this will detected long Ausklingmerkmal using a pattern recognition technique.

In other embodiments, the arrhythmia detection method extracts 200 Morphology-based features that include the ratio of PPG beats with abnormal break-ins in a PPG signal segment. For example, the method detects 200 the percentage of entry of PPG beats with abnormal break-ins. An abnormal break-in differs from a dicrotic break-in described above in that an abnormal break-in represents two incomplete abnormal heartbeats, while a dicrotic break-in represents a normal heartbeat. As in 10 to see, the identifier AN indicates the PPG pulses with abnormal break-ins. The procedure 200 assesses the PPG beats within a PPG signal segment to detect the number of beats with the abnormal break-in feature. The percentage of onset of the abnormal break-in feature is a morphology-based trait that can be used to indicate possible cardiac arrhythmias. In some embodiments, the abnormal intrusion feature is detected by a pattern recognition technique.

In other embodiments, the arrhythmia detection method extracts 200 Morphology-based features that include the standard deviation of the AC components of the rising edges and the standard deviation of the AC components of the falling edges of the PPG pulses in a PPG signal segment. A PPG waveform includes an AC component and a DC component. The AC component corresponds to variations in blood volume in synchronization with the heartbeat. The DC component arises from the optical signals reflected or transmitted by the tissue and is determined by the tissue structure as well as the venous and arterial blood volumes. The DC component shows minor changes with breathing. The fundamental frequency of the AC component varies with the heart rate and is applied across the DC baseline. In one embodiment, the method calculates 200 the AC amplitude of the rising edge of a heartbeat as the positive portion of the 1st derivative of the PPG waveform and the AC amplitude of the falling edge of the heartbeat as the negative portion of the 1st derivative of the PPG waveform. In an alternative embodiment, the PPG AC pulse waveform contour may be characterized by signal samples, peak systolic amplitude, diastolic peak amplitude, and second derived extremes of the pulse waveform. In embodiments of the present disclosure, the method assesses 200 the standard deviation of the AC components of the PPG beats in a PPG signal segment to detect an indication of possible cardiac arrhythmia.

at 214 classifies the procedure 200 the PPG signal segments using the extracted IBI features and / or the extracted morphology features associated with each PPG signal segment. The procedure 200 classifies the PPG signal segment using a machine learning model that has been previously learned based on signals and arrhythmia annotations from one or more sets of arrhythmia training data. In one embodiment, the method classifies 200 the PPG signal segments using a combination of IBI features and morphology features. Specifically, certain morphology features indicate cardiac arrhythmia. Therefore, the use of a combination of IBI and morphology-based features may be of critical importance for increasing the likelihood of arrhythmia prediction. In one embodiment, the classification step is performed by the classification module 198 in the detection module 190 of the processor 130 the device worn on a user's body 100 executed.

In some embodiments, the method uses 200 a random forest classifier to perform the classification. "Random Forest" is an ensemble method formed by combining several different independent base classifiers or decision trees. Each independent classifier is trained using the sampled record with replacement from the original training record. The features with maximum information gain are selected for further sharing. The best division features are identified from a random subset of the available features. This bagging / bootstrapping aggregation has the advantage of reducing excessive fitting so that the model can be generalized to a larger population while reducing the error rate. To allow implementation in an embedded system, in one embodiment, the random forest model uses only 3 decision trees, and the depth of each tree is 5. In this way, the method allows 200 the real-time measurement and prediction of an arrhythmia event.

As used herein, the term "machine learning model" refers to classification models that can use training to provide precise classifications. In practice, the training of a classification model on high-performance computers, and the The learned model is then used in the device where inference is performed using the model. In some embodiments, any machine learning and / or classification technique may be used to perform the classification of the PPG features described above. Briefly, embodiments of the present disclosure relate to arrhythmia detection or event prediction by machine learning, which can be incrementally refined under expert guidance. In at least one of the various embodiments, data may be fed into a machine learning model that has been learned using multiple classifiers (index, identifiers or annotations) and one or more sets of training data and / or test data.

at 216 generates the procedure 200 an event prediction result based on the extracted IBI features and / or the extracted morphology features associated with each PPG signal segment. In one embodiment, the method generates 200 an arrhythmia detection result based on the classification of the PPG segments using the IBI features and / or morphology features of the PPG segments.

In some embodiments, the arrhythmia detection method sends 200 in response to detecting that arrhythmia is present, a notification to the user. For example, the notification may be sent by an application on a mobile device and / or the wearable device.

Using the analyzes described above, the PPG-based arrhythmia detection system takes several short discrete segments of PPG signals as input, uses the inter-beat interval features and / or morphology features extracted from those segments, and provides the detection result ready. In this way, a passive detection system is realized, which the user can use all day long.

In some embodiments, the method receives 200 Movement information associated with the device wearable on a user's body ( 218 ). For example, the motion information may be obtained from a second sensor, such as an inertial measurement sensor or an accelerometer. The procedure 200 can the movement information during the PPG signal segmentation step ( 208 ) to discard PPG signal samples that are associated with a high degree of motion and therefore may be unreliable. Alternatively, the PPG sensor may be turned off during periods of heavy movement such that no PPG signal is available during these periods. Accordingly, in embodiments of the present disclosure, the PPG signal segments generated in this manner are not necessarily continuous-time, but may be separate PPG signal samples. The Arrhythmiedetektionsverfahren 200 may operate on short segments of PPG signal samples, where each segment of PPG signal samples may be time-discontinuous.

In some embodiments, the method receives 200 ECG signals ( 220 ). The procedure 200 may use the ECG signals to adjust the detected beats in the PPG signal segments. In one example, the method uses 200 the ECG signal for information on the heart rate, which can be used to adaptively adjust the decision threshold and the time interval between beats to increase the detection accuracy. At this point it should be noted that the procedure 200 the ECG signals are used only for threshold adjustment and that the determination of the presence or absence of arrhythmias is made solely using the PPG signals.

In one example, if there is sufficient arrhythmia during a given period of the PPG during fixed conditions, the user may be prompted to perform an ECG measurement, for which the sensor may also be located on the wearable device. The combination of PPG and ECG measurements may be interpreted for the presence of arrhythmia by a physician or by an ECG analysis algorithm. The determination may then be communicated to the user or stored for future cumulative analysis to identify a trend for a chronic disease.

In the above in 3 described PPG-based Arrhythmiedetektionsverfahren 200 illustrates the process 200 the use of a combination of IBI features and morphology-based arrhythmia detection features. Although the use of a combination of IBI features and morphology-based features improves detection accuracy, the PPG-based arrhythmia detection method of the present disclosure may also be implemented using only the IBI features or only the morphology-based features. IBI features or morphology-based features individually provide an indication of arrhythmias that may be used as the basis of accurate arrhythmia detection.

Aspects of this disclosure are described herein with reference to flowchart Illustrations or block diagrams are described in which each block or combination of blocks may be implemented by computer program instructions. The instructions may be communicated to a processor of a general purpose computer, a special purpose computer or other programmable data processing device to make a machine or product, and when executed by the processor, the instructions provide a means for implementing the functions, actions or events which are specified in each block or each combination of blocks in the graphs.

In this regard, each block in the flowchart or block diagrams may correspond to a module, segment, or portion of code that includes one or more executable instructions for implementing the specified logic function (s). It should also be noted that in some alternative implementations, the functions associated with a block may occur in a different order than indicated in the figures. For example, two blocks shown consecutively may in practice also be executed substantially simultaneously, or the blocks may occasionally also be executed in reverse order.

One of ordinary skill in the art will understand that aspects of this disclosure may be embodied as a device, system, method, or computer program product. Accordingly, aspects of this disclosure, which are generally referred to herein as circuits, modules, components, or systems, may be embodied in hardware, in software (including firmware, resident software, microcode, etc.), or in any combination of software and hardware including computer program products embodied in a non-transitory computer readable medium having computer readable program code embodied thereon.

The descriptions set forth above are to illustrate specific embodiments of the present disclosure and should not be construed in a limiting sense. Numerous modifications and variations are possible within the scope of the present disclosure.

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited patent literature

• US 62648821 [0001]
• US 15145 [0002]
• US 356 [0002]

Claims (23)

A method of event detection in a user wearable device (100), comprising: Receiving (202), from a first sensor (152) implemented in the user-portable body of a user device (100), photoplethysmogram (PPG) signals; Processing (204), in a processor (130), the PPG signals to obtain PPG signal samples; Detecting (206), in the processor (130), beats in the PPG signal samples; Dividing (208) the PPG signal samples into PPG signal segments; Extracting (210) at least one inter-beat interval (IBI) feature in each of the PPG signal segments; Classifying (214), in the processor (130), each of the PPG signal segments using the extracted IBI feature associated with each of the PPG signal segments and using a machine learning model; in response to the classifying, generating (216), in the processor (130), an event prediction result for the PPG signal segment of the PPG signal segments based on the extracted IBI feature; and Displaying the event prediction result on the user wearable device (100). Method according to Claim 1 , further comprising: extracting (212) at least one morphology-based feature in each of the PPG signal segments; Classifying (214), in the processor (130), each of the PPG signal segments using the extracted IBI feature and the extracted morphology-based feature associated with each of the PPG signal segments and using the machine learning model; and in response to the classifying (214), generating (216), in the processor (130), the event prediction result based on the extracted IBI feature and the extracted morphology based feature. Method according to Claim 1 wherein extracting (210) the at least one inter-beat-interval (IBI) feature in each of the PPG signal segments comprises: extracting the IBI feature by analyzing a duration of inter-beat intervals of the detected strokes into the PPG Signal segments using one or more of standard deviation, skewness, kurtosis, information entropy, inflection point ratio, root mean square sequential differences, and multiscale sample entropy. Method according to Claim 2 wherein extracting (212) the at least one morphology-based feature in each of the PPG signal segments comprises: extracting the morphology-based feature by analyzing one or both of a statistical distribution of morphology properties and a similarity between morphology features adjacent beats in each of the PPG signal segments. Method according to Claim 4 wherein extracting (212) the morphology-based feature by analyzing one or both of the following comprises: extracting the morphology-based feature by analyzing one or more of: a standard deviation of regions below the curves of the detected strokes in each of the PPG signal segments; Waveform similarities of adjacent beats in the PPG signal segments; a ratio of PPG long tail sounds in the PPG signal segments; a ratio of PPG beats with abnormal break-in in the PPG signal segments; and a standard deviation of an AC component with rising edges and falling edges of pulses in the PPG signal segments. Method according to Claim 1 wherein dividing (208) the PPG signal samples into the PPG signal segments comprises dividing the PPG signal samples into the PPG signal segments with a predetermined time of a number t of seconds. Method according to Claim 1 wherein dividing (208) the PPG signal samples into the PPG signal segments comprises dividing the PPG signal samples into the PPG signal segments by a number of beats, each of the PPG signal segments having a number n of beats. Method according to Claim 1 wherein the PPG signal samples in one or more PPG signal segments comprise PPG signal samples acquired over a non-continuous period of time. Method according to Claim 8 further comprising: receiving (218), from a second sensor (156) implemented in the device (100) wearable on a user's body, a motion signal representative of movement activity on the user-wearable device ( 100) indicates; in response to the motion signal, removing PPG signal samples associated with a high degree of motion activity from the PPG signal samples; and dividing (208) the PPG signal samples into the PPG signal segments by dividing the remaining PPG signal samples, the remaining PPG data samples being detected over the non-continuous duration of the PPG signal samples. Method according to Claim 9 wherein the first sensor (152) comprises a photoplethysmogram (PPG) sensor and the second sensor (156) comprises an accelerometer. Method according to Claim 1 , further comprising: receiving, in the processor (130), electrocardiogram (ECG) signals; Adjusting the PPG signal samples in each of the PPG signal segments using the ECG signals; and adjusting a decision threshold during classifying each of the PPG signal segments to increase a detection accuracy. Method according to Claim 1 wherein generating the event prediction result for the PPG signal segment of the PPG signal segments based on the extracted IBI feature comprises generating the event prediction result indicative of a likelihood that cardiac arrhythmias are present in a particular PPG signal segment of the PPG signal segments. Method according to Claim 1 , further comprising: in response to the event prediction result indicating a high probability that an event is present in a particular PPG signal segment of the PPG signal segments, outputting a notification to the user wearable device (100). Method according to Claim 1 wherein the processing, in the processor (130), of the PPG signals for obtaining the PPG signal samples comprises processing, in the processor (130) implemented in the user-portable device body (100), the PPG signal Includes signals for obtaining the PPG signal samples. Method according to Claim 1 wherein the processing, in the processor (130), of the PPG signals for obtaining the PPG signal samples, the processing in the processor (130) implemented in a mobile device compatible with the user wearable device (100 ) which includes PPG signals for obtaining the PPG signal samples. Method according to Claim 1 wherein the processing, in the processor (130), of the PPG signals for obtaining the PPG signal samples, the processing, in the processor (130) implemented in a cloud server, with the device wearable on a user's body (100) which includes PPG signals for obtaining the PPG signal samples. A method of event detection in a user wearable device (100), comprising: Receiving (202), from a first sensor (152) implemented in the user-portable body of a user device (100), photoplethysmogram (PPG) signals; Processing (204), in a processor (130), the PPG signals to obtain PPG signal samples; Detecting (206), in the processor (130), beats in the PPG signal samples; Dividing (208) the PPG signal samples into PPG signal segments; Extracting (212) at least one morphology-based feature in each of the PPG signal segments, the morphology-based feature being associated with statistical properties of the PPG signal samples or waveform characteristics of the PPG signal samples; Classifying (214), in the processor (130), each of the PPG signal segments using the extracted morphology-based feature associated with each of the PPG signal segments and using a machine learning model; in response to classifying (214), generating (216), in the processor (130), an event prediction result based on the extracted morphology-based feature, and Displaying the event prediction result on the user wearable device (100). Method according to Claim 17 wherein extracting (212) the at least one morphology-based feature in each of the PPG signal segments comprises: extracting the morphology-based feature by analyzing a statistical distribution of morphology characteristics and / or similarity between adjacent peak morphology features each of the PPG signal segments. Method according to Claim 18 wherein extracting the morphology based feature by analyzing the one or both thereof comprises: extracting the morphology based feature by analyzing one or more of: a standard deviation of regions below the curves of the detected beats in each of the PPG signal segments; Waveform similarities of adjacent beats in the PPG signal segments; a ratio of PPG long tail sounds in the PPG signal segments; a ratio of PPG beats with abnormal break-in in the PPG signal segments; and a standard deviation of the AC component from rising edges of the PPG pulses and a standard deviation of the AC component from falling edges of pulses in the PPG signal segments. Apparatus comprising: a sensor module (150) comprising a first sensor (152) configured to measure photoplethysmogram (PPG) signals; and a processor (130) comprising: a data processing module (192) configured to process the PPG signals to obtain PPG signal samples, detect beats in the PPG signal samples, and subdivide the PPG signal samples into PPG signal segments; an inter-beat interval detection module (194) configured to extract at least one inter-beat interval (IBI) feature in each of the PPG signal segments; a morphology detection module (196) configured to extract at least one morphology-based feature in each of the PPG signal segments; and a classification module (198) configured to classify each of the PPG signal segments using the extracted IBI feature and the extracted morphology-based feature associated with the segment and using a machine learning model, wherein the classification module of the Further, configured to generate an event prediction result based on the extracted IBI feature and the extracted morphology-based feature. Device after Claim 20 wherein the inter-beat interval detection module (194) is configured to extract the IBI feature by analyzing a duration of inter-beat intervals of the detected strokes in the PPG signal segments using one or more of: standard deviation , Skewness, kurtose, information entropy, inflection point ratio, quadratic averages of successive differences and multiscale sample entropy. Device after Claim 20 wherein the morphology detection module (196) is configured to extract the morphology-based feature by analyzing a statistical distribution of morphology characteristics and / or a similarity between morphology features of adjacent beats in each of the PPG signal segments. Device after Claim 22 wherein the morphology detection module (196) is configured to extract the morphology-based feature by analyzing one or more of: a standard deviation of regions below the curve of detected beats in each of the PPG signal segments; Waveform similarities of the adjacent beats in the PPG signal segment; a ratio of PPG beats with a long fade-out portion in the PPG signal segment; a ratio of PPG beats with abnormal break-in in the PPG signal segments; and a standard deviation of the AC component of rising edges and falling edges of beats in the PPG signal segments.

Images