A home rehabilitation management system for heart failure patients

Abstract

The present invention belongs to the technical field of medical rehabilitation, and discloses a home rehabilitation management system for heart failure patients; the present invention comprises a data acquisition and storage module, the data acquisition and storage module comprises a human-machine communication unit and a self-monitoring unit, and the heart failure data is obtained through the data obtained by the human-machine communication unit and the self-monitoring unit; a trend analysis module, which performs data analysis on the heart failure data by improving the ARIMA algorithm to obtain trend data; and performs trend analysis on the trend data by improving the Adaboost algorithm; the present invention can accurately capture the personal emotional data answered by the patient by extracting voice features, which is relatively accurate and avoids the patient from hiding his real personal emotions; the rehabilitation training process of the heart failure patient who is recovering at home can be monitored and evaluated in real time; and guidance on rehabilitation training, drug adjustment and diet adjustment can be recommended to the heart failure patient who is recovering at home.

Description

Technical Field

[0001] The present invention relates to the field of medical rehabilitation technology, and more specifically, to a home rehabilitation management system for heart failure patients. Background Art

[0002] Heart failure is a cardiovascular disease, usually caused by impaired contractile and / or diastolic function of the heart, which leads to abnormal blood flow, resulting in systemic blood congestion and insufficient blood supply to the arteries. Heart failure is not an independent disease, but the terminal stage of heart disease. Most heart failures start with left heart failure, that is, blood congestion in the lungs, leading to symptoms such as dyspnea, coughing, and sputum. There are many causes of heart failure, such as coronary heart disease, myocarditis, cardiomyopathy, and valvular heart disease. The prognosis of patients with heart failure is related to the cause, severity of the disease, and treatment plan. Generally speaking, the prognosis of heart failure is more serious, but timely and effective treatment and changes in lifestyle can improve the quality of life and survival rate of patients. Therefore, being able to track and evaluate the severity of heart failure faster, more accurately, and more conveniently is the most critical factor in adjusting treatment and prognosis measures and improving the quality of life.

[0003] However, heart failure patients face the following problems during the actual home rehabilitation process: it is inconvenient to provide medical supervision and timely communication with patients, there is a lack of guidance on diet, exercise and drug management in patients' home rehabilitation plans, and it is also inconvenient to understand patients' personal emotions in a timely manner.

[0004] In view of this, the present invention proposes a home rehabilitation management system for heart failure patients to solve the above problems. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art and to achieve the above-mentioned purpose, the present invention provides the following technical solution: a home rehabilitation management system for heart failure patients, comprising:

[0006] A data acquisition and storage module, wherein the data acquisition and storage module includes a human-machine communication unit and a self-monitoring unit, and the heart failure data is obtained through the data obtained by the human-machine communication unit and the self-monitoring unit;

[0007] The trend analysis module analyzes the heart failure data by improving the ARIMA algorithm to obtain trend data; and analyzes the trend data by improving the Adaboost algorithm to obtain the predicted value of the degree of heart failure;

[0008] The life status management module presets the training recommendation formula, recommends rehabilitation training to patients according to the training recommendation formula, and obtains the rehabilitation training evaluation results by monitoring the recommended rehabilitation training;

[0009] The drug management module adaptively adjusts the patient's drug prescription according to the predicted value of heart failure degree, self-perception data and rehabilitation training evaluation results to obtain a medication prescription;

[0010] The diet management module obtains diet recommendations based on the patient's predicted heart failure severity, rehabilitation training assessment results, current self-perception data and medication prescriptions.

[0011] Further, in the data acquisition and storage module, the human-computer communication unit is used to obtain the patient's personal emotional data and self-perception data, and the self-monitoring unit is used to obtain the patient's physiological data, exercise data and diet data;

[0012] The human-machine communication unit is used to obtain the patient's personal emotional data and self-perception data in the following manner:

[0013] The patient is asked questions through the voice of the electronic device and the patient's answer is obtained. The patient's answer is stored in voice and text form. The content of the questions includes: asking the patient about his or her self-feelings and emotional conditions;

[0014] Recognize personal emotions of patients’ answers stored in voice mode to obtain personal emotion data;

[0015] Keyword recognition is performed on the patient's answers stored in text form, and the keywords are matched with the self-perception keyword database.

[0016] Furthermore, the method of identifying personal emotions of the patient's answers stored in voice mode and obtaining personal emotion data is as follows:

[0017] The patient is asked to record speech in a specific emotional state as an initial speech, and the initial speech is preprocessed to obtain a speech sample. The specific emotional state includes happy emotion, depressed emotion, sad emotion and angry emotion;

[0018] Extracting speech features of speech samples, the speech features include tone features, pitch features, timbre features, speech pause time and word coherence;

[0019] The speech features of w groups of patients in historical time are collected; and emotional labels are marked, and the emotional labels correspond to different personal emotional values; the speech features marked with emotional labels constitute a speech feature data set, and a deep learning model is trained according to the speech feature data set to obtain a trained deep learning model; the trained deep learning model is used to perform emotion recognition on real-time speech features and output personal emotion data;

[0020] The method of extracting the speech features of the speech sample is:

[0021] Set a standard pitch, and use the average difference between the pitch in the current emotional state and the standard pitch as the pitch feature;

[0022] Set a standard pitch, and use the average difference between the pitch in the current emotional state and the standard pitch as the pitch feature;

[0023] Set a standard timbre, and take the average difference between the timbre in the current emotional state and the standard timbre as the timbre feature;

[0024] The speech sample is divided into short samples according to the start and end of the speech pause, and the time between two short samples is calculated to obtain the speech pause time;

[0025] A number of identical speech samples are collected to obtain an average speech pause time in the same speech samples, and word coherence is obtained according to a ratio of the speech pause time of the current speech sample to the average speech pause time.

[0026] Furthermore, the method of performing keyword recognition on the patient's answers stored in text form and matching the keywords with the self-perception keyword library is as follows:

[0027] A self-perception keyword database is set, and keywords in the self-perception keyword database are graded to obtain a self-perception grade, and the self-perception grade corresponds to a self-perception value;

[0028] Calculate the similarity between the text corresponding to the patient's answer and each keyword in the self-perception keyword database;

[0029] The calculated similarity of the keywords in the self-perception keyword database is processed in descending order, and the keyword with the highest similarity is used as the self-perception value corresponding to the patient's answer voice;

[0030] The process of calculating the similarity between the text corresponding to the patient's answer and each keyword in the self-perception keyword library is as follows:

[0031] The text corresponding to the patient's answer and the keywords in the self-perception keyword library are converted into binary strings, and the difference between the text corresponding to the patient's answer and the keywords in the self-perception keyword library after conversion into binary strings is calculated using a difference formula. The difference formula is:

[0032]

[0033] Among them, d(x`,y`) represents the length between the binary string of the patient's answer and the binary string of the keyword in the self-perception keyword library, z represents the number of bits of the binary string of the patient's answer, and x` k and y` kThey represent the characters at the kth position of the binary string of the patient's answer and the binary string of the keyword in the self-feeling keyword library, respectively. k ,y` k ) is 1 when the patient's answer voice binary string is not equal to the keyword binary string in the self-perception keyword library, and is 0 when they are equal.

[0034] Furthermore, the self-monitoring unit is used to obtain the patient's physiological data, exercise data and dietary data in the following manner:

[0035] Collecting the patient's physiological data through electronic devices, including electrocardiogram data, heart rate data, blood pressure data and impedance data;

[0036] Collect the patient's exercise data through any of smart watches, smartphone applications and exercise trackers, and the exercise data includes exercise pattern, exercise intensity and exercise time;

[0037] Dietary data were collected, including sodium intake, food intake, fluid intake, high-fiber food intake, and potassium- and magnesium-rich food intake.

[0038] Furthermore, the method of analyzing the heart failure data by improving the ARIMA algorithm to obtain trend data is as follows:

[0039] Personal emotion data, self-perception data, physiological data, exercise data and dietary data are taken as one heart failure data, and time series data of heart failure data corresponding to different heart failure values ​​in the historical period are collected. Initial weights are preset for personal emotion data, self-perception data, physiological data, exercise data and dietary data of heart failure data, and weighted sum is performed to make the result equal to the heart failure value. The personal emotion data, self-perception data, physiological data, exercise data and dietary data of time series data corresponding to heart failure data are analyzed in batches through the improved ARIMA model;

[0040] The linear relationship between the current value and the historical value of the personal emotional data, self-perception data, physiological data, exercise data and diet data of heart failure data is obtained in batches through the autoregressive formula. The autoregressive formula is:

[0041] Y t =c+φ1Y t-1 +φ2Y t-2 +…+φ p Y t-p +ε t ,

[0042] Among them, Y t is the current value, φ1 to φ p is the parameter of the model, c is a constant term, ε tis the error term at time t;

[0043] The personal emotion data, self-perception data, physiological data, exercise data and diet data of heart failure data are stabilized in batches through the differential operation formula. The differential operation formula is:

[0044] ΔY t =Y t -Y t-1 ,

[0045] Among them, ΔY t is the first-order difference value at time point t, Y t-1 Indicates the value at time point t-1;

[0046] The linear relationship between the current value and the prediction error of the historical value of the personal emotional data, self-perception data, physiological data, exercise data and diet data of heart failure data is obtained in batches through the moving average formula. The moving average formula is:

[0047] Y t =μ+θ1ε t-1 +θ2ε t-2 +…+θ q ε t-q +ε t ,

[0048] Among them, θ1 to θ q is the parameter of the model, μ is the mean term, ε t is the error term at the current moment;

[0049] The obtained personal emotion trend data, self-perception trend data, physiological trend data, exercise trend data and diet trend data are used as trend data.

[0050] Furthermore, the method of performing trend analysis on trend data by improving Adaboost algorithm to obtain the prediction value of heart failure degree is as follows:

[0051] The label corresponding to the trend data of the historical period is the heart failure value. The personal emotion trend data, self-feeling trend data, physiological trend data, exercise trend data and diet trend data are each taken as a type sample. The type sample includes s data points. All data points corresponding to each type sample are taken as the training data set.

[0052] Improved Adaboost algorithm formula Among them, f m (x i ) is the predicted value, b(x i ; Y m ) is the basis function, Y m is the parameter of the basis function, β mis the coefficient of the basis function, i is the index of the type sample, m is the number of iterations, x i is the data point of type i sample;

[0053] Define the loss function of the improved Adaboost algorithm

[0054] Using the training data set, iterative training is performed to obtain the weights corresponding to each type of sample;

[0055] The data acquisition and storage module collects the real-time data corresponding to the type sample in real time, inputs the real-time data of the type sample and the weights corresponding to each type sample into the weighted fusion algorithm, and outputs the predicted value of heart failure degree G(x i ):

[0056]

[0057] Among them, G(x i ) is the final weight value corresponding to the i-type sample, x i is the data point of type i sample, m1 is the number of weak classifiers, is the weight of the m1th weak classifier, is the prediction result of the m1th weak classifier for the i-type sample, It represents the weighted sum of all weak classifiers.

[0058] Furthermore, the preset training recommendation formula recommends rehabilitation training to the patient according to the training recommendation formula, and the method of obtaining the rehabilitation training evaluation result by monitoring the recommended rehabilitation training is as follows:

[0059] The training intensity is obtained through height, weight and age, and the target heart rate is obtained through heart rate trend and age. The target heart rate is the expected intensity index to be achieved by rehabilitation training through heart rate index. Then the training recommendation is obtained comprehensively. The training recommendation formula is:

[0060] G=[(Max hr -Rest hr )×0.7+Rest hr ]×α,

[0061] Among them, G is the target heart rate; Max hr is the maximum heart rate, Rest hr is the resting heart rate, and α is the training intensity.

[0062] Furthermore, the method of adaptively adjusting the patient's drug prescription according to the predicted value of the degree of heart failure, the self-perception data and the rehabilitation training evaluation results to obtain the medication prescription is:

[0063] When the patient has completed the examination at the hospital, he or she gets the examination results and a prescription for medication based on the examination results;

[0064] After the doctor gives the drug prescription, the drug prescription is entered and recorded in the personal information of the system;

[0065] The doctor adaptively adjusts the drug prescription based on the predicted value of the degree of heart failure, the current patient's self-perception data and the results of rehabilitation training evaluation to obtain a medication prescription.

[0066] Furthermore, the method of obtaining dietary recommendations based on the patient's predicted heart failure degree, rehabilitation training assessment results, current self-perception data and medication prescription is:

[0067] Obtain the patient's current predicted heart failure level, current rehabilitation training assessment results, current self-perception data, and current drug prescriptions;

[0068] Determine the diet level based on the current predicted value of heart failure severity, current rehabilitation training assessment results, current self-perception data and current drug prescriptions;

[0069] The dietary levels correspond to the heart failure levels, and each dietary level corresponds to the corresponding sodium intake, dietary level, high-fiber food intake and potassium-magnesium-rich food intake.

[0070] The technical effects and advantages of the home rehabilitation management system for heart failure patients of the present invention are as follows:

[0071] 1. By extracting voice features, the personal emotional data of the patient's voice answer can be accurately captured, which is relatively accurate, preventing the patient from hiding his true personal emotions, and facilitating the subsequent giving of corresponding feedback to the patient based on the patient's personal emotional data; by matching the patient's answer keywords, the patient's self-perception data can be obtained, which can facilitate accurate recognition of the patient's answer, avoiding the situation where the patient cannot answer accurately and cannot enter the self-perception data;

[0072] 2. In view of the current situation that there is no goal and scientific method for the home rehabilitation of heart failure patients, the system can conduct intelligent evaluation and prediction of the home rehabilitation of heart failure patients; recommend guidance on rehabilitation training, drug adjustment, and diet adjustment to heart failure patients undergoing home rehabilitation; conduct real-time monitoring and evaluation of the rehabilitation training process of heart failure patients undergoing home rehabilitation; and comprehensively manage heart failure patients undergoing home rehabilitation, adjust their living habits and health status, and maximize their quality of life. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 This is a schematic structural diagram of a home rehabilitation management system for heart failure patients according to the present invention;

[0074] Figure 2It is a structural schematic diagram of the self-monitoring module of the present invention;

[0075] Figure 3 It is a structural diagram of the life status module;

[0076] Figure 4 A flowchart for rehabilitation training;

[0077] Figure 5 It is a structural diagram of the drug management module;

[0078] Figure 6 Schematic diagram of the structure of the diet recommendation module. DETAILED DESCRIPTION

[0079] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0080] Example 1

[0081] See also Figure 1 to Figure 2 As shown, this embodiment provides a home rehabilitation management system for heart failure patients, including: a data acquisition and storage module, the data acquisition and storage module includes a human-computer interaction unit and a self-monitoring unit, and the heart failure data is obtained through the data obtained from the human-computer interaction unit and the self-monitoring unit.

[0082] In the data acquisition and storage module, the human-computer communication unit is used to obtain the patient's personal emotional data and self-feeling data, and the self-monitoring unit is used to obtain the patient's physiological data, exercise data, and diet data;

[0083] The human-machine communication unit is used to obtain the patient's personal emotional data and self-feeling data in the following ways:

[0084] The patient is asked questions through the voice of the electronic device and the patient's answer is obtained. The patient's answer is stored in voice and text form. The content of the questions includes: asking the patient about his or her self-feelings and emotional conditions;

[0085] Recognize personal emotions of patients’ answers stored in voice mode to obtain personal emotion data;

[0086] Perform keyword recognition on the patient's answers stored in text form and match the keywords with the self-perception keyword database;

[0087] The method of identifying personal emotions of patients’ answers stored in voice form and obtaining personal emotion data is as follows:

[0088] The patient is asked to record speech in a specific emotional state as the initial speech, and the initial speech is preprocessed to obtain a speech sample, the specific emotional state includes happy emotion, depressed emotion, sad emotion and angry emotion (the speech preprocessing includes denoising, segmentation and frame processing);

[0089] Extracting speech features of speech samples, the speech features include tone, pitch, timbre, speech pause time and word coherence;

[0090] The speech features of w groups of patients in historical time are collected; and emotional labels are marked, and the emotional labels correspond to different personal emotional values; the speech features marked with emotional labels constitute a speech feature data set, and a deep learning model is trained according to the speech feature data set to obtain a trained deep learning model; the trained deep learning model is used to perform emotion recognition on real-time speech features, and personal emotion data is output (different emotional states are assigned values, for example, happy emotions, depressed emotions, sad emotions, and angry emotions are assigned values ​​of 1, 2, 3, and 4 respectively).

[0091] The method of extracting speech features of speech samples is:

[0092] Set a standard pitch, and use the average difference between the pitch in the current emotional state and the standard pitch as the pitch feature;

[0093] Set a standard pitch, and use the average difference between the pitch in the current emotional state and the standard pitch as the pitch feature;

[0094] Set a standard timbre, and take the average difference between the timbre in the current emotional state and the standard timbre as the timbre feature;

[0095] The speech sample is divided into short samples according to the start and end of the speech pause, and the time between two short samples is calculated to obtain the speech pause time;

[0096] A number of identical speech samples are collected to obtain an average speech pause time in the same speech samples, and word coherence is obtained according to a ratio of the speech pause time of the current speech sample to the average speech pause time.

[0097] By extracting voice features, the personal emotional data of the patient's voice answering method can be accurately captured, which is relatively accurate and prevents the patient from hiding his true personal emotions, making it easier to give the patient corresponding feedback based on the patient's personal emotional data in the future;

[0098] It should be noted that feedback can also be given based on the patient's corresponding personal emotions. For example, soothing and calming music can be played when the patient is angry, happy music can be played when the patient is sad, and encouraging words can be given when the patient is depressed to reduce the patient's depression.

[0099] The method of performing keyword recognition on the patient's answers stored in text form and matching the keywords with the self-perception keyword database is as follows:

[0100] A self-perception keyword thesaurus is set, and keywords in the self-perception keyword thesaurus are graded to obtain a self-perception grade, and the self-perception grade corresponds to a self-perception value (different self-perception grades are assigned values, for example, 1, 2, 3, and 4 are assigned to them in order from low to high self-perception grades);

[0101] Calculate the similarity between the text corresponding to the patient's answer and each keyword in the self-perception keyword database;

[0102] The calculated similarity of the keywords in the self-perception keyword database is processed in descending order, and the keyword with the highest similarity is used as the self-perception value corresponding to the patient's answer voice;

[0103] The process of calculating the similarity between the text corresponding to the patient's answer and each keyword in the self-perception keyword database is as follows:

[0104] The text corresponding to the patient's answer and the keywords in the self-perception keyword library are converted into binary strings, and the difference between the text corresponding to the patient's answer and the keywords in the self-perception keyword library after conversion into binary strings is calculated using a difference formula. The difference formula is:

[0105]

[0106] Among them, d(x`,y`) represents the length between the binary string of the patient's answer and the binary string of the keyword in the self-perception keyword library, z represents the number of bits of the binary string of the patient's answer, and x` k and y` k They represent the characters at the kth position of the binary string of the patient's answer and the binary string of the keyword in the self-feeling keyword library, respectively. k ,y` k ) When the patient's answer voice binary string is not equal to the keyword binary string in the self-perception keyword library, it is 1, and when they are equal, it is 0. The longer the length, the lower the similarity, and the shorter the length, the higher the similarity.

[0107] By matching the key words in the patient's answer, the patient's self-perception data can be obtained, which can facilitate accurate identification of the patient's answer and avoid the situation where the self-perception data cannot be entered when the patient cannot answer accurately.

[0108] The self-monitoring unit is used to obtain the patient's physiological data, exercise data and dietary data in the following ways:

[0109] Collecting the patient's physiological data through electronic devices, including electrocardiogram data, heart rate data, blood pressure data and impedance data;

[0110] Specifically, the patient's ECG data is collected by sticking electrodes or wearing portable ECG monitoring equipment; the heart rate data is collected by a smart watch or smart phone, and the patient's blood pressure is collected by a sphygmomanometer; the patient's impedance data is collected by a bioimpedance analyzer or a multifunctional bio-monitoring device;

[0111] Collect the patient's exercise data through any of smart watches, smartphone applications and exercise trackers, and the exercise data includes exercise pattern, exercise intensity and exercise time;

[0112] Specifically, the exercise data is collected through any of smart watches, smartphone applications and sports trackers. Before the exercise starts, the exercise mode to be performed is selected or set in the collection device or application, such as running, cycling, swimming, etc. During the exercise, the relevant data of the exercise intensity, such as heart rate, speed, number of steps, etc., are recorded through the device or application to reflect the intensity and effect of the exercise, and the start and end time of the exercise are recorded, or the device or application is used to automatically record the exercise time. During the exercise, the device or application may provide real-time monitoring and feedback, such as heart rate monitoring, GPS positioning, exercise trajectory, etc., to help users understand the exercise situation in real time. After the exercise, ensure that the collected exercise data is saved in the device or application, and synchronize the data;

[0113] Dietary data were collected, including sodium intake, food intake, fluid intake, high-fiber food intake, and potassium- and magnesium-rich food intake;

[0114] The heart failure grade is calculated based on the exercise data. The formula for calculating the heart failure grade is:

[0115]

[0116] Among them, V represents the quantitative value of heart failure, M j represents the jth motion mode, α j represents the motion coefficient corresponding to the jth motion mode, I j represents the exercise intensity corresponding to the jth exercise mode, t is the exercise time, and the exercise intensity is the percentage of the heart rate during exercise to the maximum heart rate; (for example, low-intensity exercise can be defined as 50-60% of the maximum heart rate, medium-intensity exercise is 60-70%, and high-intensity exercise is more than 70%).

[0117] V is a quantitative value ranging from 0 to 100. If 0≤V≤25, it is grade I heart failure; if 25<V≤50, it is grade II heart failure; if 50<V≤75, it is grade III heart failure; if 75<V≤100, it is grade IV heart failure.

[0118] In this embodiment, the personal emotional data of the patient's voice answering method can be accurately captured by extracting voice features, which is relatively accurate and avoids the patient hiding his true personal emotions, making it easier to give the patient corresponding feedback based on his personal emotional data later; by matching the patient's answer keywords, the patient's self-perception data can be obtained, which can facilitate accurate identification of the patient's answer and avoid the situation where the patient cannot answer accurately and the self-perception data cannot be entered.

[0119] Example 2

[0120] See also Figure 1 to Figure 2 As shown, this embodiment provides a home rehabilitation management system for heart failure patients, including: a trend analysis module, which performs data analysis on heart failure data by improving the ARIMA algorithm to obtain trend data; performs trend analysis on the trend data by improving the Adaboost algorithm to obtain a predicted value of the degree of heart failure;

[0121] By analyzing the heart failure data through the improved ARIMA algorithm, the trend data is obtained as follows:

[0122] Personal emotion data, self-perception data, physiological data, exercise data and dietary data are taken as one heart failure data, and time series data of heart failure data corresponding to different heart failure values ​​in the historical period are collected. Initial weights are preset for personal emotion data, self-perception data, physiological data, exercise data and dietary data of heart failure data, and weighted sum is performed to make the result equal to the heart failure value. The personal emotion data, self-perception data, physiological data, exercise data and dietary data of time series data corresponding to heart failure data are analyzed in batches through the improved ARIMA model;

[0123] It should be noted that the specific values ​​corresponding to the heart failure values ​​can be obtained through ejection fraction, BNP measurement or physician assessment;

[0124] The linear relationship between the current value and the historical value of the personal emotional data, self-perception data, physiological data, exercise data and diet data of heart failure data is obtained in batches through the autoregressive formula. The autoregressive formula is:

[0125] Y t =c+φ1Y t-1 +φ2Y t-2 +…+φ p Y t-p +ε t ,

[0126] Among them, Y t is the current value, φ1 to φ p is the parameter of the model, c is a constant term, ε t is the error term at time t;

[0127] The personal emotion data, self-perception data, physiological data, exercise data and diet data of heart failure data are stabilized in batches through the differential operation formula. The differential operation formula is:

[0128] ΔY t =Y t -Y t-1 ,

[0129] Among them, ΔY t is the first-order difference value at time point t, Y t-1 Indicates the value at time point t-1;

[0130] The linear relationship between the current value and the prediction error of the historical value of the personal emotional data, self-perception data, physiological data, exercise data and diet data of heart failure data is obtained in batches through the moving average formula. The moving average formula is:

[0131] Y t =μ+θ1ε t-1 +θ2ε t-2 +…+θ q ε t-q +ε t ,

[0132] Among them, θ1 to θ q is the parameter of the model, μ is the mean term, ε t is the error term at the current moment;

[0133] The obtained personal emotion trend data, self-perception trend data, physiological trend data, exercise trend data and diet trend data are used as trend data.

[0134] By improving the Adaboost algorithm to analyze the trend data, the method to obtain the predicted value of heart failure degree is as follows:

[0135] The label corresponding to the trend data of the historical period is the heart failure value. The personal emotion trend data, self-feeling trend data, physiological trend data, exercise trend data and diet trend data are each taken as a type sample. The type sample includes s data points. All data points corresponding to each type sample are taken as the training data set.

[0136] Improved Adaboost algorithm formula Among them, f m (x i) is the predicted value, b(x i ; Y m ) is the basis function, Y m is the parameter of the basis function, β m is the coefficient of the basis function, i is the index of the type sample, m is the number of iterations, x i is the data point of type i sample;

[0137] Define the loss function of the improved Adaboost algorithm

[0138] Using the training data set, iterative training is performed to obtain the weights corresponding to each type of sample;

[0139] The data acquisition and storage module collects the real-time data corresponding to the type sample in real time, inputs the real-time data of the type sample and the weights corresponding to each type sample into the weighted fusion algorithm, and outputs the predicted value of heart failure degree G(x i ):

[0140]

[0141] Among them, G(x i ) is the final weight value corresponding to the i-type sample, x i is the data point of type i sample, m1 is the number of weak classifiers, is the weight of the m1th weak classifier, is the prediction result of the m1th weak classifier for the i-type sample, Represents the weighted sum of all weak classifiers:

[0142] In this embodiment, by improving the ARIMA algorithm, the trend of heart failure data can be better analyzed, and more accurate information about the patient's health status can be provided. By analyzing the trend data through the improved Adaboost algorithm, a more accurate prediction value of the degree of heart failure can be obtained; by combining multiple weak classifiers to construct a strong classifier, the accuracy of the prediction value of the degree of heart failure can be improved; by monitoring and predicting the patient's predicted value of the degree of heart failure, personalized health management suggestions and treatment plans can be provided to the patient, thereby improving the quality of life of heart failure patients and reducing the risk of complications.

[0143] Example 3

[0144] See also Figures 1 to 6 As shown, this embodiment provides a home rehabilitation management system for heart failure patients, including:

[0145] The life status management module presets the training recommendation formula, recommends rehabilitation training to patients according to the training recommendation formula, and obtains the rehabilitation training evaluation results by monitoring the recommended rehabilitation training;

[0146] The drug management module adaptively adjusts the patient's drug prescription according to the predicted value of heart failure degree, self-perception data and rehabilitation training evaluation results to obtain a medication prescription;

[0147] The diet management module obtains diet recommendations based on the patient's predicted heart failure severity, rehabilitation training assessment results, current self-perception data and medication prescriptions.

[0148] A training recommendation formula is preset, and rehabilitation training is recommended to the patient according to the training recommendation formula. The method of obtaining the rehabilitation training evaluation result by monitoring the recommended rehabilitation training is as follows:

[0149] The training intensity is obtained through height, weight and age, and the target heart rate is obtained through heart rate trend and age. The target heart rate is the expected intensity index to be achieved by rehabilitation training through heart rate index. Then the training recommendation is obtained comprehensively. The training recommendation formula is:

[0150] G=[(Max hr -Rest hr )×0.7+Rest hr ]×α,

[0151] Where: G is the target heart rate; Max hr is the maximum heart rate; the maximum heart rate is calculated by the maximum target heart rate and the actual maximum heart rate. The maximum target heart rate is the difference between 220 and age, and the actual maximum heart rate is the maximum heart rate during walking; Rest hr is the resting heart rate, α is the training intensity, and the training intensity can be set according to the heart failure prediction value or heart failure grade;

[0152] Monitor the recommended rehabilitation training, obtain monitoring results, evaluate the monitoring results, and obtain rehabilitation training evaluation results;

[0153] Get the maximum heart rate of the monitoring result, calculate the heart rate deviation value between the maximum heart rate of the monitoring result and the maximum heart rate corresponding to the training recommended formula, get the rehabilitation training evaluation result according to the heart rate deviation value, set the heart rate deviation value interval, the heart rate deviation value interval corresponds to the rehabilitation training evaluation result, the smaller the heart rate deviation value, the higher the output value of the rehabilitation training evaluation level, for example, the heart rate deviation value is 0, if the heart rate deviation value interval is [1, 10], then the rehabilitation training evaluation result is 10, for example, the heart rate deviation value is 10, if the heart rate deviation value interval is [1, 10], then the rehabilitation training evaluation result is 1, for example, the heart rate deviation value is 30, if the heart rate deviation value interval is [1, 10], then the rehabilitation training evaluation result is 3;

[0154] It should be noted that after the rehabilitation training is completed, the training effect, relevant parameters during the training process, and multi-parameter comprehensive evaluation of the training effect are used, and then the trend changes of the training effects are integrated to formulate the next rehabilitation training recommendation; when the training effect is poor, the training index is reduced, and the first reduction of the training index is mainly to lower the target heart rate; when the training effect is improved compared with the previous training effect, the target heart rate is upgraded to a normal index; comprehensively evaluate the effects of more than two rehabilitation trainings (including this one), if the training effect shows an upward trend, the training index is advanced, and the advanced ratio is 10%;

[0155] Physical training process: first wear heart rate and data acquisition equipment to collect and analyze training data in real time during rehabilitation training; open the rehabilitation training module, the system will automatically connect to the device and automatically import recommended training data; click to start training and enter the rehabilitation training stage. During this process, relevant data acquisition equipment will collect training data in real time and conduct real-time analysis until the end of training; after the training, the rehabilitation training will be evaluated and the rehabilitation training evaluation results will be given.

[0156] The patient's drug prescription is adaptively adjusted according to the predicted value of heart failure degree, self-perception data and rehabilitation training evaluation results, and the medication prescription is obtained as follows:

[0157] When the patient has completed the examination at the hospital, he or she gets the examination results and a prescription for medication based on the examination results;

[0158] After the doctor gives the drug prescription, the drug prescription is entered and recorded in the personal information of the system;

[0159] The doctor makes adaptive adjustments to the drug prescription based on the predicted value of the degree of heart failure, the current patient's self-perception data and the results of rehabilitation training assessment to obtain a medication prescription;

[0160] Specifically, a regression model is used to establish a relationship between drug prescription and the predicted value of heart failure severity, the current patient's self-perception data and rehabilitation training evaluation results, and the drug prescription is adjusted based on the relationship.

[0161] Based on the patient's predicted heart failure level, rehabilitation training assessment results, current self-perception data, and medication prescriptions, dietary recommendations are as follows:

[0162] Obtain the patient's current predicted heart failure level, current rehabilitation training assessment results, current self-perception data, and current drug prescriptions;

[0163] Determine the diet level based on the current predicted value of heart failure severity, current rehabilitation training assessment results, current self-perception data and current drug prescriptions;

[0164] The dietary level corresponds to the heart failure level, and each dietary level corresponds to the corresponding sodium intake, dietary level, high-fiber food intake, and potassium- and magnesium-rich food intake;

[0165] Specifically, the adjustment method of heart failure grade from low to high is: reduction of sodium salt intake, reduction of fluid intake, increase of high-fiber food intake and increase of potassium and magnesium-rich food intake. The above reductions and increases are all in the form of percentages.

[0166] Each dietary level corresponds to a dietary value threshold range, which is used to assign a value to the current drug prescription. A regression model is used to establish a relationship between the dietary value and the current predicted value of heart failure severity, the current rehabilitation training assessment results, the current self-perception data, and the current drug prescription. Dietary recommendations are given to patients based on the dietary level corresponding to the dietary value obtained from the relationship.

[0167] In this embodiment, rehabilitation training is recommended to patients through a preset training recommendation formula combined with monitoring results, and the training effect is evaluated based on the deviation between the monitoring results and the expected value, which can help patients to conduct rehabilitation training more effectively and adjust the training plan according to the evaluation results; in combination with the predicted value of the degree of heart failure, self-perception data and rehabilitation training evaluation results, the patient's drug prescription is adaptively adjusted, and the drug adjustment based on individual conditions can better control the symptoms of heart failure and improve the treatment effect; by combining the predicted value of the degree of heart failure, rehabilitation training evaluation results, self-perception data and medication prescriptions, a personalized diet management plan is output, which can help patients better control their diet, thereby maintaining the patient's physical health and treatment effect.

[0168] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0169] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technical users in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A home rehabilitation management system for heart failure patients, characterized in that: include, A data acquisition and storage module, wherein the data acquisition and storage module includes a human-machine communication unit and a self-monitoring unit, and the heart failure data is obtained through the data obtained by the human-machine communication unit and the self-monitoring unit; The trend analysis module analyzes the heart failure data by improving the ARIMA algorithm to obtain trend data; The trend data was analyzed by improving the Adaboost algorithm to obtain the predicted value of heart failure degree; The life status management module presets the training recommendation formula, recommends rehabilitation training to patients according to the training recommendation formula, and obtains the rehabilitation training evaluation results by monitoring the recommended rehabilitation training; The drug management module adaptively adjusts the patient's drug prescription according to the predicted value of heart failure degree, self-perception data and rehabilitation training evaluation results to obtain a medication prescription; The diet management module provides diet recommendations based on the patient's predicted heart failure severity, rehabilitation training assessment results, current self-perception data, and medication prescriptions; In the data acquisition and storage module, the human-computer communication unit is used to obtain the patient's personal emotional data and self-perception data, and the self-monitoring unit is used to obtain the patient's physiological data, exercise data and diet data; The human-machine communication unit is used to obtain the patient's personal emotional data and self-perception data in the following manner: The patient is asked questions through the voice of the electronic device and the patient's answer is obtained. The patient's answer is stored in voice and text form. The content of the questions includes: asking the patient about his or her self-feelings and emotional conditions; Recognize personal emotions of patients’ answers stored in voice mode to obtain personal emotion data; Perform keyword recognition on the patient's answers stored in text form and match the keywords with the self-perception keyword database; The method of performing keyword recognition on the patient's answers stored in text form and matching the keywords with the self-perception keyword database is as follows: A self-perception keyword database is set, and keywords in the self-perception keyword database are graded to obtain a self-perception grade, wherein the self-perception grade corresponds to the self-perception data; Calculate the similarity between the text corresponding to the patient's answer and each keyword in the self-perception keyword database; The calculated similarity of the keywords in the self-perception keyword database is processed in descending order, and the keyword with the highest similarity is used as the self-perception data corresponding to the patient's answer voice; The process of calculating the similarity between the text corresponding to the patient's answer and each keyword in the self-perception keyword library is as follows: The text corresponding to the patient's answer and the keywords in the self-perception keyword library are converted into binary strings, and the difference between the text corresponding to the patient's answer and the keywords in the self-perception keyword library after conversion into binary strings is calculated using a difference formula. The difference formula is: Among them, d(x`,y`) represents the length between the binary string of the patient's answer and the binary string of the keyword in the self-perception keyword library, z represents the number of bits of the binary string of the patient's answer, and x` k and y` k They represent the characters at the kth position of the binary string of the patient's answer and the binary string of the keyword in the self-feeling keyword library, respectively. k ,y` k ) is 1 when the patient's answering voice binary string is not equal to the keyword binary string in the self-perception keyword library, and is 0 when they are equal; The preset training recommendation formula is used to recommend rehabilitation training to the patient according to the training recommendation formula, and the method of obtaining the rehabilitation training evaluation result by monitoring the recommended rehabilitation training is as follows: The training intensity is obtained through height, weight and age, and the target heart rate is obtained through heart rate trend and age. The target heart rate is the expected intensity index to be achieved by rehabilitation training through heart rate index. Then the training recommendation is obtained comprehensively. The training recommendation formula is: G=[(Max hr -Rest hr )×0.7+Rest hr ]×α, Among them, G is the target heart rate; Max hr is the maximum heart rate, Rest hr is the resting heart rate, α is the training intensity; The recommended rehabilitation training is monitored to obtain monitoring results, and the monitoring results are evaluated to obtain rehabilitation training evaluation results.

2. A home rehabilitation management system for heart failure patients according to claim 1, characterized in that: The method of identifying personal emotions of the patient's answers stored in voice mode and obtaining personal emotion data is as follows: The patient is asked to record speech in a specific emotional state as an initial speech, and the initial speech is preprocessed to obtain a speech sample. The specific emotional state includes happy emotion, depressed emotion, sad emotion and angry emotion; Extracting speech features of speech samples, the speech features include tone features, pitch features, timbre features, speech pause time and word coherence; Collect the speech characteristics of group w patients in historical time; The emotion labels are marked, and the emotion labels correspond to different personal emotion values; the speech features marked with emotion labels constitute a speech feature data set, and a deep learning model is trained according to the speech feature data set to obtain a trained deep learning model; the trained deep learning model is used to perform emotion recognition on real-time speech features and output personal emotion data; The method of extracting the speech features of the speech sample is: Set a standard pitch, and use the average difference between the pitch in the current emotional state and the standard pitch as the pitch feature; Set a standard pitch, and use the average difference between the pitch in the current emotional state and the standard pitch as the pitch feature; Set a standard timbre, and take the average difference between the timbre in the current emotional state and the standard timbre as the timbre feature; The speech sample is divided into short samples according to the start and end of the speech pause, and the time between two short samples is calculated to obtain the speech pause time; A number of identical speech samples are collected to obtain an average speech pause time in the same speech samples, and word coherence is obtained according to a ratio of the speech pause time of the current speech sample to the average speech pause time.

3. A home rehabilitation management system for heart failure patients according to claim 2, characterized in that: The self-monitoring unit is used to obtain the patient's physiological data, exercise data and dietary data in the following manner: Collecting the patient's physiological data through electronic equipment, including electrocardiogram data, heart rate data, blood pressure data and impedance data; Collect the patient's exercise data through any of smart watches, smartphone applications and exercise trackers, and the exercise data includes exercise pattern, exercise intensity and exercise time; Dietary data were collected, including sodium intake, food intake, fluid intake, high-fiber food intake, and potassium- and magnesium-rich food intake.

4. A home rehabilitation management system for heart failure patients according to claim 3, characterized in that: The method of analyzing the heart failure data by improving the ARIMA algorithm to obtain trend data is as follows: Personal emotion data, self-perception data, physiological data, exercise data and dietary data are taken as one heart failure data, and time series data of heart failure data corresponding to different heart failure values ​​in the historical period are collected. Initial weights are preset for personal emotion data, self-perception data, physiological data, exercise data and dietary data of heart failure data, and weighted sum is performed to make the result equal to the heart failure value. The personal emotion data, self-perception data, physiological data, exercise data and dietary data of time series data corresponding to heart failure data are analyzed in batches through the improved ARIMA model; The linear relationship between the current value and the historical value of the personal emotional data, self-perception data, physiological data, exercise data and diet data of heart failure data is obtained in batches through the autoregressive formula. The autoregressive formula is: Y t =c+φ1Y t-1 +φ2Y t-2 +···+φ p Y t-p +e t , Among them, Y t is the current value, φ1 to φ p is the parameter of the model, c is a constant term, ε t is the error term at time t; The personal emotion data, self-perception data, physiological data, exercise data and diet data of heart failure data are stabilized in batches through the differential operation formula. The differential operation formula is: ΔY t =Y t -Y t-1 , Among them, ΔY t is the first-order difference value at time point t, Y t-1 Indicates the value at time point t-1; The linear relationship between the current value and the prediction error of the historical value of the personal emotional data, self-perception data, physiological data, exercise data and diet data of heart failure data is obtained in batches through the moving average formula. The moving average formula is: Y t =μ+θ1ε t-1 +θ2ε t-2 +···+θ q e t-q +e t , Among them, θ1 to θ q is the parameter of the model, μ is the mean term, ε t is the error term at the current moment; The obtained personal emotion trend data, self-perception trend data, physiological trend data, exercise trend data and diet trend data are used as trend data.

5. A home rehabilitation management system for heart failure patients according to claim 4, characterized in that: The method of performing trend analysis on trend data by improving Adaboost algorithm to obtain the predicted value of heart failure degree is as follows: The label corresponding to the trend data of the historical period is the heart failure value. The personal emotion trend data, self-feeling trend data, physiological trend data, exercise trend data and diet trend data are each taken as a type sample. The type sample includes s data points. All data points corresponding to each type sample are taken as the training data set. Improved Adaboost algorithm formula Among them, f m (x i ) is the predicted value, b(x i ; Y m ) is the basis function, Y m is the parameter of the basis function, β m is the coefficient of the basis function, i is the index of the type sample, m is the number of iterations, x i is the data point of type i sample; Define the loss function of the improved Adaboost algorithm Using the training data set, iterative training is performed to obtain the weights corresponding to each type of sample; The data acquisition and storage module collects the real-time data corresponding to the type sample in real time, inputs the real-time data of the type sample and the weights corresponding to each type sample into the weighted fusion algorithm, and outputs the predicted value of heart failure degree G(x i ): Among them, G(x i ) is the final weight value corresponding to the i-type sample, x i is the data point of type i sample, m1 is the number of weak classifiers, is the weight of the m1th weak classifier, is the prediction result of the m1th weak classifier for the i-type sample, It represents the weighted summation of all weak classifiers.

6. A home rehabilitation management system for heart failure patients according to claim 5, characterized in that: The method of adaptively adjusting the patient's drug prescription according to the predicted value of the degree of heart failure, the self-perception data and the rehabilitation training evaluation results to obtain the medication prescription is: When the patient has completed the examination at the hospital, he or she gets the examination results and a prescription for medication based on the examination results; After the doctor gives the drug prescription, the drug prescription is entered and recorded in the personal information of the system; The doctor adaptively adjusts the drug prescription based on the predicted value of the degree of heart failure, the current patient's self-perception data and the results of rehabilitation training evaluation to obtain a medication prescription.

7. A home rehabilitation management system for heart failure patients according to claim 1, characterized in that: The method of obtaining dietary recommendations based on the patient's predicted heart failure degree, rehabilitation training assessment results, current self-perception data and medication prescriptions is as follows: Obtain the patient's current predicted heart failure level, current rehabilitation training assessment results, current self-perception data, and current drug prescriptions; Determine the diet level based on the current predicted value of heart failure severity, current rehabilitation training assessment results, current self-perception data and current drug prescriptions; The dietary levels correspond to the heart failure levels, and each dietary level corresponds to the corresponding sodium intake, dietary level, high-fiber food intake and potassium-magnesium-rich food intake.

Images