CN103300819B - Study patient monitoring and interfering system - Google Patents

Abstract

A kind of patient monitoring and interfering system, including: interface, for receiving the data representing multiple different parameters from multiple different sensors, the sensor that the plurality of different sensor is included in sick bed and the sensor being attached to patient, this sensor includes: heart rate sensor, respiration pickup, and the pressure transducer of instruction bed pressure spot.Study processor, by recording patient parameter value within a period of time and analyzing recorded parameter value to determine their scope, determines the normal range of multiple different patient parameter group received for this patient.Data processor determine this different patient parameter group received whether exceed determined by normal range determining in response to this and in response to the type of parameter in this set and the medical record information of this patient, initiate the adjustment to sick bed and at least one of the following: (a) changes and give the medicine of patient and (b) changes to worker's alarm patient parameter.

Description

Study patient monitoring and interfering system

This is to be faced the non-of Provisional Application serial number 61/611,260 that on March 15th, 2012 submits to by C. Schultz et al. Time application.

Invention field

The present invention relates to use patient monitoring and the interfering system neutralizing the sensor being attached to patient at sick bed, it uses Study processor determines the normal range of one group of patient parameter received and in response to the ginseng exceeding the normal range determined Number adjusts sick bed, changes and gives the medicine of patient and send alarm to worker.

Background technology

Preferably make monitoring and intervening can support in hospital, family, other controlled and organized environment and Patient medical health care in uncontrolled environment (the such as scene of the accident, disaster area and other outdoor environment).Such Arrange down, substantial amounts of casualty may be had to need quickly and identify and severity classification.Such situation is for monitoring patient All it is challenging to transporting them.According to the system of inventive principle process these need and the problem that is associated support from Substantial amounts of patient in real time to the seamless concurrent collection of different types of medical data and support for these patients from Dynamic nursing intervention.These not enough and relevant problems are processed according to the system of inventive principle.

Summary of the invention

System provides a kind of transparent and hospital patient watch-dog of intelligence, and it uses system unit and wire or wirelessly The multiple sensors being connected to this unit perform to relate to generating the process of alarm, data base and display function and performing patient's number According to monitoring, storage and knowledge base function.Patient monitoring and interfering system include: represent for receiving from multiple different sensors The data of multiple different parameters, these multiple different sensors are included in sick bed and neutralize and be attached to the sensor of patient and (include Heart rate sensor, respiration pickup and the pressure transducer of instruction bed pressure spot).Study processor is by remembering within a period of time Record patient parameter value also analyzes recorded parameter value to determine their scope, determines a different set of receiving for patient Normal range of patient parameter.Data processor determines that the patient parameter received that this group is different is normal determined by exceeding Scope also determines in response to this and in response to the type of parameter in this set and the medical record information of patient, initiates the tune of sick bed Whole and following at least one: (a) changes the medicine giving patient, and (b) changes to worker's alarm patient parameter.

Accompanying drawing explanation

Fig. 1 is shown according to the doctor of the sensor being attached in sick bed or being wirelessly attached to patient of embodiments of the invention Institute configures.

Fig. 2 is shown according to patient monitoring and the interfering system of embodiments of the invention.

Fig. 3 is shown according to the normal mode including automatically learning to depend on the sensing data of patient of embodiments of the invention Formula for patient automatically and have the monitoring of guide and the process chart of intervention.

Fig. 4 is shown according to the method processing the sensing data obtained of embodiments of the invention, and the method includes adaptive Induction sensor selects, classification feature selects and responds or action selects.

Fig. 5 be shown according to embodiments of the invention for automatically monitoring patient and identify at the home care of unusual condition The example flow diagram of reason.

Fig. 6 illustrates by the flow chart of the process performed according to patient monitoring and the interfering system of embodiments of the invention.

Detailed description of the invention

Fig. 1 illustrates hospital's configuration of the sensor being combined in sick bed 103 or being wirelessly attached to patient 105, this hospital Configuration realizes transparent and intelligence hospital patient monitoring.System in one embodiment combines sensor in sick bed 103 And infrastructure, in infrastructure, it is provided with system unit 107(by each sick bed and is also referred to as sensor bridge).Physics bed accessory There is the unit of similar system and be wired or wirelessly connected to multiple sensors of this unit.Bed is wirelessly connected to basis and sets Execute 109(such as, Ethernet) and perform process for each on the computer be connected to ethernet network.System unit 107 Obtain wired simulation input and numeral input from sensor, measurement is carried out digital sample and routes them to teleprocessing Stand (such as, server).Remote server performs generate the process of alarm, data base and display function and perform patient data's prison Control, storage and knowledge base function.Server achieve from nurse station, from physicians location and patients home by bed to patient's number According to comprehensive security webserver connectivity.

Fig. 2 illustrates patient monitoring and the interfering system 10 of the network including monitor node 43,45,47,49 and 51, this monitoring Node individually comprises the connectivity that cooperation realizes in the most scattered environment (such as including the scene of the accident) together Process and communication unit 107(Fig. 1).Doctor and other medical worker pass through ad hoc network (that is, for specifically by system 10 The network set up with potential interim function) communicate with back-end server 30.Server 30 includes at data processor 15, study Reason device 25, storage vault 17, interface 27 and display 19.Interface 27 receives from multiple different sensors and represents multiple different parameters Data, these multiple different sensors are included in sick bed and neutralize and be attached to the sensor of patient and (include heart rate sensor, exhale Inhale sensor and the pressure transducer of instruction bed pressure spot).In one embodiment, server 30 is the equipment of networking, and it refers to Show the possible connection to other healthcare resources (including patient record and drug information), and special to other health care The connection of family.Study processor 25 is by recording patient parameter value and analyzing recorded parameter value to determine within a period of time Their scope, thus the normal range of one group of multiple different patient parameter received is determined for patient.Unit 25 also from The data set learning standard of similar patient, and healthcare workers determines canonical parameter in one embodiment.Data Processor 15 determine the patient parameter received that this group is different exceed determined by normal range and determine in response to this and In response to type and the medical record information of patient of parameter in this set, processor 15 initiate sick bed is adjusted and following extremely Few one: (a) changes the medicine giving patient, and (b) changes to worker's alarm patient parameter.

Patient sensor parameter value data that storage vault 17 was stored in a period of time, determined by remember within a period of time Patient's normal range and the instruction of the patient parameter group of record exceed, in response to patient parameter value, the normal range determined and perform Action and the data of response.Display 19 provides physicians with message and alarm.Even if in the case of there is no broadband connection (such as In disaster area), the autonomous character of this ad hoc network provides local connectivity.Unit 107 obtains data, to it in this locality Carry out pretreatment and on network 21, send necessary information.Therefore, the load on network and the storage on server 30 need Ask and be minimized.The definition of sensor information necessary for each is applied is different and is adaptive or use Family is selectable.Bigger data volume (such as original sensor data) can be locally stored in unit 107 and can be through Connected (such as USB connects) by such as high bandwidth and be accessed.

The scope of the sensor integrated with unit 107 includes: be programmed into intelligent bed or intelligent room (mattress, medicated pillow, gear Plate, wall) in wired sensor.This sensor includes: microphone (for breathing, heart beating, the sound of stomach), vibration (for Pulse, tremble, onste, muscle spasm and ballism), pressure is (for the local tightened, body position, tissue can be compressed High pressure points, determines whether patient leaves bed), temperature (for fever, blood circulation, comfortable), chemical constitution (for urine, Antiperspirant, saliva) and the specialty that uses at special time (for oximetry, EEG, metabolism, protein group, genomic expression) Liner.Additional portable sensor can be used and connected by ad hoc by wireless technology.Accordingly it is also possible at bed Monitor patient in addition.Such as, the number that the supply of the sensor in smart machine as panel computer and smart phone is additional According to.Additionally, wrist device measuring pulse, temperature, blood pressure, lung Oxygenated parameters (oxygenization), galvanic skin response (GSR) conductivity limited protein group/genome chip sampled data is provided.Similar equipment be adhered to other extremity or Chest realizes analyzing to monitor such as diet, the effect of gastrointestinal program or the reaction to medicine the sound breathing stomach function regulating. Additionally, body sensor (including accelerometer) is for the signal walked about of monitoring, such as (react or movement) speed, stability (such as, whether patient waves or walk haltingly and blush), vigilance or sleepy, gait and posture.Photographing unit (on bed, glasses or On smart machine) it is used for eye tracks, palor and the tracking of facial expression (such as, instruction is painful or fears).Transaudient Device (on bed or smart machine) is for assessing voice (such as, detecting apoplexy), health creak sound and crack, the compassion of patient The sound of pain or keyword (such as, help or pain).Additionally, smart machine is for the additional input from patient and order (such as, increasing temperature, help).

Mutual by unit 107(or worker) actuator automatically commanded controls sick bed position and (reduces heart disease to send out Make or the damage that causes of cardiac shock, prevent from breathing and snoring problem) and medicament dispenser (be used for avoiding life-threatening Situation).System 10 in one embodiment provides the remote household care monitoring of patient.

Fig. 3 illustrate the normal mode that includes automatically learning to depend on the sensing data of patient for patient automatically and There are the monitoring of guide and the process chart of intervention.Following after the beginning at step 303 place, data processor 15 is in step If determine iteratively in 305 new sensing data whether can with and data be available, then interface 27 is in step 307 Obtain sensing data and record sensing data and be used for being used behind and filing.Processor 15 identifies in a step 309 and is obtained Feature in the sensing data taken and in step 311 tagsort is become normal or abnormal in for patient's instruction just use The often model 316 of state.System also realizes for having a patient or specific within preset time or under one group of environment The doctor of the parameter value of the special circumstances of group patient selects.Normal patient data are being obtained in step 319 learning processor 25 Model 316 uses the training dataset from patient He the data of (optionally) expertise 313.If in step 323 Processor 15 identifies that the feature in acquired sensing data is exception the generation indicating event, leads to the most in step 326 Cross audition, vision or haptic alerts or communication information notifies these data of worker.In response in step 341 by data with alert Check cause worker order, initiate medical intervention (such as, the use of medication management, Medical Equipment) in step 346 And in step 349, record and the medical intervention of sensing data are stored in storage vault 17.In response in step 341 Instruction need not worker's order of intervening, step 351 place termination and with the audition of patient and other worker- Visual communication is selectively initiated in step 343.

Processor 15 is compared to identify by the value by patient parameter or obtained from parameter and threshold value in step 329 Emergency, automatically analyzes the event detected in step 323.If be detected that emergency, then in step 331 Initiate medical intervention action (such as, basic first aid, foot via bed order raise, carry oxygen) and in step 336 Record and the medical intervention of sensing data are stored in storage vault 17.Similarly, if not detecting tight in step 329 Anxious situation, then initiate comfortable action for patient in step 333.In response to the success of the action determined in step 339, process Flow process returns to beginning and step 305 and repeats to process.If not detecting successfully, then handling process is from notifying process 326 Advance.Additionally, in response to worker's status request 361, in step 363, interface 27 obtains sensing data, in step 366 Feature and handling process that processor 15 is extracted in sensing data continue from worker's notifying process 326.

Study processor 25 automatically learns to depend on the normal mode sensing data of patient and an embodiment Middle employing human expert supervises and is input in study process.The crucial decision of system and the crucial decision of overseer user It is recorded and is used in learning model and help treatment in the future.Study processor 25 obtain live patient data, learn across (normally) patient's particular range of the different types of parameter of the most multiple dimensions, monitors abnormal conditions, and control at emergency room and The equipment that the medical science around patient in home care environment is relevant.Use for telemedicine, remotely diagnosis, remotely program With the mutual connectivity to health care personnel, system use intelligence assistant's (unit 107) network to come unusual situation and Trend carries out process and cloud storage, search, query generation, the discovery based on pattern of advanced person.(do not have and nurse system transparent Or doctor's is mutual) perform to include by obtaining different types of patient data constantly from sensor from the unit 107 of bed And provide data to transmit in a distributed fashion to data processor 15 via the network of dynamically configuration, store and process number According to.

Processor 15 local, process data and obtain reasoning and measure (by will in a distributed system or remotely The amount that sensing data combination is correlated with reasoning) and by learning the patient's specific nominal sensor number specific to given patient According to performing diagnosis demonstration.Processor 15 checks sensor function to find by such as sensor degradation to cause not the most constantly Correct reading.System determines that patient's particular safety limits (such as, minimum, maximum and rest heart rate), by by current for patient shape State and processor 25 are compared to remotely via the pattern that sensing data is learnt or perform reality at patient location Time data abnormality detection, this sensing data be given patient from normal condition with from by having the people similar with patient Mouthful statistics situation (include the age, body weight, highly, sex, Pregnancy status) obtain with the patient population of similar medical condition Data collection.System also uses received medical science and social knowledge to perform abnormality detection and show measurement and examine Disconnected data (on network devices and be authorized to be connected on the equipment of grid).

System 10 performs other number in large scale system knowledge base (on data base and one or more processing equipment) According to obtaining and processing function, it is included in data base storage data, from patient data's study, the function of execution auxiliary.System 10 Bridge and the realization of setting up the information about the auxiliary portable equipment being connected to patient that obtains are logical with the audition-vision of patient Letter.System performs action, the most automatically call contact, such as, if be detected that abnormal data, patient attempt leave bed or Wake up.System according to from patient sensing data control environment (such as temperature) or activate message function, control medicine divide Join with electronic equipment (such as TV, microwave) if be detected that breathing problem then rotates sleeping patient and performs simple anxious Rescue function (raising upper body or the lower part of the body the most in bed).System also performs such as provide input to management personnel, law and set Execute the action of management group and assess pathophoresis pattern.

System 10 disposes the most netted (mesh) communication construction of robust, and it does not has infrastructure support the most wherein Arrange in make system equipment can maintain connectivity collaboratively.Such as, in disaster area, patient can be placed on system and make Can stretcher on, it is collected, analyzes and act on and directly collect data from patient.Each system equipment sets also by by system The standby mesh topology created cooperatively maintains the connectivity of center system.Owing to these equipment had both shown as data Gatherer also shows as data and turns originator, so such mesh topology further expands wireless range.Therefore, set from origin system The standby each packet generated selects optimal available route by jumping on other system equipments some.By equipment certainly Body automatically and independently performs the establishment of mesh topology, configures and maintain operation, and does not has any needs of human interaction.Cause This, it is not necessary to technical staff occurs at the website of deployment.

Fig. 4 illustrates the method processing the sensing data obtained, and the method includes self-adapted sensor selection, classification feature Select and respond and action selection.The method has and depends adaptively on the medical condition of patient, current sensor reading, depends on Rely and determine in the nurse of precedent and the expert of doctor and can use the stage of amount of training data.The first step shown in hurdle 403 In Zhou, data processor 15(Fig. 2) in response to patient's medical condition of patient and (a) current sensor parameter value, (b) can At least one in the decision of the clinician in medical science case relatively and (c) available amount of training data, and from available biography Sensor is adaptive selected multiple different sensor, to provide optimal treatment in specific cases.One group of sensor quilt Select such as blood pressure, heart rate, health vibration, the pressure in given body position and the humidity water of a part for patient surface Flat, for monitoring the significant content of medical condition for patient.Use in one embodiment and to be configured by clinician Look-up table performs this adaptive selection.

In exemplary operation, for the patient with heart, it is important that monitoring blood pressure and heart rate.Hold Obtain continuously from the data of selected sensor and deliver the data to shown in the step on hurdle 405 by study The training of reason device 25 execution and sorting phase.Study processor 25 is in response to (i) from the patient of available record of sensor Data volume and (ii) from least one in the type of the patient data of the available record of sensor, connects for multiple differences The patient parameter group received be adaptive selected from multiple difference in functionalitys by study processor use for determining (a) normal model Enclose and the function of at least one in (b) abnormal ranges.The type of the patient data of this record includes that normal and healthy patient passes Sensor reading and abnormal patient sensor reading.The plurality of function includes (a) one-class support vector machine, (b) class interconnection vector At least two in machine and (c) multiclass interconnection vector machine.Such as at-J. Shawe-Taylor and N. Cristianini. Kernel Methods for Pattern Analysis. Cambidge University Press, New York, NY, 2004. (pages 211 and following etc.) describe SVM.Such as at-C. M. Bishop. Pattern Recognition and Machine Learning (Information Science and Statistics). Springer, New York, NY, 2006.(pages 345 and following etc.) in describe RVM.Processor 25 is in response to the available letter about patient Cease and select adaptively between different learning machines.Such as, if limited patient data be recorded (by with in advance Fixed threshold value is compared and is relatively determined) and be available from the data of normal/health sensor reading, then select a class kernel Support vector machine (class SVM).

As follows, orderRepresent withThe d sensor reading of included in N training data example Or the vector of feature.Additionally, orderWithRepresent the soft degree of SVM and support vector weight.Additionally, orderRepresent symmetric positive definite kernel matrixElement.The cost function of one class SVM by

Given, obey:With。

In the data sample group recorded before, have found the weight making this cost function minimize.In classifying step In, class SVM assesses whether new data instance is included in the given threshold value learnt beforeHypersphere in, in kernel spacing Radius include

And biasAs follows:

The border of the point being positioned at this classification inner side and outer side is illustrated in the heart rate on hurdle 405 in blood pressure drawing.If had The sufficient training data being collected carrys out the probability in Modelling feature space, then by processor 25 with another machine learning method (such as interconnection vector machine (RVM)) replaces class SVM.Rather than only returning normal or anomaly classification, RVM causes current state The estimation of posterior probability.Such as, output may indicate that the probability of 95% is that sensing data represents normal patient situation.If There are the information about specific exceptions situation or data, then these can use multiclass learning method to be trained to.Different enforcement Example uses different sensor selection, study and improved methods and is not limited to SVM and RVM.

In step shown in hurdle 407, the result of sort program is used to select adaptively by data processor 15 Select action and response.In the case of determining that multiple different patient parameter group received exceedes the normal range determined, at data Reason device is adaptive selected the action will being performed in response at least one of the following: (a) available from sensor Patient data's amount of record, (b) from the function selected by the available type of patient data of record of sensor, (c) The criticality of patient parameter of type, the medical condition of (d) patient and (e) different reception.Such as, as long as study processes Device 25 does not have the data volume (threshold quantity less than predetermined) of abundance, and a class SVM machine learning method is just used for from selection The patient parameter of one group of sensor acquisition is classified.Underconfidence in this case, when detecting abnormal, in classification To initiate urgent action.Therefore the one group of action simplified is chosen and notifies care-giver or nurse.Data processor 15 as by Clinician initiates operation and the data identifying the action taked is carried out record as being asked.If pattern is tested Survey and data are used for modeling, then the behavior of the clinician recorded initiated in case in the future by study processor 25. If there is enough data be given patient just illustrating Deviant Behavior sufficient credibility (such as, having the probability of 98%) and And in response to the medical condition of patient and the criticality of sensing data, processor 15 notifies the doctor in charge of patient.Such as, If patient has heart and blood pressure is raised to abnormal level, then initiated action immediately by processor 15.

In emergency situations, overseer notify ambulance and advance to simultaneously with the position of contact patients.System also warp Controlled the medicament distribution to patient by the automatic vending machine remotely controlled and automatically remind patient take medicine and verify medicine quilt Take.Such as, if in case of emergency or patient has pain, overseer provides attached via the smart phone being linked to system The medicine added.In this approach, overdose is prevented from and doctor knows exactly which in the case of needs Emergency Assistance When take which kind of medicine at point.System performs, controls and assesses the training module for patient of doctor's regulation.Example As, system periodically initiates interactive game, follows the trail of including the eye motion that recovers for apoplexy or muscular training and for multiple Former joint mobility.Assess test result and predict recovery time.If recovering to be obstructed, then patient and doctor are notified Progress or the needs intervened.System is automatically or remote auto ground or manually mutual and control the device of domestic in response to overseer Tool.Such as, if patient is cold or perspires, the most automatically adjust room temperature, if patient starves, overseer activate microwave.System is also Can be used together with intelligent baby bed, this intelligent baby bed such as monitoring the sleep of child, the position of child (such as exists On side, stomach or back) and abnormal pattern of crying prevent the child of burst dead.

In another embodiment, system, by efficiently controlling the environment of each and accelerating recovery time, monitors And the sensing data (such as blood pressure, blood oxygen saturation SPO2 and heart rate) of emergency room (ER) patient measured by improving.Such as, According to the sick person's development measured, the temperature in bed can be automatically controlled at, automatically activate message function can improve blood Circulate or other measurement can be carried out and avoid a skin ulcer or other harmonic motion problem.Additionally, when detecting the sound of snoring, system is automatic Ground rotates the patient during sleeping.This is used for preventing barrier sleep apnea (OSA).This imbalance be typically characterised by blood oxygen Deficiency, it can cause the oxygen less than 50% saturated.And, the arrhythmia of heart is typically caused by OSA.For right side mental and physical efforts Exhaustion, pulmonary hypertension, stagnated blood of liver, angle edema and dizziness dispose some patients OSA in advance.By control have multiple individually The air mattress of Controllable Air bag realizes rotating.If heart attack or apoplexy are detected, then system performs the most anxious Rescue, such as raise patient's upper body (pressure reducing on involved area) if or cardiovascular system apoplexy, then raise The foot (realizing the more preferable blood circulation in head) of patient.The management immediately of such measure minimizes response time And reduce the infringement as result the most potentially until specialty rescue arrives.

(such as disaster area is (such as, by fire, earthquake, hurricane, flood, nuclear accident, probably at hazardous environment for system monitoring Be afraid of attack etc. to cause)) in people.Such as, in the case of having a large amount of victim and infrastructure to damage, it is often impossible to vertical I.e. help everyone.The state of system monitoring difference patient also helps to identify the disease needing to help immediately of major injuries People.And, system is set up local communication networking and realizes the real time access to patient data.This can be used to assessment in a district Additional help in territory need and therefore help total emergency response.System also monitor aid (such as, fireman, soldier, Doctor, technical staff) state.Such as, consequential emergency (such as, if in a fire house collapse Bury fireman) in the case of, the position following the trail of aid realizes helping immediately.And, assignor's holding in specific location The information of continuous time promotes possible is exposed to toxin, communicable infection and the determination of radiation.Additionally, it is attached to help The sensor of person is used to draw environmental aspect map, and (such as, temperature, ambient sound, radiation, pathogenic bacteria, fuel gas, CO2 are dense Degree) and other people of warning of hazardous region.Pressure, it is lost and to be exposed to the grade of toxic substance also monitored and for making Disaster aid in turn, and distributes their somatotonia comparably and prevents overworked and collapse.

System monitors the sleep of patient by the action of assessment heart rate, breathing, pulse and patient.Each patient is different And be assumed there is different medical conditions.Systems stay ground adapt to and learn for each patient at predefined thing Nominal sensor parameter value output in reason border (such as, minimum, maximum heart rate).If patient wakes up, irritated, attempt Leave bed or abnormality sensor data are shown, then notifying overseer and the local alarm activating in bed.Abnormality sensor number According to include such as patient in past this irregular not shown or the heart rate that increases or slight breathing problem.In these feelings Under condition, the photographing unit that overseer is activated by smart phone while working with other patient and accesses in patients home.Prison The person of superintending and directing assesses situation and talks with patient (or to come via the such as video conference of another exchange method by the speaker in room Communication) such as stop patient and leave bed, inquire for abnormal conditions and provide oral about possible reason or for patient Support.

In the example of operation, John is 65 years old male, and he suffer from heart disease in the past ten years, and John is by unit 107 monitoring, if this unit 107 constantly and is accurately followed the trail of his heart beat and has any change or unusual row For then alerting him.System is collected various types of health and fitness informations from the health of John and (is included temperature, pulse frequency and motion number According to) for following the trail of unexpected falling and health collision.The bed of John is intelligent bed, has some sensings being embedded in bed structure Device, including detecting the chemical sensor of urine, perspiration and saliva and detect motion and the high pressure on body part Pressure transducer.Intelligent bed directly communicates with unit 107 and periodically transmission sensor reading.Unit 107 is from some Source (includes being positioned directly in the local sensor on unit 107 and all remote source such as embedded within the sensor on intelligent bed) Collection information.Unit 107 is checked this information and analyzes it in real time, and estimates the state of patient.Systems stay ground is from patient Sensor reading carries out learning and identifying that the patient of patient is normal by analyzing the historical data from patient sensor collection Parameter readings.System access provide the general data storehouse of health data for various types of diseases and conditions and other Information source.This information of system acquisition and by the historical data of this information and patient and other static data (age of such as patient, Class origin and health records) combine and find normal reading and the sensor reading of patient.Therefore, system adapts itself to From different background and age group and various health problem and the patient of disease.

The intelligent bed of John is also the motor-driven bed of the function with the position for automatically adjusting the head of a bed and tailstock.Two Before individual month, when John lies on his bed, there is main loop systemic shock in him.His immovable and nobody Can cry for help.Unit 107 identifies abnormal blood pressure, heart beating and adnormal respiration and about this situation immediately to emergency medical Personnel send alarm.Before ambulance arrives, system by by the sign of sensor reading and general medical science burning issue and Symptom is compared to identify potential problem, thus utilize preciousness minute, and detect that John may be experienced by circulation Systemic shock.Making foot be higher than head (being commonly called as Trend) additionally, system is known to raise foot, this is standard The shock of the type is relaxed in first aid position, and activates intelligent bed motor to raise the foot of patient and to reduce the head of patient Portion.In this case, system serves as Emergency paramedic intelligently to improve patient care.

Fig. 5 illustrates the example flow of the home care process for monitoring patient automatically and the unusual condition identifying patient Figure.In step 506, followed by the beginning in step 503 place, one group of biography that data processor 15 selects from storage 504 acquisition The sensor parameters data of sensor, and extract data characteristics and pattern in step 509.In step 511, learn processor 25 To the patient data's feature extracted with pattern is classified and determines whether the data of classification indicate specific disease in the step 514 Normal or the Deviant Behavior of people.In response to normal classification, the patient data's feature extracted in step 521 and pattern are used to change Enter patient model and in step 528, store this model, processing and return to step 509.In step 521 from improvement processes Get rid of abnormal data.Under some situations identified in combination based on sensor reading, the normal value of these readings is Depend on the normal value of the data value from other sensor.

System identification is for the group of the most abnormal value of given medical condition the determination of exception in some cases Including a series of step rather than only search or single formula.Additionally, a series for the treatment of includes one group of action and is System again monitors the sensor input of complexity when being adapted to specific cases.System includes for by manufacturer, brand, use The time limit, position, environment detect the different ability of sensor, and sensor data analysis is adapted to the sensor that detects Difference.If determining Deviant Behavior in the step 514, then send police in step 526 learning processor 25 to worker Report and data processor 15 determines whether to take predetermined action in response to Deviant Behavior in step 517.If processed Device 15 takes predetermined action in response to Deviant Behavior, is initiated action by processor 15 the most in step 534, and process returns to Step 506.If processor 15 determines does not takes predetermined action in response to Deviant Behavior, then processor in step 531 15 for sharing the patient population of demographic characteristic (age, body weight, height, sex) with this patient by patient's medical science shape Condition data and history (obtaining from data base 561,563) carry out relevant to medical symptom and treatment data.

System 10 is by checking different types of information (including the medical sensor reading being attached to patient body and bed) Automatically identify the unusual condition of patient.Static information (the such as age is gathered from medical data base 563,561 and information centre And race), general medical science information and statistical information.Being correlated with in response to this, in step 537, processor 15 determines patient's medical science shape Condition is the most critical and if it is notifies worker in step 541.In step 547, processor 15 further determines whether Need first-aid intervention and if it is initiate to take action in step 549 (such as to be adjusted by bed and raise foot relative to head Portion).Process and continue from step 506.

System obtains patient parameter and according to record from the sensor of multiple patients of being attached to and the sensor of patients room Complex patterns detection error, determine that deviation whether substantially and is suitably taken action from different on selections and perform Using as response.System uses the favourable combination of different types of sensing data, process and the method for combination sensor data There is provided data pattern for obtain from learning model for this patient or the data value of patient population identification normal mode and The preassigned pattern of scope compares.Comparing in response to this, system is taken action, and moves including via the actuating of patient bed parts Dynamic patient.

The instruction that systems stay ground record obtains from the sensor of multiple patients of being attached to and the sensor of patients room is sick The patient parameter data of people's health status the deviation of the complex patterns of detection and record.System is based on predetermined adaptive threshold Determine that deviation the most substantially and (includes managing medicine, adjusting patient from the action relating to taking specific to patient Bed and send alarm to personnel) the look-up table that is associated of parametric model in select suitably action.System is within a period of time Carry out learning also from the historical patient particular data obtained by the sensor of multiple patients of being attached to and the sensor of patients room Diagnose new data and act on it.

System receives data from sensor, processes sensing data in combination and produces intermediate object program, by this intermediate object program Predetermined patient particular safety (normally) scope relevant with threshold value, scope, the data being previously recorded and patient compares, Carry out the pattern of detection of complex and determine the most obvious with the deviation of normal range.System uses look-up table or predetermined rule Obvious result is mapped to action, and (look-up table hurdle and sensing data, from sensing data, associated threshold, action task The data that obtain of combination, need the data performing specific action to be associated；This specific action includes managing medicine, specific with this Mode adjusts bed and generates alarm by such as communicating with destination via phone, IP address, Email).System is from prison The data of control carry out study and improve process constantly.

Study processor 25 processes training data group to learn the patient data of the model of statistical knowledge, this statistical knowledge bag Include the patient's particular data about patient's present situation (normality) in complexity, multidimensional, leap time and space, and this study processes Device 25 monitoring may indicate that "abnormal" situation the event needing the medical science made to determine as response.System is by changing bed Temperature, back/pedal position, the activation of message function change the environment of patient, thus control to improve Medical shape Condition or the comfortable parameter of patient.System based on according to multiple forms by patient worn or be embedded in environment (bed, room) Substantial amounts of sensor extract the information of the medical condition about patient and by jointly utilizing expertise and such as making The change detected in the medical condition of patient by the abnormality detection of the data-driven of oneclass classification.System includes realizing patient And user circle of the intervention (including but not limited to provide medicine) of the mutual and care-giver in case of emergency between care-giver Face.System use intelligence assistant (unit 107) network unusual situation and trend are carried out advanced person process and cloud storage, Search, inquiry, discovery based on pattern are, and use for telemedicine, remotely diagnosis, remotely programming and mutual to strong The connectivity of health nursing staff.

System uses the efficient mesh communication framework of robust, and it makes system set in not having the arranging of infrastructure support For maintaining connectivity collaboratively.Such as, in disaster area, patient is placed on the stretcher that system enables, and it is collected, divides Analyse and act on and directly collect data from patient.Each system equipment is also by being created cooperatively by system equipment Mesh topology maintains the connectivity of center system.Show the most at the same time owing to these equipment are not showed only as data collector Originator is turned, so such mesh topology processes the demand of wireless range further for data.Therefore, generate from origin system equipment Each packet select optimal available route by running through other system equipments some to carry out jumping over.By equipment self Automatically and independently perform the establishment of mesh topology, configure and maintain operation, and there is no any needs of human interaction.

Fig. 6 shows the flow chart of the process performed by patient monitoring and interfering system.Followed by step 60 1 In the step 602 started, data processor 15 is in response to patient's medical condition of patient and (a) current sensor parameter value, (b) At least one in the available amount of training data of the decision of the clinician in comparable medical science case and (c), and from can Sensor in be adaptive selected multiple different sensor.In one embodiment, data processor 15 is in response to many The single parameter of individual different parameters exceedes predetermined threshold value and selects the patient of multiple different reception from multiple different groups Parameter group.Sensor includes the chemical sensor of the chemical parameters for sensing body fluid and multiple different patient received Parameter group includes chemical parameters.Sensor also includes the sensor being positioned at least one following position: (a) mattress, (b) Medicated pillow and the supporting member of (c) bed, the sensor associated including metabolic system and protein group or genomic expression sensing Device.Sensor includes the microphone of at least one for detecting in breathing, heart beating and stomach sound and automatically analyzes expression The signal data of the sound detected is to provide the data processor of audio parameter, and the patient parameter group of multiple reception includes This audio parameter.Sensor also includes for detecting pulse, tremble and the vibrating sensor of at least one in onste and from Analyze the data processor of the parameter representing the signal data of vibration detected to provide from vibration to obtain and multiple dynamicly The patient parameter group of different receptions includes the parameter obtained from vibration.Sensor includes for sense blood pressure, blood oxygen saturation The vital sign sensor of at least one in SPO2, ECG signal and temperature and the patient parameter group of multiple different reception Including sign of life parameter.

In step 605, interface 27 receives the data representing multiple different parameters from multiple different sensors, and these are many Individual different sensor is included in sick bed and neutralizes and be attached to the sensor of patient and (include heart rate sensor, respiration pickup and refer to Show the pressure transducer of a pressure spot).The partial high pressure power that bed pressure spot includes tightening, body position and tissue can be compressed At least one in point.Study processor 25 is in response to (i) from patient's number of available record of sensor in step 607 According to amount with (ii) from least one in the type of the patient data of the available record of sensor, for multiple different receptions Patient parameter group be adaptive selected from multiple difference in functionalitys by study processor use for determining (a) normal range (b) function of at least one in abnormal ranges.Normal scope is from having the demographics (bag similar with patient Include the age, body weight, highly, sex, Pregnancy status) obtain with the patient population of similar medical condition.Training data is permissible Both come from this patient (capturing particular condition) and also come from other patients (capturing both common scenarios, health and exception) Promote faster multicategory classification and diagnosis.The type of the patient data of this record includes with Types Below, including normal and healthy Patient sensor reading and abnormal patient sensor reading.The plurality of different function includes (a) one-class support vector machine, (b) At least two in one class interconnection vector machine and (c) multiclass interconnection vector machine.Study processor 25 carry out learning from case and Supervision and data from new case are fed into update this study.But not study in diagnostic mode, and simply learn Applying and be in production model, processor 25 such as includes grader, evaluator, diagnostor and treatment in one embodiment Specifier.

In step 610, study processor 25 is by recording patient parameter value and using in analysis note within a period of time Function selected when the parameter value of record is to determine their scope, is just determining the patient parameter group of different receptions for patient Often scope.In step 613 data processor 15 determine the patient parameter group of this different reception exceed determined by normal model Enclose and determine in response to this and in response to the type of parameter in this set and the medical record information of patient and different receptions The criticality of patient parameter, be adaptive selected the action that will be performed.Multiple predetermined action include: initiate to adjust Sick bed, change give patient medicine, to worker's alarm patient parameter change, as by indicated by clinician labelling From the parameter of sensor and labelling from the parameter of sensor to use in training.Action also includes automatically rotating Patient supports to breathe, raise back or foot, offer medicine, remind patient and monitoring ingestion of medicines amount and chaotic, In undermanned environment, (such as disaster area) is by precedence arrangement treatment.The process of Fig. 6 terminates in step 631 place.

Processor used herein is performed for storing machine-instructions on a computer-readable medium to hold Row task and can include any one in hardware and firmware or the equipment of combination.Processor can also include that storage is used Memorizer in the executable machine readable instructions of the task of execution.Processor is by handling, analyze, revise, change or sending Information uses for executable program or information equipment, and/or passes through information router to outut device, acts on information. Processor can use or include the ability of computer, controller or microprocessor, such as, and uses executable instruction to prop up Join processor to perform the special function not performed by general computer.Processor can couple with other processor any (electrically and/or include can execution unit), it is achieved that the mutual and/or communication between them.Computer program instructions can (include being not limited to general purpose computer or special-purpose computer or other is for producing the place able to programme of machine being loaded onto computer Reason device) on so that the computer programming instruction establishment performed on computer or other processing meanss able to programme is used for realizing The mode of specific function in (one or more) block of (one or more) flow chart.User interface processor or maker It is known element, including electronic circuit or software or combination for generating display image or their part.User Interface includes the display image that one or more realization is mutual with the user of processor or miscellaneous equipment.

As used herein, executable application includes for such as arranging processor in response to user instruction or input (function or the out of Memory of the function of such as operating system, context data acquisition system process system to realize predetermined function System function) code or machine readable instructions.Executable program be code snippet or machine readable instructions, subprogram or its It maybe can perform the part of application for performing the different code section of one or more specific process.These process can be wrapped Include and receive input data and/or parameter, in the input data received, perform operation and/or in response to the input ginseng received Number performs function and provides the output data as result and/or parameter.As used herein, graphic user interface (GUI) bag Include and one or more generated and realize the display image mutual with the user of processor or miscellaneous equipment and phase by video-stream processor The data acquisition of association and process function.

UI also includes that executable program maybe can perform application.Executable program maybe can perform application domination video-stream processor Generate the signal representing UI display image.These signals are fed into the display device of the display image for being watched by user. Executable program maybe can perform to apply also from user input device (such as keyboard, mouse, light pen, touch screen or any other permit Family allowable provides the device of data to processor) receive signal.Under the control that executable program maybe can perform application, process Device is in response to the signal manipulation UI display image received from input equipment.In this mode, user uses input equipment with aobvious Diagram picture is mutual, it is achieved mutual with the user of processor or miscellaneous equipment.Can in response to user command automatically or all Ground or partly execution functions herein and process step.Operate in response to executable instruction or equipment and perform automatically to perform Activity (including step), it is not necessary to the direct promotional activities of user.

The system of Fig. 1-6 and process are not exclusive.Other system, process and journey can be obtained according to the principle of the present invention Sequence order table completes identical target.Although describing the present invention by reference to specific embodiment it should be appreciated that, The embodiment being illustrated and described herein and change are solely for the purpose of illustration.Without departing from the scope of the invention, Can be by those of skill in the art's amendment to current design.System self-adaption ground selects one group of sensing data, study Processor function processes selected sensing data to determine the normal patient scope of data of sensing data, and, it is System exceedes, in response to sensing data, the normal range determined and is adaptive selected action to be performed.Additionally, replaceable Embodiment in, process and apply on the network of the unit that can be positioned in linked, diagram 2 one or more (such as, distribution Formula) in processing equipment.Any one in the function provided in figs. 1-6 and step can use hardware, software or the group of the two Close and realize.Under the regulation of 35 U.S.C. 122 the 6th section, do not explain the claims herein key element, unless used phrase " be used for ... device " be clearly set forth this element.

Claims (20)

1. patient monitoring and an interfering system, including:

Interface, for receiving the data representing multiple different parameters, the plurality of different sensing from multiple different sensors The sensor that device is included in sick bed and the sensor being attached to patient, described sensor includes:

Heart rate sensor,

Respiration pickup, and

The pressure transducer of instruction bed pressure spot；

Study processor, for by recording patient parameter value and analyzing recorded parameter value to determine it within a period of time Scope, determine the normal range of multiple described different patient parameter group received for described patient；And

Data processor, for determine this different patient parameter group received exceed determined by normal range；And respond Determine and in response to the type of parameter in this set and the medical record information of described patient in this, initiate the adjustment to sick bed and At least one of the following:

A () changes the medicine giving patient, and

B () changes to worker's alarm patient parameter；

Wherein, described data processor is in response to (a) current sensor parameter value, (b) facing in comparable medical science case At least one in the decision of bed doctor and (c) available amount of training data and patient's medical condition of patient, can from predetermined Sensor in be adaptive selected the plurality of different sensor.

System the most according to claim 1, wherein,

Described data processor exceed predetermined threshold value in response to the single parameter of the plurality of different parameter and from multiple not Same group selects the plurality of described different patient parameter group received.

System the most according to claim 1, wherein,

Described sensor includes the chemical sensor of the chemical parameters for sensing body fluid and the plurality of described different connecing The patient parameter group received includes described chemical parameters.

System the most according to claim 1, wherein,

Described sensor includes the sensor being positioned at least one following position: (a) mattress, (b) medicated pillow and (c) bed Supporting member.

System the most according to claim 1, wherein,

Described sensor includes for detecting the microphone of at least one in breathing, heart beating and stomach sound or vibration and automatically The signal data of the sound that ground analysis expression detects is to provide the described data processor of audio parameter, and the plurality of institute State the different patient parameter groups received and include described audio parameter.

System the most according to claim 1, wherein,

Described sensor includes for detecting pulse, tremble and at least one (a) vibrating sensor and (b) in onste takes the photograph Camera, described data processor is automatically analyzed and is represented that the signal data of the vibration detected provides the parameter obtained from vibration And the plurality of described different patient parameter group received includes the described parameter obtained from vibration.

System the most according to claim 1, wherein,

Described sensor includes the life of at least one in sense blood pressure, blood oxygen saturation SPO2, ECG signal and temperature Sign sensor and the plurality of described different patient parameter group received include sign of life parameter.

System the most according to claim 1, wherein,

At least one in the partial high pressure force that described bed pressure spot includes tightening, body position and tissue can be compressed.

System the most according to claim 1, wherein,

Described sensor includes the sensor that metabolic system is associated, and

Described data processor automatically initiate following at least one:

A) automatically rotate patient to support to breathe,

B) patients back or foot are automatically raised,

C) it is automatically provided medicine, and

D) determine in response to described, automatically by precedence arrangement treatment.

System the most according to claim 1, wherein,

Described sensor includes protein group or genomic expression sensor.

11. systems according to claim 1, wherein,

Described normal range is to obtain from the patient population with the demographics similar with patient and similar medical condition Arrive, described demographics include the age, body weight, highly, sex, Pregnancy status.

12. systems according to claim 1, wherein,

Described study processor in response to (i) from sensor available record patient data's amount and (ii) from sensor Available record patient data type at least one, for the plurality of described different patient received ginseng Array be adaptive selected from multiple different predetermined functions by described study processor use for determining (a) normal model Enclose and the function of at least one in (b) abnormal ranges.

13. systems according to claim 12, wherein,

The type of the patient data of described record includes the normal and type of healthy patient sensor reading and abnormal patient's sensing The type of device reading.

14. systems according to claim 12, wherein,

The plurality of different predetermined function includes that (a) one-class support vector machine, (b) class interconnection vector machine and (c) multiclass are closed At least two in connection vector machine.

15. systems according to claim 12, wherein,

Described data processor is in response to determined by determining that the plurality of described different patient parameter group received exceedes Normal range, described data processor is adaptive selected the action will being performed in response at least one of the following: A () is from patient data's amount of the available record of sensor, (b) from the class of the patient data of the available record of sensor The type of the function selected by type and (c).

16. systems according to claim 12, wherein,

Described data processor is in response to determined by determining that the plurality of described different patient parameter group received exceedes Normal range, described data processor predetermined on is adaptive selected in response at least one of the following from multiple The action that will be performed: the medical condition of (a) patient and the criticality of (b) described different patient parameter received.

17. systems according to claim 16, wherein,

The plurality of predetermined action includes: initiate to adjust sick bed, change gives the medicine of patient, to worker alarm patient ginseng Number changes, as by indicated by clinician labelling from the parameter of sensor and labelling from the parameter of sensor with Just use in training.

18. systems according to claim 1, including:

Linking described sensor and the mesh communication network of described system, described network enables described system equipment to tie up collaboratively Hold connectivity.

19. 1 kinds of patient monitorings and interfering system, including:

Interface, for receiving the data representing multiple different parameters, the plurality of different sensing from multiple different sensors The sensor that device is included in sick bed and the sensor being attached to patient；

Study processor, for by recording patient parameter value and analyzing recorded parameter value to determine it within a period of time Scope, determine the normal range of multiple described different patient parameter group received for described patient；And

Data processor, in response to determined by determining that the plurality of described different patient parameter group received exceedes Normal range, will be held in response at least one of the following from multiple different predetermined on being adaptive selected The action of row: the medical condition of (a) patient, and the criticality of (b) described different patient parameter received, described action Give the medicine of patient including (i) initiating adjustment sick bed, (ii) change and (iii) change to worker's alarm patient parameter In at least one；

Wherein, described data processor is in response to (a) current sensor parameter value, (b) facing in comparable medical science case At least one in the decision of bed doctor and (c) available amount of training data and patient's medical condition of patient, and from predetermined Available sensor is adaptive selected the plurality of different sensor.

20. 1 kinds of patient monitorings and interfering system, including:

Interface, for receiving the data representing multiple different parameters, the plurality of different sensing from multiple different sensors The sensor that device is included in sick bed and the sensor being attached to patient；

Study processor, is used for

In response to (i) from the plurality of different sensor available record patient data's amount and (ii) from described many At least one in the type of the patient data of the available record of individual different sensor, for the plurality of described different The patient parameter group received is adaptive selected for determining (a) normal range, (b) from multiple different predetermined functions The function of at least one in the unusual combination of abnormal ranges and (c) scope, and

During by recording patient parameter value within a period of time and using parameter value that analysis recorded to determine their scope Selected function, determines normal range or the degree of multiple described different patient parameter group received for described patient； And

Data processor, in response to determine the plurality of described different patient parameter group received determined by exceeding normal Scope, will be performed in response at least one of the following from multiple different predetermined on being adaptive selected Action: (a) is from patient data's amount of the available record of sensor, (b) from patient's number of the available record of sensor According to type and (c) selected by the type of function；

Wherein, described data processor is in response to (a) current sensor parameter value, (b) facing in comparable medical science case At least one in the decision of bed doctor and (c) available amount of training data and patient's medical condition of patient, and from predetermined Available sensor is adaptive selected the plurality of different sensor.