A Method and System for Assisting in Decision-making of Personalized Rehabilitation Programs

Abstract

The present application provides a method and system for assisting in decision-making of personalized rehabilitation programs, which relates to the field of intelligent medical technology and includes: receiving wearable user status monitoring information; obtaining a first functional disability status assessment result; obtaining a second functional disability status assessment result; obtaining recommended sleep posture information; obtaining a recommended rehabilitation exercise program; and sending the recommended sleep posture information and the recommended rehabilitation exercise program to the management terminal of elderly patients with comorbid AKI. Through the present application, the technical problem in the prior art that due to the failure to fully consider the gradually changing health status of patients during the rehabilitation process, the multiple health problems of patients cannot be accurately and personalized considered, which may lead to insufficient pertinence of the rehabilitation program can be solved. By personalizing the adjustment of sleep and exercise programs, more scientific and personalized rehabilitation guidance is provided for users with different health states, achieving the technical effect of improving the applicability of the rehabilitation program to users.

Description

Technical Field

[0001] This application relates to the field of intelligent medical technology, and in particular, to a method and system for assisting in making decisions on personalized rehabilitation programs. Background Art

[0002] With the advent of an aging society, the elderly population faces challenges of multiple chronic diseases and comorbidities. Among them, acute kidney injury (AKI), as a common clinical symptom, has become one of the main causes of death and disability worldwide. AKI not only seriously affects the renal function of patients, but also often accompanies the decline of other organ functions, leading to multiple system dysfunction. Therefore, the rehabilitation and care of AKI patients, especially elderly patients with comorbid AKI, need to comprehensively consider the functional status of each organ and formulate personalized and precise rehabilitation programs.

[0003] With the development of medical technology, intelligent rehabilitation assistance decision-making systems have emerged. By comprehensively considering the vital signs, disease conditions, and physiological function data of patients, they assist medical staff in formulating personalized rehabilitation programs, thereby improving the rehabilitation effect and the quality of life of patients. Traditional rehabilitation programs are usually static and do not fully consider the gradually changing health conditions of patients during the rehabilitation process. For elderly patients with comorbid AKI, due to the existence of multiple complications such as heart disease and respiratory problems, traditional programs cannot accurately and personalizedly consider the multiple health problems of patients, which may lead to insufficient pertinence of the rehabilitation program. Summary of the Invention

[0004] The purpose of this application is to provide a method and system for assisting in making decisions on personalized rehabilitation programs to solve the technical problem in the prior art that due to the failure to fully consider the gradually changing health conditions of patients during the rehabilitation process, the multiple health problems of patients cannot be accurately and personalizedly considered, which may lead to insufficient pertinence of the rehabilitation program.

[0005] In view of the above problems, this application provides a method and system for assisting in making decisions on personalized rehabilitation programs.

[0006] In a first aspect, the present application provides a method for assisting in making decisions on personalized rehabilitation programs. The method for assisting in making decisions on personalized rehabilitation programs is implemented through a system for assisting in making decisions on personalized rehabilitation programs. Among them, the method for assisting in making decisions on personalized rehabilitation programs includes: receiving wearable user status monitoring information from a wearable monitoring device, where the wearable user status monitoring information includes daily activity status monitoring information and rehabilitation exercise status monitoring information; analyzing the functional disability status based on the sleep status monitoring information, breathing status monitoring information, and swallowing status monitoring information of the daily activity status monitoring information to obtain a first functional disability status assessment result; collecting the heart, lungs, brain, and kidney status monitoring information of the rehabilitation exercise status monitoring information for functional disability status analysis to obtain a second functional disability status assessment result; optimizing the sleep posture according to the first functional disability status assessment result to obtain recommended sleep posture information; optimizing the exercise program according to the second functional disability status assessment result to obtain a recommended rehabilitation exercise program; and sending the recommended sleep posture information and the recommended rehabilitation exercise program to the management terminal of elderly patients with comorbid AKI.

[0007] In a second aspect, the present application further provides a system for assisting in making decisions on personalized rehabilitation programs, which is used to execute the method for assisting in making decisions on personalized rehabilitation programs as described in the first aspect. Among them, the system for assisting in making decisions on personalized rehabilitation programs includes: a user status monitoring module, configured to receive wearable user status monitoring information from a wearable monitoring device, where the wearable user status monitoring information includes daily activity status monitoring information and rehabilitation exercise status monitoring information; a first functional disability analysis module, configured to analyze the functional disability status based on the sleep status monitoring information, breathing status monitoring information, and swallowing status monitoring information of the daily activity status monitoring information to obtain a first functional disability status assessment result; a second functional disability analysis module, configured to collect the heart, lungs, brain, and kidney status monitoring information of the rehabilitation exercise status monitoring information for functional disability status analysis to obtain a second functional disability status assessment result; a sleep posture optimization module, configured to optimize the sleep posture according to the first functional disability status assessment result to obtain recommended sleep posture information; an exercise program optimization module, configured to optimize the exercise program according to the second functional disability status assessment result to obtain a recommended rehabilitation exercise program; and a program sending module, configured to send the recommended sleep posture information and the recommended rehabilitation exercise program to the management terminal of elderly patients with comorbid AKI.

[0008] One or more technical solutions provided in the present application have at least the following technical effects or advantages:

[0009] Receive wearable user status monitoring information, where the wearable user status monitoring information includes daily activity status monitoring information and rehabilitation exercise status monitoring information; perform functional disability status analysis based on the sleep status monitoring information, breathing status monitoring information, and swallowing status monitoring information in the daily activity status monitoring information to obtain a first functional disability status assessment result; collect the heart, lung, brain, and kidney status monitoring information in the rehabilitation exercise status monitoring information for functional disability status analysis to obtain a second functional disability status assessment result; optimize the sleep posture according to the first functional disability status assessment result to obtain recommended sleep posture information; optimize the exercise plan according to the second functional disability status assessment result to obtain a recommended rehabilitation exercise plan; send the recommended sleep posture information and the recommended rehabilitation exercise plan to the management terminal for elderly patients with comorbid AKI. Through real-time collection and analysis of multi-source data of users, perform functional disability status assessment, and personalize the adjustment of sleep and exercise plans, providing more scientific and personalized rehabilitation guidance for users in different health states, achieving the technical effect of improving the applicability of the rehabilitation plan to users.

[0010] The above description is only an overview of the technical solution of this application. In order to be able to understand the technical means of this application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features, and advantages of this application more obvious and understandable, the following specifically gives the specific implementation manners of this application. It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of this application, nor is it used to limit the scope of this application. Other features of this application will become easily understood through the following description. Brief Description of the Drawings

[0011] In order to more clearly illustrate the technical solutions in this application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings described below are only exemplary, and for those of ordinary skill in the art, without creative efforts, other drawings can also be obtained according to the provided drawings.

[0012] Figure 1 It is a schematic flowchart of a method for assisting in making a personalized rehabilitation plan of this application.

[0013] Figure 2 It is a schematic structural diagram of a system for assisting in making a personalized rehabilitation plan of this application.

[0014] Description of the reference numerals: User status monitoring module 11, First functional disability analysis module 12, Second functional disability analysis module 13, Sleep posture optimization module 14, Exercise plan optimization module 15, Plan sending module 16. Detailed Description of the Preferred Embodiments

[0015] By providing a method and system for assisting in making decisions on personalized rehabilitation programs, this application solves the technical problem in the prior art that due to the failure to fully consider the gradually changing health status of patients during the rehabilitation process, it is impossible to accurately and personalized consider the multiple health problems of patients, which may lead to insufficient pertinence of the rehabilitation program. Through real-time collection and analysis of multi-source data of users, the functional disability status is evaluated, the sleep and exercise programs are adjusted personalized, and more scientific and personalized rehabilitation guidance is provided for users in different health states, achieving the technical effect of improving the applicability of the rehabilitation program to users.

[0016] Next, the technical solutions in this application will be described clearly and completely with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments of this application. It should be understood that this application is not limited by the example embodiments described here. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of this application. Additionally, it should be noted that for the sake of description, only the parts related to this application are shown in the accompanying drawings rather than all of them.

[0017] Example 1. Please refer to the attached Figure 1 , this application provides a method for assisting in making decisions on personalized rehabilitation programs. Among them, the method for assisting in making decisions on personalized rehabilitation programs is applied to a system for assisting in making decisions on personalized rehabilitation programs. The method for assisting in making decisions on personalized rehabilitation programs specifically includes the following steps:

[0018] Step 1: Receive the wearable user status monitoring information from the wearable monitoring device. Among them, the wearable user status monitoring information includes daily activity status monitoring information and rehabilitation exercise status monitoring information.

[0019] Specifically, through the communication connection with the wearable monitoring device, the wearable user status monitoring information can be received. Among them, the wearable monitoring device usually includes various sensors, such as vibration sensors, accelerometers, gyroscopes, heart rate sensors, etc., which can collect and transmit the physiological data of users in real time. In this application, the vibration status of the user is collected through the vibration sensor of the wearable monitoring device, and then the vibration characteristics are analyzed to determine the daily activity status monitoring information or rehabilitation exercise status monitoring information of the user, and the wearable user status monitoring information is generated with the daily activity status monitoring information or rehabilitation exercise status monitoring information.

[0020] Generally speaking, the monitoring parameters corresponding to the daily activity status monitoring information and the rehabilitation exercise status monitoring information are different. The monitoring parameters corresponding to the daily activity status monitoring information are the activity data of the user in daily life, including sleep status, breathing status, swallowing status, etc., which can reflect the user's activity level and whether there are living disabilities. The rehabilitation exercise status monitoring information is the exercise performance of the user during the rehabilitation process, including the cardiac, pulmonary, cerebral, and renal status parameters during rehabilitation training. By comprehensively analyzing the daily activity status and the rehabilitation exercise status monitoring information, the activity ability and rehabilitation progress of the user can be comprehensively understood, providing data support for the subsequent optimization of the sleep and rehabilitation programs.

[0021] Step 2: Analyze the functional disability status based on the sleep status monitoring information, breathing status monitoring information, and swallowing status monitoring information of the daily activity status monitoring information to obtain the first functional disability status assessment result.

[0022] Specifically, the sleep status monitoring information is usually obtained through devices such as smart bracelets and smart watches in wearable monitoring devices. The sleep status monitoring information is obtained by monitoring the user's sleep quality, sleep cycle, and the proportion of deep sleep and light sleep. The breathing status monitoring information is usually collected through a smart chest strap or a breathing monitoring device in the wearable monitoring device, recording data such as the user's breathing frequency, breathing depth, and oxygen saturation to form the breathing status monitoring information. The swallowing status monitoring information is a key piece of data for evaluating whether the patient has swallowing difficulties (such as swallowing disorders or swallowing function decline). It is usually collected through a dedicated swallowing monitoring device in the wearable monitoring device. The swallowing status monitoring information can include swallowing frequency, swallowing duration, and muscle activity during swallowing (electromyogram data), etc.

[0023] By synthesizing the sleep status monitoring information, breathing status monitoring information, and swallowing status monitoring information, analyze the functional disability status of the user by establishing a respiratory disability status assessment model, and infer the type of functional disability of the user as the first functional disability status assessment result. For example, assume that the first functional disability status assessment result of an elderly patient is moderate respiratory dysfunction accompanied by mild swallowing disorders. This means that the user's respiratory function is damaged to a certain extent. Through this multi-dimensional data analysis, not only can the functional status of the patient be accurately evaluated, but also detailed basis can be provided for personalized sleep programs.

[0024] Step 3: Collect the cardiac, pulmonary, cerebral, and renal status monitoring information of the rehabilitation exercise status monitoring information to analyze the functional disability status and obtain the second functional disability status assessment result.

[0025] Specifically, the cardiopulmonary encephalorenal status monitoring information of the rehabilitation exercise status monitoring information includes the monitoring data corresponding to the heart, lungs, brain, and kidneys, specifically including data related to heart health such as heart rate, electrocardiogram, blood pressure, etc., parameters related to lung health such as vital capacity, blood oxygen saturation, respiratory rate, etc., parameters related to brain health such as electroencephalogram, nerve reflex test data, etc., and parameters related to kidney health such as urine information, kidney ultrasound data, creatinine and blood urea nitrogen data in the blood. The same method as that used to obtain the first functional disability status assessment result is adopted to analyze the cardiopulmonary encephalorenal status monitoring information by training an organ disability status assessment model, synthesize the health status data of each organ, and evaluate the organ disability type as the second functional disability status assessment result, so as to provide a basis for subsequent rehabilitation treatment and help achieve a more accurate and personalized rehabilitation plan.

[0026] Step Four: Optimize the sleep posture according to the first functional disability status assessment result to obtain recommended sleep posture information.

[0027] Specifically, when the first functional disability status assessment result is the user's respiratory tract disability type, the core goal of sleep posture optimization is to improve the patient's rest quality and reduce possible functional disabilities through scientific sleep posture adjustment. For example, for patients with respiratory problems (such as sleep apnea, shortness of breath, etc.), a side-lying or semi-reclined posture is recommended to reduce airway obstruction and promote smooth breathing. For patients with poor swallowing function, it is recommended to appropriately elevate the head to avoid the risk of aspiration or asphyxiation. Specifically, by analyzing the user's sleep posture records in the big data set, the effective sleep postures adopted by users with similar health conditions can be found as the recommended sleep posture information, which can help users obtain higher-quality sleep and thus prevent the deterioration of respiratory tract disabilities.

[0028] Step Five: Optimize the exercise plan according to the second functional disability status assessment result to obtain a recommended rehabilitation exercise plan.

[0029] Specifically, the result of the second functional disability status assessment is the type of organ disability of the user. The core goal of the exercise program is to formulate a more targeted rehabilitation exercise plan based on the health conditions of the patient's important organs such as the heart, lungs, brain, and kidneys, avoid health risks caused by excessive exercise or improper exercise, and maximize the promotion of the rehabilitation effect. According to the result of the second functional disability status assessment, the type, intensity, and duration of exercise are reasonably selected to help the patient improve or maintain cardiopulmonary function, promote blood circulation, enhance muscle strength, and improve the overall health condition. When optimizing the exercise program, according to the result of the second functional disability assessment and the user's medical record information, a dataset of the rehabilitation exercise program records of the same-cluster users can be collected, that is, a set of rehabilitation exercise programs of other users with the same medical record and the same type of organ functional disability is obtained. Further, the trigger frequency of any rehabilitation exercise program in the dataset of the rehabilitation exercise program records of the same-cluster users is statistically analyzed, that is, the proportion of any rehabilitation exercise program in the dataset of the rehabilitation exercise program records of the same-cluster users. The rehabilitation exercise program with a trigger frequency greater than or equal to the trigger frequency threshold is selected as the recommended rehabilitation exercise program to provide scientific and personalized exercise guidance for the user, which can not only help the patient recover physical function but also reduce the risks during exercise and promote their comprehensive rehabilitation.

[0030] Step Six: Send the recommended sleep posture information and the recommended rehabilitation exercise program to the management terminal for elderly patients with comorbid AKI.

[0031] Specifically, the management terminal for elderly patients with comorbid AKI, as an interaction interface between the user and the rehabilitation system, can receive, store, and display rehabilitation suggestions, and at the same time provide functional support for the user to execute and give feedback. Sending the recommended sleep posture information and the recommended rehabilitation exercise program to the management terminal for elderly patients with comorbid AKI requires the aid of modern medical information systems and communication technologies. Before sending, the recommended information is formatted, for example, generating a structured data file (such as JSON or XML format), and using encryption technologies (such as AES encryption) to protect data privacy. The data is sent to the management terminal for elderly patients with comorbid AKI through existing secure communication protocols (such as HTTPS or MQTT protocol). The integrity and security of the data are ensured during the transmission process. The management terminal for elderly patients with comorbid AKI intuitively displays the recommended sleep posture information and the recommended rehabilitation exercise program in a vivid and graphic form. For example, the sleep suggestions may be presented in the form of a schematic diagram of the human posture, and the exercise program is displayed in the form of a daily schedule or step-by-step guidance. The patient can obtain scientific and personalized rehabilitation guidance in a timely manner, and at the same time, the operability of the program is greatly improved.

[0032] Furthermore, Step One of this application includes:

[0033] Collect user vibration state monitoring information through the vibration sensor of the wearable monitoring device; extract time-domain features from the user vibration state monitoring information to obtain user vibration time-domain features; perform Fourier transform on the user vibration state monitoring information to obtain user vibration frequency-domain features; classify the activity type according to the user vibration frequency-domain features and the user vibration time-domain features to obtain an activity type calibration result; wherein, the activity type calibration result is one of the daily activity state monitoring information and the rehabilitation exercise state monitoring information; classify and collect the wearable user state monitoring information according to the activity type calibration result.

[0034] Specifically, the vibration sensor provides valuable physiological data by sensing the minute vibration changes during the user's activities or movements, which can reflect the exercise intensity, frequency, and amplitude of the user in different states. The vibration state monitoring information, including data such as vibration amplitude and frequency, can be directly collected through the vibration sensor. Time-domain feature extraction refers to extracting basic information such as amplitude, frequency, and waveform from the vibration signal as the user vibration time-domain features. For example, if the vibration frequency of a certain user shows a certain periodic pattern during activities, its frequency feature will be recorded, which may be expressed as the number of vibrations per second. Then, perform Fourier transform on the user vibration state monitoring information to obtain the user vibration frequency-domain features. Fourier transform is an existing mathematical tool that can convert the time-domain signal into a frequency-domain signal and reveal the frequency components of the signal. In the frequency domain of the vibration signal, wave peaks of different frequencies can often be seen, which are used as the user vibration frequency-domain features. For example, the low-frequency region may correspond to the stationary state or slow activities, while the high-frequency region represents fast movement or high-intensity activities. This frequency-domain feature can effectively reflect the intensity and pattern of the movement and provide a basis for subsequent activity type classification.

[0035] By combining the user vibration frequency-domain features and the user vibration time-domain features, the activity type of the user can be classified. The process of activity type classification usually involves machine learning algorithms, such as using methods like support vector machine (SVM), decision tree, or neural network, to compare and train the extracted features with the pre-labeled activity type data. Through training, it can be determined whether the user's current activity type belongs to the daily activity state (such as walking, standing, sitting, sleeping, etc.) or the rehabilitation exercise state (such as gait training, balance exercise, strength training, etc.). Finally, according to the activity type calibration result, the wearable state monitoring information of the user is classified and collected, ensuring that different types of exercise states can be accurately monitored and corresponding interventions or suggestions can be made as needed.

[0036] Furthermore, the present application further includes the following steps:

[0037] Collect user vibration state record data and activity type identification data; extract time-domain features from the user vibration state record data to obtain user vibration state time-domain record features; perform Fourier transform on the user vibration state record data to obtain user vibration state frequency-domain record features; use the user vibration state time-domain record features and the user vibration state frequency-domain record features as inputs and the activity type identification data as supervision to train an activity type classifier.

[0038] Specifically, the user vibration state record data and the activity type identification data are sample data obtained by professional technicians in the field through testing. By testing, vibration state data corresponding to different activity types of different users are obtained and data annotation is performed. The vibration state record data records the vibration intensity and frequency changes at each moment. Based on the corresponding user's activity type, data annotation is performed to obtain the corresponding activity type identification data. Further, time-domain features are extracted from the user vibration state record data. The time-domain features include, but are not limited to, statistics such as the mean, variance, maximum value, minimum value, and peak value of the signal, which can reveal the basic trend and fluctuation of the vibration signal, and obtain user vibration state time-domain record features. Next, Fourier transform is performed on the user vibration state record data to convert it into frequency-domain features, revealing the different frequency components and their intensities in the signal, and obtaining user vibration state frequency-domain record features.

[0039] Use the user vibration state time-domain record features and the user vibration state frequency-domain record features as inputs and perform supervised learning using the activity type identification data to train an activity type classifier. The activity type classifier can be constructed based on existing machine learning models, such as neural networks, support vector machines, etc. The goal is to learn the relationship between the time-domain and frequency-domain features extracted from the vibration data and the calibrated activity type, so as to accurately predict the user's activity type in new uncalibrated data. For example, 5000 pieces of user vibration record data were used in the experiment, and these data were paired with the calibrated activity types. Through multiple iterative trainings, a classification model that can accurately distinguish activities such as walking, sitting, and running was finally constructed. The accuracy rate during the training process gradually increased with the optimization of the training and finally reached an activity classification accuracy rate of 95%.

[0040] Finally, input the obtained user vibration frequency-domain features and user vibration time-domain features into the activity type classifier for analysis, and output the activity type calibration result to provide more accurate data support for the user's personalized rehabilitation plan.

[0041] Further, step two of this application includes:

[0042] Input the sleep state monitoring information, the respiratory state monitoring information, and the swallowing state monitoring information into a respiratory disability condition assessment model for analysis to obtain the type of respiratory disability, designated as the first functional disability condition assessment result.

[0043] Specifically, the comprehensive analysis of sleep state monitoring information, respiratory state monitoring information, and swallowing state monitoring information is crucial for evaluating the respiratory disability condition of a patient. These data can reflect the respiratory function and swallowing ability of the patient during daily activities and rehabilitation. After integrating these monitoring information, they are input into a respiratory disability condition assessment model for further analysis. The respiratory disability condition assessment model is usually trained through various algorithms. By combining the sleep, respiratory, and swallowing information of the user, it determines the type of respiratory disability, designated as the first functional disability condition assessment result. For example, machine learning techniques may be used to learn the clinical characteristics of different patients through a large amount of training data to judge the type of respiratory disability. According to the assessment result, a personalized rehabilitation plan can be further provided, such as improving the sleep posture, to improve the user's respiratory function and reduce the impact of respiratory disability on daily life.

[0044] Furthermore, the present application further includes the following steps:

[0045] Construct a multi-classification loss function and an ensemble loss function; train several respiratory disability condition assessment sub-classifiers according to the multi-classification loss function; train the fully connected weights of the several respiratory disability condition assessment sub-classifiers according to the ensemble loss function; perform a full connection on the several respiratory disability condition assessment sub-classifiers according to the fully connected weights to obtain the respiratory disability condition assessment model.

[0046] Specifically, the multi-classification loss function includes various existing loss functions, such as the cross-entropy loss function, the Hinge loss function (for support vector machines), the adversarial loss function, etc. Specifically, those skilled in the art can select various existing loss functions by themselves. It is prior art and will not be elaborated here one by one. The various loss functions can be the loss functions used in the training of different machine learning models. Further, several respiratory disability condition assessment sub-classifiers can be trained according to the multi-classification loss function, that is, one loss function is used to train one respiratory disability condition assessment sub-classifier. The functions of the several respiratory disability condition assessment sub-classifiers are the same, which is to analyze the corresponding type of respiratory disability based on the user's sleep, respiratory, and swallowing information, but there may be differences in accuracy.

[0047] Under the framework of the multi-classification loss function, several sub-classifiers for evaluating respiratory disability conditions are trained. Specifically, those skilled in the art can obtain a large number of sample data of sleep state monitoring information, respiratory state monitoring information, swallowing state monitoring information, and respiratory disability type annotation data with corresponding relationships based on the prior art. These sample data can be grouped to obtain multiple sets of training samples. Based on the multi-classification loss function and existing machine learning models, multiple sets of training samples are used for training and testing respectively to obtain several sub-classifiers for evaluating respiratory disability conditions that are trained to convergence.

[0048] Next, the stability and accuracy of the prediction are improved by connecting several sub-classifiers for evaluating respiratory disability conditions through an integrated loss function. The integrated loss function usually weights the losses of several sub-classifiers for evaluating respiratory disability conditions, that is, multiple sets of connection weights can be randomly generated. Each set of connection weights includes the weights corresponding to several sub-classifiers for evaluating respiratory disability conditions respectively, and the sum of the weights of several sub-classifiers for evaluating respiratory disability conditions is 1. That is, based on the integrated loss function, the losses of several sub-classifiers for evaluating respiratory disability conditions are comprehensively calculated according to their respective weights to obtain a set of connection weights with the minimum integrated loss as the final full connection weights of several sub-classifiers for evaluating respiratory disability conditions, so as to ensure the performance of the overall model. The full connection weights refer to the weighted relationship when several sub-classifiers for evaluating respiratory disability conditions output, which determines how the final prediction result is integrated from the outputs of each sub-classifier.

[0049] Full connection is a common layer structure in neural networks. The outputs of several sub-classifiers for evaluating respiratory disability conditions are connected together through weights to form a unified prediction model, that is, the respiratory disability condition evaluation model, which can comprehensively judge the respiratory disability type of patients according to the input multi-source data (such as sleep state, respiratory state, swallowing state, etc.), providing model support for the evaluation of functional disability conditions.

[0050] Furthermore, the present application further includes the following steps:

[0051] The integrated loss function is: ;

[0052] ;

[0053] Wherein, represents the first integrated loss of a certain set of connection weights, represents the second integrated loss of a certain set of connection weights, represents the loss of the i-th classifier, represents the sub-classifier loss threshold, Characterize the weighted mean weight of the i-th classifier; based on the first integrated loss threshold and the second integrated loss threshold, combined with the integrated loss function, train the fully connected weights of the several respiratory disability status assessment sub-classifiers.

[0054] Specifically, in the above integrated loss function, Characterize the loss of the i-th classifier, that is, the final loss value obtained after training the i-th classifier to convergence through the multi-classification loss function and the corresponding training samples. Characterize the sub-classifier loss threshold, that is, a preset loss threshold, which can be understood as the expected loss threshold in an ideal situation, and is specifically set by professionals in this field. Characterize the weighted mean weight of the i-th classifier, that is, the weight of the i-th classifier in the randomly generated multiple sets of connection weights, and is also the target that needs to be trained and optimized through the integrated loss function. Is the standard deviation symbol, , where N is the total number of several respiratory disability status assessment sub-classifiers.

[0055] The first integrated loss threshold and the second integrated loss threshold are the expected first integrated loss and the expected second integrated loss set by professionals in this field, such as 0.01. Randomly generate multiple sets of connection weights, calculate the multiple sets of connection weights through the integrated loss function, and obtain a set of connection weights that meet the first integrated loss threshold and the second integrated loss threshold as the fully connected weights of the several respiratory disability status assessment sub-classifiers.

[0056] Specifically, the integrated loss function can also be used as the fitness function, and meeting the first integrated loss threshold and the second integrated loss threshold is used as the iteration stop condition. Through existing optimization algorithms, such as the particle swarm optimization algorithm, the genetic algorithm, etc., perform iterative optimization of the link weights, so as to obtain the fully connected weights that meet the first integrated loss threshold and the second integrated loss threshold. The optimization algorithm is a commonly used technical means for those skilled in the art and will not be elaborated here.

[0057] Similarly, the method for obtaining the first functional disability status assessment result can be adopted to construct an organ disability status assessment model, analyze the monitoring information of the heart, lungs, brain, and kidneys, and evaluate the organ disability type as the second functional disability status assessment result. Specifically, a multi-classification loss function and an ensemble loss function are also used to obtain the monitoring sample data of the heart, lungs, brain, and kidneys and the corresponding organ disability types as sample data. The sample data is divided, and a multi-classification loss function is used to train several sub-classifiers according to the division result. Then, the fully connected weights of several sub-classifiers are trained using the ensemble loss function, and several sub-classifiers are fully connected to obtain an organ disability status assessment model. Furthermore, the monitoring information of the heart, lungs, brain, and kidneys is analyzed to evaluate the organ disability type, that is, the second functional disability status assessment result.

[0058] Further, step four of this application includes:

[0059] Based on the first functional disability status assessment result and combined with the user's medical record information, collect the sleep posture record data set of users in the same cluster; extract the trigger frequency set of the sleep posture record data set of users in the same cluster; select the sleep postures with the trigger frequency set greater than or equal to the trigger frequency threshold and add them to the recommended sleep posture information.

[0060] Specifically, first collect the user's medical record information, for example, the user's past medical history, chronic diseases, surgical records, etc. to form the user's medical record information. According to the first functional disability assessment result and the user's medical record information, based on the existing big data technology, collect the sleep postures of other users with the same first functional disability assessment result and user's medical record information as this user to form a sleep posture record data set of users in the same cluster. The users in the sleep posture record data set of users in the same cluster have similar respiratory disability type problems and similar medical record information, making the analysis more accurate. The next step is to deeply analyze the sleep posture record data set of users in the same cluster, extract the trigger frequencies of various sleep postures, and form a trigger frequency set. The trigger frequency refers to the proportion of a certain specific sleep posture in the user group of the same cluster. For example, there are 1000 users' sleep postures in the sleep posture record data set of users in the same cluster, and among them, the side-lying posture appears 800 times. Then the corresponding trigger frequency is 0.8, which reflects the effectiveness of this posture in patients with a certain specific health problem. The trigger frequency threshold is set by those skilled in the art, such as 0.7. Select the sleep postures with the trigger frequency greater than or equal to the trigger frequency threshold and add them to the recommended sleep posture information. Through this data-driven method, the recommendation of sleep postures not only has data support but also can be precisely optimized according to the actual needs of patients, maximizing the sleep adjustment effect.

[0061] Similarly, the same method can be adopted to collect the rehabilitation exercise plan record dataset of users in the same cluster according to the second functional disability assessment result and the user's medical record information, further count the trigger frequency of any rehabilitation exercise plan in the rehabilitation exercise plan record dataset of users in the same cluster, and select the rehabilitation exercise plan with a trigger frequency greater than or equal to the trigger frequency threshold as the recommended rehabilitation exercise plan to provide users with scientific and personalized exercise guidance.

[0062] In summary, the personalized rehabilitation plan assisted decision-making method provided by this application has the following technical effects:

[0063] Receiving the wearable user status monitoring information from the wearable monitoring device, where the wearable user status monitoring information includes daily activity status monitoring information and rehabilitation exercise status monitoring information; analyzing the functional disability status based on the sleep status monitoring information, breathing status monitoring information, and swallowing status monitoring information of the daily activity status monitoring information to obtain the first functional disability status assessment result; collecting the heart, lung, brain, and kidney status monitoring information of the rehabilitation exercise status monitoring information to analyze the functional disability status and obtain the second functional disability status assessment result; optimizing the sleep posture according to the first functional disability status assessment result to obtain the recommended sleep posture information; optimizing the exercise plan according to the second functional disability status assessment result to obtain the recommended rehabilitation exercise plan; sending the recommended sleep posture information and the recommended rehabilitation exercise plan to the management terminal for elderly patients with comorbid AKI. Through real-time collection and analysis of multi-source data of users, the functional disability status is evaluated, and the sleep and exercise plans are adjusted personalized to provide more scientific and personalized rehabilitation guidance for users in different health states, achieving the technical effect of improving the applicability of the rehabilitation plan to users.

[0064] Embodiment 2, based on the same inventive concept as the personalized rehabilitation plan assisted decision-making method in the foregoing embodiment, this application also provides a personalized rehabilitation plan assisted decision-making system. Please refer to the appendix Figure 2 The personalized rehabilitation plan assisted decision-making system includes:

[0065] A user status monitoring module 11, configured to receive wearable user status monitoring information from a wearable monitoring device, where the wearable user status monitoring information includes daily activity status monitoring information and rehabilitation exercise status monitoring information.

[0066] A first functional disability analysis module 12, configured to analyze the functional disability status based on the sleep status monitoring information, breathing status monitoring information, and swallowing status monitoring information of the daily activity status monitoring information to obtain the first functional disability status assessment result.

[0067] The second functional disability analysis module 13 is used to analyze the functional disability status of the cardiopulmonary, cerebral, and renal status monitoring information collected for the rehabilitation exercise status monitoring information, and obtain the second functional disability status evaluation result.

[0068] The sleep posture optimization module 14 is used to optimize the sleep posture according to the first functional disability status evaluation result, and obtain the recommended sleep posture information.

[0069] The exercise plan optimization module 15 is used to optimize the exercise plan according to the second functional disability status evaluation result, and obtain the recommended rehabilitation exercise plan.

[0070] The plan sending module 16 is used to send the recommended sleep posture information and the recommended rehabilitation exercise plan to the management terminal of the elderly patients with comorbid AKI.

[0071] Furthermore, the user status monitoring module 11 in the personalized rehabilitation plan auxiliary decision-making system is further used for:

[0072] Collect the user vibration status monitoring information through the vibration sensor of the wearable monitoring device; extract the time-domain characteristics of the user vibration status monitoring information to obtain the user vibration time-domain characteristics; perform Fourier transform on the user vibration status monitoring information to obtain the user vibration frequency-domain characteristics; classify the activity types according to the user vibration frequency-domain characteristics and the user vibration time-domain characteristics to obtain the activity type calibration result; wherein, the activity type calibration result is one of the daily activity status monitoring information and the rehabilitation exercise status monitoring information; classify and collect the wearable user status monitoring information according to the activity type calibration result.

[0073] Furthermore, the user status monitoring module 11 in the personalized rehabilitation plan auxiliary decision-making system is further used for:

[0074] Collect the user vibration status record data and the activity type identification data; extract the time-domain characteristics of the user vibration status record data to obtain the user vibration status time-domain record characteristics; perform Fourier transform on the user vibration status record data to obtain the user vibration status frequency-domain record characteristics; use the user vibration status time-domain record characteristics and the user vibration status frequency-domain record characteristics as inputs, and use the activity type identification data as supervision to train the activity type classifier.

[0075] Furthermore, the first functional disability analysis module 12 in the personalized rehabilitation plan auxiliary decision-making system is further used for:

[0076] Input the sleep state monitoring information, the respiratory state monitoring information, and the swallowing state monitoring information into a respiratory disability condition assessment model for analysis to obtain the type of respiratory disability, which is set as the first functional disability condition assessment result.

[0077] Furthermore, the first functional disability analysis module 12 in the personalized rehabilitation plan assistance decision-making system is further configured to:

[0078] Construct a multi-classification loss function and an integrated loss function; train a number of respiratory disability condition assessment sub-classifiers according to the multi-classification loss function; train the fully connected weights of the number of respiratory disability condition assessment sub-classifiers according to the integrated loss function; perform full connection on the number of respiratory disability condition assessment sub-classifiers according to the fully connected weights to obtain the respiratory disability condition assessment model.

[0079] Furthermore, the first functional disability analysis module 12 in the personalized rehabilitation plan assistance decision-making system is further configured to:

[0080] The integrated loss function is: ;

[0081] ;

[0082] Wherein, represents the first integrated loss at a certain set of connection weights, represents the second integrated loss of a certain set of connection weights, represents the loss of the i-th classifier, represents the sub-classifier loss threshold, represents the weighted mean weight of the i-th classifier; based on the first integrated loss threshold and the second integrated loss threshold, combined with the integrated loss function, train the fully connected weights of the number of respiratory disability condition assessment sub-classifiers.

[0083] Furthermore, the sleep posture optimization module 14 in the personalized rehabilitation plan assistance decision-making system is further configured to:

[0084] Based on the first functional disability condition assessment result, combined with the user's medical record information, collect the sleep posture record data set of the same cluster of users; extract the trigger frequency set of the sleep posture record data set of the same cluster of users; select the sleep postures with the trigger frequency set greater than or equal to the trigger frequency threshold and add them to the recommended sleep posture information.

[0085] Each embodiment in this specification is described in a progressive manner, and the key point of each embodiment is the difference from other embodiments. The foregoing Figure 1The personalized rehabilitation plan assisted decision-making method and specific examples in Embodiment 1 are equally applicable to the personalized rehabilitation plan assisted decision-making system in this embodiment. Through the above detailed description of the personalized rehabilitation plan assisted decision-making method, those skilled in the art can clearly know the personalized rehabilitation plan assisted decision-making system in this embodiment. Therefore, for the sake of brevity of the specification, it will not be elaborated here. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0086] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0087] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is also intended to include these changes and modifications.

Claims

1. A method for assisting in making decisions on personalized rehabilitation programs, characterized in that Applied to a personalized rehabilitation plan assisted decision-making system, the personalized rehabilitation plan assisted decision-making system is communicatively connected to a wearable monitoring device, and includes: Receiving wearable user status monitoring information from the wearable monitoring device, where the wearable user status monitoring information includes daily activity status monitoring information and rehabilitation exercise status monitoring information; Performing functional disability status analysis based on the sleep status monitoring information, breathing status monitoring information, and swallowing status monitoring information of the daily activity status monitoring information to obtain a first functional disability status assessment result; Collecting the heart, lung, brain, and kidney status monitoring information of the rehabilitation exercise status monitoring information to perform functional disability status analysis and obtain a second functional disability status assessment result; Optimizing the sleep posture according to the first functional disability status assessment result to obtain recommended sleep posture information; Optimizing the exercise plan according to the second functional disability status assessment result to obtain a recommended rehabilitation exercise plan; Sending the recommended sleep posture information and the recommended rehabilitation exercise plan to the management terminal of elderly patients with comorbid AKI; Receiving wearable user status monitoring information from the wearable monitoring device, including: Collecting user vibration status monitoring information through the vibration sensor of the wearable monitoring device; Performing time-domain feature extraction on the user vibration status monitoring information to obtain user vibration time-domain features; Performing Fourier transform on the user vibration status monitoring information to obtain user vibration frequency-domain features; Performing activity type classification based on the user vibration frequency-domain features and the user vibration time-domain features to obtain an activity type calibration result; Wherein the activity type calibration result is one of the daily activity status monitoring information and the rehabilitation exercise status monitoring information; Performing classified collection of the wearable user status monitoring information according to the activity type calibration result; Optimizing the sleep posture according to the first functional disability status assessment result to obtain recommended sleep posture information, including: Based on the first functional disability status assessment result, combining with the user's medical record information, collecting the sleep posture record data set of the same cluster of users; Extracting the trigger frequency set of the sleep posture record data set of the same cluster of users; Selecting the sleep postures in the trigger frequency set that are greater than or equal to the trigger frequency threshold and adding them to the recommended sleep posture information; Performing activity type classification based on the user vibration frequency-domain features and the user vibration time-domain features to obtain an activity type calibration result, including: Collecting user vibration status record data and activity type identification data; Performing time-domain feature extraction on the user vibration status record data to obtain user vibration status time-domain record features; Performing Fourier transform on the user vibration status record data to obtain user vibration status frequency-domain record features; Using the user vibration status time-domain record features and the user vibration status frequency-domain record features as inputs and the activity type identification data as supervision to train an activity type classifier.

2. The personalized rehabilitation plan assisted decision-making method according to claim 1, wherein Performing functional disability status analysis based on the sleep status monitoring information, breathing status monitoring information, and swallowing status monitoring information of the daily activity status monitoring information to obtain a first functional disability status assessment result, including: Input the sleep state monitoring information, the respiratory state monitoring information, and the swallowing state monitoring information into a respiratory disability status assessment model for analysis to obtain the type of respiratory disability, which is set as the first functional disability status assessment result.

3. The personalized rehabilitation plan assisted decision-making method according to claim 2, wherein Input the sleep state monitoring information, the respiratory state monitoring information, and the swallowing state monitoring information into a respiratory disability status assessment model for analysis to obtain the type of respiratory disability, which previously included: Construct a multi-classification loss function and an ensemble loss function; Train a number of respiratory disability status assessment sub-classifiers according to the multi-classification loss function; Train the fully connected weights of the number of respiratory disability status assessment sub-classifiers according to the ensemble loss function; Fully connect the number of respiratory disability status assessment sub-classifiers according to the fully connected weights to obtain the respiratory disability status assessment model.

4. The personalized rehabilitation plan assisted decision-making method according to claim 3, wherein The ensemble loss function is: ； ； Among them, represents the first integrated loss of a certain set of connection weights, represents the second integrated loss of a certain set of connection weights, represents the loss of the i-th classifier, represents the sub-classifier loss threshold, represents the weighted mean weight of the i-th classifier.

5. A personalized rehabilitation plan assisted decision-making system, characterized in that, For implementing the steps of the personalized rehabilitation plan assistance decision-making method according to any one of claims 1 to 4, the personalized rehabilitation plan assistance decision-making system includes: A user state monitoring module for receiving wearable user state monitoring information from a wearable monitoring device, where the wearable user state monitoring information includes daily activity state monitoring information and rehabilitation exercise state monitoring information; A first functional disability analysis module for performing functional disability status analysis based on the sleep state monitoring information, the respiratory state monitoring information, and the swallowing state monitoring information of the daily activity state monitoring information to obtain a first functional disability status assessment result; A second functional disability analysis module for performing functional disability status analysis on the cardiac, pulmonary, cerebral, and renal state monitoring information of the rehabilitation exercise state monitoring information to obtain a second functional disability status assessment result; A sleep posture optimization module for optimizing the sleep posture according to the first functional disability status assessment result to obtain recommended sleep posture information; An exercise plan optimization module for optimizing the exercise plan according to the second functional disability status assessment result to obtain a recommended rehabilitation exercise plan; A plan sending module for sending the recommended sleep posture information and the recommended rehabilitation exercise plan to the management terminal of elderly patients with comorbid AKI.

Images