CN110893099B

CN110893099B - Intelligent early warning system for sudden diseases in sleep state

Info

Publication number: CN110893099B
Application number: CN201911174449.9A
Authority: CN
Inventors: 薛红涛; 吴蒙; 周嘉文
Original assignee: Jiangsu University
Current assignee: Jiangsu Weihang Automobile Communication Technology Co ltd
Priority date: 2019-11-26
Filing date: 2019-11-26
Publication date: 2022-05-20
Anticipated expiration: 2039-11-26
Also published as: CN110893099A

Abstract

The invention discloses an intelligent early warning method for sudden illness in sleep state in the field of medical treatment, which comprises the steps of constructing a diagnosis model by using a Bayesian network intelligent algorithm, taking an intelligent pillow as a carrier, acquiring sign information data in the sleep state by using a sensor in the intelligent pillow, comparing the sign information data with sign data in a normal state, acquiring a corresponding sign information data fluctuation value set by using a sign data processing module, and inputting the sign information data fluctuation value into a Bayesian network diagnosis module for diagnosing abnormal state of a body; the early warning module is combined with the diagnosis result to make corresponding early warning response, timely cure the sudden illness, accurately diagnose the sudden illness and the grade of the symptom of people in the sleeping process, and fully consider the change of the symptom caused by the accumulation of time segments, thereby avoiding the diagnosis error caused by the fluctuation of sign data in short time such as the change of sleeping posture, dreaming and the like.

Description

Intelligent early warning system for sudden diseases in sleep state

Technical Field

The invention relates to the field of medical treatment, relates to a disease early warning method, and particularly relates to an early warning system for sudden diseases when people are in a sleep state.

Background

One third of the life time of people is spent in sleep, with the acceleration of life rhythm, the serious influence is brought to people's sleep quality and health to overload working pressure etc. for example, the latent apnea syndrome that arouses can lead to problems such as hypertension, coronary heart disease, apoplexy and sudden death, seriously threatens people's health. The death cases caused by the failure of timely effective medical treatment due to sudden diseases are frequently seen, and therefore, various health monitoring devices are provided, which can monitor various physiological indexes of people in daily life, count various physiological indexes to form health reports and feed back the health reports in time, so that people can know the self health trends in time. However, as the number of elderly people living alone increases, especially for the middle-aged and elderly people suffering from myocardial infarction, cerebral hemorrhage, heart disease, etc., it is a necessary measure to monitor their physical status during sleep, and there is a strong desire for timely and effective medical treatment for patients suffering from sudden sleep disorders. The Chinese patent application No. 201711455386.5 provides a health detection method of an intelligent pillow based on a hotel and the intelligent pillow, physiological parameter information of a sleeper is acquired in real time and analyzed through a sensor installed on the intelligent pillow, if the physiological parameter information exceeds a preset range and the duration exceeding the preset range is longer than a first preset duration, first alarm information used for prompting that the body is abnormal is sent, people can know whether the body is abnormal in time, and treatment measures can be taken at the optimal time under the condition that the body is abnormal. However, the method sets that the body is judged to be abnormal as long as one physiological data exceeds a threshold value, multiple physical sign data are not comprehensively considered, and the severity of the body abnormality of people in sleeping cannot be judged; meanwhile, the method ignores errors caused by fluctuation of physical sign data due to factors such as sleep posture change, dreaming and the like in the sleeping process of people, and the diagnosis accuracy is not high.

Disclosure of Invention

Aiming at the defects of the existing method, the invention provides an intelligent early warning system for sudden diseases in a sleep state, which comprehensively considers the fluctuation of a plurality of sign data and sign data to judge the severity of sudden diseases and diseases when people sleep, and has high diagnosis accuracy.

The technical scheme adopted by the invention is to realize the following steps:

step A: reading normal human body sign data in a network big data platform, taking an average value of the normal human body sign data as a human body sign fluctuation value, establishing sample data, and constructing a Bayesian network diagnosis model taking the human body sign fluctuation value as input and the conditional probability of a body state as output; the method comprises the steps that a Bayesian network diagnosis model is arranged in a Bayesian network diagnosis module, the input end of the Bayesian network diagnosis module is connected with the output end of a sign data processing module, the input end of the sign data processing module is connected with a sensor, the output end of the Bayesian network diagnosis module is respectively connected with the input ends of a first early warning module and a comprehensive diagnosis module, the output end of the comprehensive diagnosis module is respectively connected with a second early warning module and an initial parameter updating module, and the first early warning module and the second early warning module send information to a user side APP in a wireless mode; an intelligent early warning system is formed by a sensor, a sign data processing module, a Bayesian network diagnosis module, a comprehensive diagnosis module, an initial parameter updating module, a first early warning module and a second early warning module, and the intelligent early warning system is integrally arranged in a pillow for a user to sleep;

the diagnostic model building module fluctuates the heart rate5 human body sign fluctuation values of data delta H, pulse fluctuation data delta P, blood pressure fluctuation data delta BP, blood oxygen saturation fluctuation data delta BS and body temperature fluctuation data delta T are used as child nodes, and the body state S is used_iAs a parent node, i ═ 0,1,2, a bayesian network structure, body state S, is constructed_iIs divided into a normal state S₀Mild disease state S₁And severe disease state S₂Three kinds of the above;

conditional probability P (S) of body state_i|△H_j,△P_m,△BP_n,△BS_p,△T_q) More than or equal to 50 percent, judging the corresponding body state S_i，P(S_i|△H_j,△P_m,△BP_n,△BS_p,△T_q) Is expressed as Δ H ═ Δ H_j，ΔP＝ΔP_m，ΔBP＝ΔBP_n，ΔBS＝ΔBS_p，ΔT＝ΔT_qUnder the condition of (1), the physical state is S_iThe conditional probability of (a); p (S)_i) Indicating that the body state is S_iThe total probability of (c); p (. DELTA.H)_j)，P(△P_m)，P(△BP_n)，P(△BS_p)，P(△T_q) Each represents Δ H ═ Δ H_j，ΔP＝ΔP_m，ΔBP＝ΔBP_n，ΔBS＝ΔBS_p，ΔT＝ΔT_qThe total probability of (c); p (. DELTA.H)_j|S_i) Is shown in the body state of S_iUnder the condition of (1), Δ H ═ Δ H_jThe probability of (d); p (DELTA P)_m|S_i) Is shown in the body state of S_iUnder the condition of (1), Δ P ═ Δ P_mThe conditional probability of (a); p (Delta BP)_n|S_i) Is shown in the body state of S_iUnder the condition (1), Δ BP ═ Δ BP_nThe conditional probability of (a); p (Delta BS)_p|S_i) Is shown in the body state of S_iUnder the condition of (1), Δ BS ═ Δ BS_pThe conditional probability of (a); p (Delta T)_q|S_i) Is shown in the body state of S_iUnder the condition of (1), Δ T ═ Δ T_qThe conditional probability of (1), 1,2, j is 1,2,3,4,5,6,7, m is 1,2,3,4,5,6,7, n is 1,2,3,4,5,6,7, p is 1,2,3,4,5,6,7, q is 1,2,3,4,5,6, 7;

and B: capturing packets within a current single time slice using a sensorUser physical sign data including heart rate H, pulse P, blood pressure BP, blood oxygen saturation BS and body temperature T form a physical sign data set D { H, P, BP, BS, T }; the physical sign data processing module receives a physical sign data set D { H, P, BP, BS, T } collected by a sensor and an internal initial human body normal physical sign information data set D'₀{H'₀,P'₀,BP'₀,BS'₀,T'₀Subtracting one from the other correspondingly to obtain a corresponding human body sign information fluctuation value set delta D, delta H, delta P, delta BP, delta BS, delta T;

and C: the Bayesian network diagnosis module diagnoses the body state of the user by utilizing the human body sign information fluctuation value set delta D {. DELTA H, {. DELTA P,. DELTA BP,. DELTA BS,. DELTA T } to obtain the normal state S₀Mild disease state S₁Severe disease state S₂The diagnosis result of (1); the Bayesian network diagnosis module outputs a voltage signal V corresponding to the diagnosis result₀、V₁、V₂Normal state S₀Corresponds to V₀Mild disease state S₁Corresponds to V₁Severe disease state S₂Corresponds to V₂(ii) a Diagnosing as a normal state S₀The corresponding physical sign data set D { H, P, BP, BS, T } is a normal physical sign data set D'_a；

Step D: the Bayesian network diagnosis module converts the voltage signal V into a voltage signal₂The voltage signal V is transmitted to a first early warning module₀、V₁And a normal signs data set D'_aThe first early warning module sends out a notice through a user side APP in the comprehensive diagnosis module, and the comprehensive diagnosis module sends a voltage signal V₀、V₁And a normal signs data set D'_aStoring;

the comprehensive diagnosis module stores the voltage signal V₀、V₁And a normal signs data set D'_aAnd then, the intelligent early warning system enters the next time slice, the steps B-D are repeated, and the comprehensive diagnosis module continuously stores the voltage signal V₀、V₁And a normal signs data set D'_aAll the voltage signals stored in the set continuous M time slices are V₁Or a voltage signal V₀And V₁Voltage signal V alternately appearing and in N successive time slices₁Appear for M times, and the ratio of M/N is less than or equal to 50 percent<1, judging the physical state of the user to be mild disease S₁。

The invention adopts the technical scheme that the method has the beneficial effects that:

1. the intelligent pillow is used as a carrier, the sensor in the intelligent pillow is used for collecting sign information data in a sleeping state, the sign information data is compared with sign data in a normal state, a corresponding sign information data fluctuation value set is obtained through a sign data processing module, and the sign information data fluctuation value is input into a Bayesian network diagnosis module for diagnosing abnormal body states; the early warning module makes corresponding early warning response by combining the diagnosis result, timely treats sudden diseases, and particularly provides intelligent, accurate and rapid medical protection for the solitary old people who are prone to disease.

2. The invention builds the diagnosis model by using the Bayesian network intelligent algorithm, can accurately diagnose sudden diseases and symptoms of people in the sleeping process, and fully considers the symptoms change caused by time segment accumulation, thereby avoiding the diagnosis error caused by the fluctuation of sign data in short time such as sleeping posture change, dreaming and the like.

Drawings

FIG. 1 is a block diagram of the construction of a Bayesian network diagnostic module;

FIG. 2 is a diagram of the Bayesian network of FIG. 1;

FIG. 3 is a block diagram of a hardware structure connection between an intelligent early warning system for sudden illness in sleep state and a user side APP;

FIG. 4 is a schematic diagram of a first integrated diagnostic rule;

FIG. 5 is a schematic diagram of a second integrated diagnostic rule;

fig. 6 is a flow chart of the operation of the present invention.

Detailed Description

The method comprises four stages, wherein the first stage is to construct a Bayesian network diagnosis model; the second stage is based on a Bayesian network diagnosis model, and the body state of the user in a single time slice is diagnosed; the third stage is to combine a plurality of time segments to comprehensively judge the body state of the user; and the fourth stage is to update the initial parameters after the comprehensive judgment of the body state of the user is finished. Referring to fig. 6, the specific steps are as follows:

in the first stage, a Bayesian network diagnostic model is constructed. Referring to fig. 1, a large amount of normal human body sign data in a network big data platform is read through a diagnosis model building module, and sample data is built. The normal human body sign data considers the fluctuation conditions of various possible human body sign data, comprehensively and comprehensively analyzes various possible human body sign fluctuations, and takes the average value as the human body sign fluctuation value.

The sample data of the human body sign fluctuation value are as follows:

1. heart rate fluctuation data ah. The plurality of possible heart rate fluctuation data are respectively Delta H₁，ΔH₂，ΔH₃，ΔH₄(ΔH₄＝0)，ΔH₅，ΔH₆，ΔH₇And Δ H₁<ΔH₂<ΔH₃<ΔH₄<ΔH₅<ΔH₆<ΔH₇. The heart rate fluctuation data Delta H is Delta H₁、ΔH₂、ΔH₃、ΔH₄、ΔH₅、ΔH₆、ΔH₇Average of these 7 data.

2. Pulse fluctuation data Δ P. The pulse fluctuation data is respectively delta P₁，ΔP₂，ΔP₃，ΔP₄(ΔP₄＝0)，ΔP₅，ΔP₆，ΔP₇And Δ P₁<ΔP₂<ΔP₃<ΔP₄<ΔP₅<ΔP₆<ΔP₇. The pulse fluctuation data Δ P is Δ P₁、ΔP₂、ΔP₃、ΔP₄、ΔP₅、ΔP₆、ΔP₇Average of these 7 data.

3. Blood pressure fluctuation data Δ BP. The data of various possible blood pressure fluctuations are respectively delta BP₁，ΔBP₂，ΔBP₃，ΔBP₄(ΔBP₄＝0)，ΔBP₅，ΔBP₆，ΔBP₇And Δ BP₁<ΔBP₂<ΔBP₃<ΔBP₄<ΔBP₅. The blood pressure fluctuation data Delta BP is Delta BP₁、ΔBP₂、ΔBP₃、ΔBP₄、ΔBP₅、ΔBP₆、ΔBP₇Average of these 7 data.

4. Blood oxygen saturation fluctuation data Δ BS. The multiple possible blood oxygen saturation fluctuation data are respectively delta BS₁，ΔBS₂，ΔBS₃，ΔBS₄(ΔBS₄＝0)，ΔBS₅，ΔBS₆，ΔBS₇And Δ BS₁<ΔBS₂<ΔBS₃<ΔBS₄<ΔBS₅<ΔBS₆<ΔBS₇. Fluctuation data of blood oxygen saturation Δ BS is Δ BS₁、ΔBS₂、ΔBS₃、ΔBS₄、ΔBS₅、ΔBS₆、ΔBS₇Average of these 7 data.

5. Body temperature fluctuation data Δ T. The data of various possible body temperature fluctuations are respectively delta T₁，ΔT₂，ΔT₃，ΔT₄(ΔT₄＝0)，ΔT₅，ΔT₆，ΔT₇And Δ T₁<ΔT₂<ΔT₃<ΔT₄<ΔT₅<ΔT₆<ΔT₇. Body temperature fluctuation data DeltaT is DeltaT₁、ΔT₂、ΔT₃、ΔT₄、ΔT₅、ΔT₆、ΔT₇Average of these 7 data.

Referring to fig. 2, the diagnostic model building module takes 5 human body sign fluctuation values of heart rate fluctuation data Δ H, pulse fluctuation data Δ P, blood pressure fluctuation data Δ BP, blood oxygen saturation fluctuation data Δ BS, and body temperature fluctuation data Δ T as sub-nodes, and takes a possibly-occurring body state S_iAs a parent node, a Bayesian network structure is constructed in which a body state S_iIs divided into a normal state S₀Mild disease state S₁And severe disease state S₂These three types, i ═ 0,1, 2. Said normal state S₀Refers to the physical health state of the user, or the user has diseasesBut in a stable, non-diseased state; mild disease state S₁The sign data set of the user fluctuates to some extent compared with the sign data set in a normal state, but the life is not endangered; severe disease state S₂The sign data set of the user fluctuates greatly in a short time compared with the sign data set in a normal state, and the life of the user is possibly threatened.

The diagnostic model building module adopts the following Bayesian formula to calculate the S of different body states under the condition of different sample data_iThereby constructing a conditional probability table. The Bayesian equation is as follows:

where i is 0,1,2, j is 1,2,3,4,5,6,7, m is 1,2,3,4,5,6,7, n is 1,2,3,4,5,6,7, P is 1,2,3,4,5,6,7, q is 1,2,3,4,5,6,7, P (S) (1, 2,3,4,5,6,7, q is 1,2, 7, P (S) (i.e., P is n, P_i|△H_j,△P_m,△BP_n,△BS_p,△T_q) Is expressed as Δ H ═ Δ H_j，ΔP＝ΔP_m，ΔBP＝ΔBP_n，ΔBS＝ΔBS_p，ΔT＝ΔT_qUnder the condition of (1), the physical state is S_iThe conditional probability of (a); p (S)_i) Indicating that the body state is S_iThe total probability of (c); p (. DELTA.H)_j)，P(△P_m)，P(△BP_n)，P(△BS_p)，P(△T_q) Respectively represent Δ H ═ Δ H_j，ΔP＝ΔP_m，ΔBP＝ΔBP_n，ΔBS＝ΔBS_p，ΔT＝ΔT_qThe total probability of (c); p (. DELTA.H)_j|S_i) Is shown in the body state of S_iUnder the condition of (1), Δ H ═ Δ H_jThe probability of (d); p (DELTA P)_m|S_i) Is shown in the body state of S_iUnder the condition of (1), Δ P ═ Δ P_mThe conditional probability of (a); p (Delta BP)_n|S_i) Is shown in the body state of S_iUnder the condition (1), Δ BP ═ Δ BP_nThe conditional probability of (a); p (Delta BS)_p|S_i) Is shown in the body state of S_iUnder the condition of (1), Δ BS ═ Δ BS_pThe conditional probability of (a); p (Delta T)_q|S_i) Is shown in the body state of S_iUnder the condition of (1), Δ T ═ Δ T_qThe conditional probability of (2). The probability values may be obtained through sample data statistics.

And (4) combining the Bayesian network structure and the conditional probability table to construct a Bayesian network diagnosis model. The built model takes the fluctuation value of human body physical sign as input and the body state S_iIs output, when the conditional probability P (S) is output_i|△H_j,△P_m,△BP_n,△BS_p,△T_q) Not less than 50%, the corresponding body state S can be determined_i。

Referring to fig. 3, the bayesian network diagnosis model to be constructed is arranged in the bayesian network diagnosis module, the output end of the bayesian network diagnosis module is connected with the output end of the sign data processing module, the input end of the sign data processing module is connected with the sensor, the output end of the bayesian network diagnosis module is respectively connected with the input ends of the first early warning module and the comprehensive diagnosis module, the output end of the comprehensive diagnosis module is respectively connected with the second early warning module and the initial parameter updating module, and the output ends of the first early warning module and the second early warning module send information to the user side APP in a wireless mode. Therefore, the intelligent early warning system is formed by the sensor, the sign data processing module, the Bayesian network diagnosis module, the comprehensive diagnosis module, the initial parameter updating module, the first early warning module and the second early warning module, and is integrally installed in a pillow for a user to sleep, so that the intelligent early warning system is called as an intelligent pillow. Wherein, the sensor is human sign information sensor for gather the sign data of user when sleeping, user side APP generally is patient family members, and first, second early warning module output early warning information gives user side APP, in order to remind patient family members.

And a second stage of diagnosing the physical state of the user in a single time segment based on the Bayesian network diagnosis model. The method comprises the following steps:

when the intelligent early warning system in the intelligent pillow works, 1 minute is taken as the length of a time slice, and the time slice is the time of one-time circulating work of the system and comprises the whole process of collecting physical sign data, making diagnosis, storing data and making corresponding early warning by the system.

And acquiring user sign data in a current single time slice by using a sensor to form a sign data set D { H, P, BP, BS, T }. The physical sign data comprise heart rate H, pulse P, blood pressure BP, blood oxygen saturation BS and body temperature T. The data set is defined as a vital signs data set D { H, P, BP, BS, T }.

The sensor outputs the collected user sign data information to the sign data processing module, the sign data processing module receives a sign data set D { H, P, BP, BS, T } collected by the sensor and an initial human body normal sign information data set D 'arranged in the sign data processing module'₀{H'₀,P'₀,BP'₀,BS'₀,T'₀Comparing, respectively carrying out corresponding one-to-one subtraction calculation, and obtaining corresponding human body physical sign information fluctuation value sets delta D { [ delta ] H, [ delta ] P, [ delta ] BP, [ delta ] BS, [ delta ] T }, namely, [ delta ] H { [ H-H'₀，△P＝P-P'₀，△BP＝BP-BP'₀，△BS＝BS-BS'₀，△T＝T-T'₀. Wherein, H'₀,P'₀,BP'₀,BS'₀,T'₀Respectively is the average value of all normal human body sign data collected in the last working period of the intelligent early warning system.

The physical sign data processing module inputs a physical sign data set D { H, P, BP, BS, T } and a human body physical sign information fluctuation value set delta D { [ delta ] H, [ delta ] 0P, [ delta ] 1BP, [ delta ] BS, [ delta ] T } into the Bayesian network diagnosis module together, the Bayesian network diagnosis module diagnoses the body state of the user by using the human body physical sign information fluctuation value set delta D { [ delta ] H, [ delta ] P, [ delta ] BP, [ delta ] BS, [ delta ] T to obtain a diagnosis result, and the diagnosis result is the normal state S of the body₀Mild disease state S₁Severe disease state S₂. The Bayesian network diagnosis module outputs a voltage signal V corresponding to the diagnosis result₀、V₁、V₂In which S is₀Corresponds to V₀，S₁Corresponds to V₁，S₂Corresponds to V₂. Meanwhile, when the diagnosis result is obtained, the normal state S is diagnosed₀The corresponding physical sign data set D { H, P, BP, BS, T } is defined as a normal physical sign data set D'_a。

Bayesian network diagnosis module sends voltage informationNumber V₂Inputting the voltage signal V into a first early warning module₀、V₁And a normal signs data set D'_aAnd inputting the data to a comprehensive diagnosis module. The first early warning module receives the information that the body is in a serious disease state S₂Voltage signal V of₂And sending a notice through the user side APP to inform the family members, and simultaneously alarming to a 120 emergency center to request emergency treatment. The comprehensive diagnosis module receives a voltage signal V₀、V₁When it is, it indicates that the body is in a normal state S₀Or mild disease state S₁. Normal state S diagnosed by a single time slice₀Or mild disease state S₁Possibly caused by the fluctuation of feature data in a short time such as the change of sleeping posture, dreaming and the like of the user, misdiagnosis may be generated, and the diagnosis accuracy is difficult to ensure, so that the comprehensive diagnosis of the physical state of the user needs to be carried out by combining a plurality of time slices.

And in the third stage, the body state of the user is comprehensively judged by combining a plurality of time slices, and the method specifically comprises the following steps:

when the comprehensive diagnosis module receives the voltage signal V₀And V₁And normal sign data set D'_aThen, the voltage signal V is converted into a voltage signal₀、V₁And a normal signs data set D'_aAnd (4) storing. And the system enters the next time slice, and then the second stage of work is repeated in a circulating way, namely the next minute of the intelligent early warning system is taken as the length of the time slice, the physical sign data of the user in the current single time slice is collected by utilizing the sensor, and the physical state of the user in the single time slice is diagnosed. Thus, the comprehensive diagnosis module continuously stores the voltage signal V₀、V₁And a normal signs data set D'_aAt this point, a may take 1,2,3 … … to represent the number of advances for a single time slice in the system cycle. And counting the voltage signal V from the current time slice₀、V₁The number of occurrences. The comprehensive diagnosis module takes a preset time threshold value M as a reference, and if the voltage signals stored in the continuous M time slices are all V₁If the diagnosis results of the M time slices are mild pathology, the physical state of the user can be determined to be mild pathology S₁A diagnostic gauge as shown in FIG. 4Then drawing; or a voltage signal V₀And V₁Alternately occurring, and in N successive time slices, the voltage signal V₁The diagnosis result of M times of occurrence, namely M time slices, is mild disease S₁Wherein 50 percent is less than or equal to M/N<1, the physical state of the user can be judged to be mild disease S₁Such as the diagnostic rule graph shown in fig. 5.

When a plurality of time slices are passed, the diagnosis result is finally determined as a mild disease state S₁Then, the mild disease state S₁Corresponding voltage signal V₁The voltage is transmitted to a second early warning module which receives a voltage signal V₁And then, a notice is sent out through the user side APP to inform the family members of the patient and remind the family members to carry out corresponding medical care measures on the patient.

In particular, during the third stage of comprehensive diagnosis, the result is a severe pathology S, as diagnosed in a subsequent time segment₀If so, the diagnosis is stopped, the first early warning module informs the family members through the user side APP, and simultaneously alarms the 120 emergency center to request emergency treatment.

The fourth stage is an initial parameter updating stage, and comprises the following specific steps:

step 1: after the third stage is finished, the system finishes the diagnosis of the body state of the user and corresponding early warning, and the comprehensive diagnosis module stores the voltage signal V₀And normal signs data set D'_aThe voltage signal is transmitted to an initial parameter updating module which receives the voltage signal V₀And normal signs data set D'_aAnd then, calculating the average value of all normal sign data acquired in the system operation process to obtain a new normal sign data set as a human body normal sign data set D 'in the next system operation'₀. The method of updating data is as follows:

assuming that the diagnosis in Q time slices is in a normal state in the total operation process of the system, the initial parameter updating module is operated to extract Q normal state sign data sets D 'stored in the comprehensive diagnosis module'₁{H'₁,P'₁,BP'₁,BS'₁,T'₁}，D'₂{H'₂,P'₂,BP'₂,BS'₂,T'₂}，……，D'_Q{H'_Q,P'_Q,BP'_Q,BS'_Q,T'_QAnd averaging the corresponding data in the data set to obtain a human body normal sign data set D'₀{H'₀,P'₀,BP'₀,BS'₀,T'₀And the data set is used as a human body normal state sign data set when the system operates again. Wherein, the first and the second end of the pipe are connected with each other,

y＝1,2,3...Q。

the initial parameter updating module updates the updated normal sign data set D'₀{H'₀,P'₀,BP'₀,BS'₀,T'₀Inputting the data to a sign data processing module, wherein the sign data processing module stores the updated normal sign data set which is used as a human body normal sign data set D 'in the next system operation'₀And comparing the next time of system operation with the physical sign data set D { H, P, BP, BS, T } acquired by the sensor to obtain a human body physical sign information fluctuation value set delta D { [ delta ] H, [ delta ] P, [ delta ] BP, [ delta ] BS, [ delta ] T } of the next time of system operation, and referring to the step two.

Claims

1. An intelligent early warning system for sudden diseases in a sleep state is characterized in that the intelligent early warning system realizes the following steps:

the diagnosis model building module takes 5 human body sign fluctuation values of heart rate fluctuation data delta H, pulse fluctuation data delta P, blood pressure fluctuation data delta BP, blood oxygen saturation fluctuation data delta BS and body temperature fluctuation data delta T as sub-nodes, and takes the body state S as a sub-node_iAs a parent node, i ═ 0,1,2, a bayesian network structure, body state S, is constructed_iIs divided into a normal state S₀Mild disease state S₁And severe disease state S₂Three kinds of the above;

conditional probability P (S) of body state_i|△H_j,△P_m,△BP_n,△BS_p,△T_q) More than or equal to 50 percent, judging the corresponding body state S_i，P(S_i|△H_j,△P_m,△BP_n,△BS_p,△T_q) Is expressed as Δ H ═ Δ H_j，ΔP＝ΔP_m，ΔBP＝ΔBP_n，ΔBS＝ΔBS_p，ΔT＝ΔT_qUnder the condition of (1), the physical state is S_iThe conditional probability of (a); p (S)_i) Indicating that the body state is S_iThe total probability of (c); p (. DELTA.H)_j)，P(△P_m)，P(△BP_n)，P(△BS_p)，P(△T_q) Each represents Δ H ═ Δ H_j，ΔP＝ΔP_m，ΔBP＝ΔBP_n，ΔBS＝ΔBS_p，ΔT＝ΔT_qThe total probability of (c); p (. DELTA.H)_j|S_i) Is shown in the body state of S_iUnder the condition of (1), Δ H ═ Δ H_jThe probability of (d); p (DELTA P)_m|S_i) Is shown in the body state of S_iUnder the condition of (1), Δ P ═ Δ P_mThe conditional probability of (a); p (Delta BP)_n|S_i) Is shown in the body state of S_iStrip ofUnder the condition, Δ BP ═ Δ BP_nThe conditional probability of (a); p (Delta BS)_p|S_i) Is shown in the body state of S_iUnder the condition of (1), Δ BS ═ Δ BS_pThe conditional probability of (a); p (Delta T)_q|S_i) Is shown in the body state of S_iUnder the condition of (1), Δ T ═ Δ T_qThe conditional probability of (1), 1,2, j is 1,2,3,4,5,6,7, m is 1,2,3,4,5,6,7, n is 1,2,3,4,5,6,7, p is 1,2,3,4,5,6,7, q is 1,2,3,4,5,6, 7;

and B: acquiring user sign data including heart rate H, pulse P, blood pressure BP, blood oxygen saturation BS and body temperature T in a current single time slice by using a sensor to form a sign data set D { H, P, BP, BS, T }; the physical sign data processing module receives a physical sign data set D { H, P, BP, BS, T } collected by a sensor and an internal initial human body normal physical sign information data set D'₀{H'₀,P’₀,BP’₀,BS'₀,T’₀Subtracting one from the other correspondingly to obtain a corresponding human body sign information fluctuation value set delta D, delta H, delta P, delta BP, delta BS, delta T;

Step D: the Bayesian network diagnosis module converts the voltage signal V into a voltage signal₂The voltage signal V is transmitted to a first early warning module₀、V₁And a normal signs data set D'_aThe voltage signal V is transmitted to the comprehensive diagnosis module, the first early warning module sends out a notice through a user side APP, and the comprehensive diagnosis module enables the voltage signal V to be transmitted to the comprehensive diagnosis module₀、V₁And a normal signs data set D'_aStoring;

the comprehensive diagnosis module stores the voltage signal V₀、V₁And a normal signs data set D'_aAnd then, the intelligent early warning system enters the next time slice, the steps B-D are repeated, and the comprehensive diagnosis module continuously stores the voltage signal V₀、V₁And a normal signs data set D'_aAll the voltage signals stored in the set continuous M time slices are V₁Or a voltage signal V₀And V₁Voltage signal V alternately appearing and in N successive time slices₁Appear for M times, and the ratio of M/N is less than or equal to 50 percent<1, judging the physical state of the user as mild disease S₁。

2. The intelligent early warning system for sudden illness in sleep state as claimed in claim 1, wherein: mild disease state S₁Corresponding voltage signal V₁And the second early warning module sends out a notice through the user side APP.

3. The intelligent early warning system for sudden illness in sleep state as claimed in claim 2, wherein: the comprehensive diagnosis module stores the voltage signal V₀And normal signs data set D'_aAnd outputting to an initial parameter updating module, wherein the initial parameter updating module calculates the average value of all normal sign data in the running process of the intelligent early warning system to obtain a new normal sign data set and uses the new normal sign data set as a next human normal sign data set D 'in the running process of the intelligent early warning system'₀。