Elderly health situation monitoring and early warning system and device based on digital twinborn

Abstract

The invention relates to the field of intelligent health situation monitoring, in particular to an elderly health situation monitoring and early warning system and device based on digital twinning, and the system comprises a data collection module, a digital twinning module, a risk prediction module and an early warning module. Wherein the data acquisition module acquires physiological and environmental data in real time; the digital twin module constructs an environment and individual model which dynamically reflects the physical state of the old people, predicts the environment risk and adjusts the state of the model; the risk prediction module evaluates health risks according to the vital sign data and a twinborn model; and the early warning module customizes a health management scheme or sends early warning to nursing personnel according to the evaluation result. The system has the advantages of real-time performance and accuracy, realization of personalized health management, effective prediction of environmental risks, timely early warning and presentation of technical advantages. Meanwhile, physiological and environmental factors are comprehensively considered, comprehensive health management support is provided for old people, and health monitoring and risk early warning efficiency is remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent health situation monitoring, and in particular to a system and device for monitoring and early warning the health situation of the elderly based on digital twins. Background Art

[0002] As the global population aging trend becomes more and more obvious, the health problems faced by the elderly are becoming more and more prominent. Health challenges such as chronic disease management, fall risk prevention and control, motor function decline and cognitive decline have become important factors that plague the quality of life of the elderly. Traditional health management methods are difficult to provide accurate, timely and personalized intervention measures in the face of these complex and diverse health problems, and cannot effectively meet the actual needs of the elderly.

[0003] In recent years, sensor-based health monitoring technology has gradually emerged, which has improved this situation to a certain extent. With the help of the integration of various sensors, this type of technology realizes the real-time collection of human health data, providing more immediate data support for health management. However, the mainstream health monitoring systems on the market still have many limitations: First, most systems only focus on the monitoring of a single health indicator, such as heart rate, blood pressure or daily steps, and over-rely on wearable devices, making it difficult to fully and deeply show the complexity and diversity of the physical condition of the elderly, and unable to form a complete health view; second, the existing systems generally lack comprehensive management of the health data and environmental information of the elderly, resulting in data in isolation, ignoring the important impact of environmental factors on the health of the elderly. For example, slippery floors increase the risk of falls, and insufficient indoor light may affect cognitive function, but these environmental information has not been effectively combined with health data, greatly limiting the accuracy and comprehensiveness of health monitoring.

[0004] In addition, the existing health management system also lacks personalized health management solutions. It is difficult to formulate targeted intervention strategies based on the specific health conditions and individual needs of the elderly. At the same time, the long-term tracking and analysis of the health conditions of the elderly is insufficient, and the trend of health changes cannot be effectively identified, making it difficult to optimize the health management plan according to the actual health evolution of the elderly.

[0005] Therefore, there is an urgent need for further exploration and innovation to develop a more comprehensive, accurate, real-time and personalized health management system to effectively improve the health management level of the elderly and provide stronger guarantees for their quality of life. Summary of the invention

[0006] In order to solve the above technical problems, the present invention provides a system and device for monitoring and early warning of elderly health status based on digital twins, the system comprising: The data collection module is used to collect monitoring data of the elderly in real time, including their vital signs data and the environmental data they are in; A digital twin module, for constructing a digital twin model and an environmental model that can respectively reflect the changes in the physical state of the elderly in real time based on the vital sign data and the environmental data, wherein the environmental model evaluates the potential risk results of the elderly in the current environment by predicting the environmental data, and adjusts the output state of the digital twin model based on the potential risk results; A risk prediction module, used to obtain health risk assessment results for the elderly based on the current output state of the digital twin model; And an early warning module is used to provide customized health management plans for the elderly or send risk warning information to caregivers based on the health risk assessment results.

[0007] In one embodiment of the present invention, based on the current output state of the digital twin model, a health risk assessment result of the elderly is obtained, including: Perform feature extraction on the vital sign data and the environmental data after preprocessing to obtain feature data, and dynamically assign corresponding weights to each monitoring data in the feature data; Based on the current output state of the digital twin model, the monitoring data is fused and calculated according to the weights to obtain state data that can represent the current physical condition of the elderly; Based on the status data, a health risk assessment result for the elderly is obtained.

[0008] In one embodiment of the present invention, the monitoring data is fused and calculated according to the weights to obtain status data that can represent the current physical condition of the elderly, including: Constructing a reinforcement learning model, and inputting the real-time monitoring data of the elderly, the historical performance data of the sensor that collects the monitoring data, and the simulation output results of the digital twin model as system state data into the reinforcement learning model; The monitoring data is processed using a neural network model, and the neural network model is expressed as , where X is the input monitoring data, F is the health status score output by the neural network model, are neural network model parameters; Based on the health status score, the evaluation value of the historical performance data and the impact evaluation value of the environmental data, a reward function R is constructed: , in, 、 、 They are all weight coefficients, H is the evaluation value of historical performance data, and P is the impact evaluation value of environmental data; When training the reinforcement learning model, the goal is to maximize the value of the reward function, perform iterative training, and continuously adjust the weight of each monitoring data to obtain a trained reinforcement learning model; Based on the trained reinforcement learning model, the current system state data is input to obtain the action with the maximum reward value, i.e., the weight adjustment strategy; According to the weight adjustment strategy, the monitoring data is weighted and fused to obtain status data that can represent the current physical condition of the elderly. .

[0009] In one embodiment of the present invention, iterative training is performed with the goal of maximizing the value of the reward function, and the weight of each monitoring data is continuously adjusted to obtain a trained reinforcement learning model, including: Define the state space S and the weight-adjusted action space A, where Indicates any time i The system status data obtained includes the monitoring data of the elderly, the historical performance data of the sensors that collect the monitoring data, and the simulation output results of the digital twin model. Indicates any time i The calculated weight adjustment strategy includes the weights assigned to the monitoring data; Initialize the parameters of the current Q network and use the current Q network to calculate the current state The Q-values ​​of all possible actions , and select actions according to the ε-greedy strategy , perform actions in the environment , the reward is calculated by the reward function and select the next state , according to the reward and the next state , using the target Q network to calculate the selected action The target Q value y; Use the target Q value y and the Q value predicted by the current Q network Calculate the loss value, and update the parameters of the current Q network by back-propagating by minimizing the loss value until the network converges or reaches the maximum iteration round, and obtain the trained reinforcement learning model; The current Q network is a neural network structure of multiple convolutional layers and multiple fully connected layers, and the structure of the target Q network is the same as that of the current Q network.

[0010] In one embodiment of the present invention, the expression of the target Q value y is as follows: , in, is the discount factor, According to the status , execute the action with the maximum Q value, represents the target Q network.

[0011] In one embodiment of the present invention, the monitoring data is weighted and fused according to the weight adjustment strategy to obtain status data that can represent the current physical condition of the elderly. ,as follows: , in, For sensor i The monitoring data after preprocessing at time t, n is the total number of sensors collecting vital signs data and environmental data; Expressed as The weight assigned.

[0012] In one embodiment of the present invention, the weight as follows: , in, is the historical performance data of sensor i; Represents the simulation output of the digital twin model, i.e., the sensor i Simulated performance in current health state, , is the health status of the elderly at time t, indicating the acquired vital signs data and environmental data, Is with sensor i Performance data at specified health states; is sensor i at time t The redundant information evaluation value.

[0013] In one embodiment of the present invention, the redundant information evaluation value The acquisition method includes: calculating the correlation coefficient between any monitoring data of the elderly, and obtaining the redundant information evaluation value of the monitoring data through the correlation coefficient ,as follows: , in, For sensor j The monitoring data after preprocessing at time t, Is to seek and The covariance of is to find the standard deviation of x.

[0014] Based on the same inventive concept, the present invention also provides an elderly health status monitoring and early warning device based on digital twins, including the system, user interaction module, server side and multiple sensors communicatively connected to the server side described in the above technical scheme; wherein, the system is mounted in the server side, and the system uses the computing resources in the server side to calculate the acquired sensor data to obtain the health risk assessment results of the elderly, and presents the health risk assessment results to the caregivers through the user interaction module.

[0015] In one embodiment of the present invention, the device further comprises a wireless communication module, and the user interaction module and the server perform data transmission via the wireless communication module.

[0016] The above technical solution of the present invention has the following advantages compared with the prior art: 1. Comprehensive data management: This invention builds a dynamic and accurate digital twin model for each elderly person through high-precision, real-time collection of vital signs data (such as heart rate, blood pressure, body temperature, etc.) and capture of environmental information (such as indoor and outdoor temperature, humidity, light intensity, etc.). This model not only comprehensively tracks the health status of the elderly, but also deeply reflects the complex interactive relationship between their physiological changes and environmental factors, enabling medical staff to understand the health dynamics of the elderly and the impact of their environment in real time, effectively solving the problem that the existing elderly care service technology lacks comprehensive management of the elderly's health data and environmental information, resulting in data isolation and difficulty in forming a comprehensive health view.

[0017] 2. Personalized health intervention: This invention uses machine learning algorithms to analyze real-time data, accurately predict the health risks that the elderly may face (such as cardiovascular disease attacks, tendency to fall, etc.), automatically generate personalized health intervention suggestions (such as diet adjustments, exercise recommendations, medication reminders, etc.), and quantitatively evaluate the health risk level through hidden danger indexes to ensure that medical staff can formulate corresponding intervention measures. Once a potential health crisis or fall risk is detected, the system immediately triggers the early warning mechanism and quickly notifies family members or professional caregivers through preset communication channels to ensure that intervention measures are taken quickly to avoid adverse events.

[0018] 3. Long-term data accumulation and trend analysis: The present invention has the function of long-term data accumulation and trend analysis. Through the continuous tracking and analysis of the health data of the elderly (such as weekly and monthly health reports), the system can continuously optimize the health management plan to make it more in line with the individual's health status and needs. The system can identify the evolution of health trends (such as weight changes, activity changes, etc.) and potential risks, and generate personalized health trend charts to help the elderly and their families intuitively understand health changes. In addition, the present invention can regularly provide health assessment reports to help medical staff develop more scientific long-term health management plans. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to make the content of the present invention more clearly understood, the present invention is further described in detail below according to specific embodiments of the present invention in conjunction with the accompanying drawings, wherein: Figure 1 It is a structural schematic diagram of a digital twin-based elderly health situation monitoring and early warning system provided by the present invention; Figure 2 This is a workflow diagram of an elderly health situation monitoring and early warning system based on digital twins provided by the present invention; Figure 3 It is the influence weight of some sensors and their collected data for monitoring the vital signs data of the elderly in the present invention; Explanation of the figure numbers in the specification: 100, data acquisition module; 200, digital twin module; 300, risk prediction module; 400, early warning module. DETAILED DESCRIPTION

[0020] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it, but the embodiments are not intended to limit the present invention. Embodiment 1

[0021] See also Figure 1 and Figure 2 As shown, the present invention provides a digital twin-based elderly health status monitoring and early warning system, the system comprising: The data acquisition module 100 is used to collect monitoring data of the elderly in real time, including vital signs data such as heart rate, blood pressure, body temperature, movement trajectory, gait, body position change, and environmental data such as temperature, humidity, and ground conditions; The digital twin module 200 is used to construct a digital twin model and an environmental model that can reflect the changes in the physical state of the elderly in real time based on the vital sign data and the environmental data, respectively. The environmental model predicts the environmental data to evaluate the potential risk results of the elderly in the current environment, and adjusts the output state of the digital twin model based on the potential risk results; The risk prediction module 300 is used to obtain the health risk assessment result of the elderly based on the current output state of the digital twin model; And the early warning module 400 is used to provide customized health management plans for the elderly or send risk warning information to caregivers based on the health risk assessment results.

[0022] It can be seen from the above technical solution that through the data acquisition module 100, the system can obtain the vital signs data and environmental data of the elderly in real time, ensure the timeliness and accuracy of the information, and provide a basis for timely intervention and early warning. The digital twin model 200 construction module creates a personalized digital model for each elderly person, which can accurately reflect the changes in their physical state, making health management more accurate and personalized. The risk prediction module 300 not only considers the vital signs data of the elderly, but also conducts a comprehensive assessment in combination with environmental data, comprehensively analyzes the health status and potential risks of the elderly, and improves the accuracy and comprehensiveness of the early warning. The early warning module 400 can automatically provide customized health management plans for the elderly or send risk warning information to caregivers based on the results of the health risk assessment, realizing intelligent health management and risk prevention and control.

[0023] For multiple sensors monitoring different vital signs data, the monitoring results of various vital sign indicator sensors (such as heart rate, body temperature, blood oxygen saturation, etc.) under different health conditions have different impacts. For example, the heart rate sensor may be interfered with in a high-noise environment, resulting in unstable measurements; while the body temperature sensor usually has higher accuracy and may be more important in a specific health state.

[0024] In the present invention, some sensors involved in obtaining the vital signs data of the elderly and the weight coefficients assigned to the data collected by these sensors are as follows: Figure 3 As shown in the figure. Since the data between sensors are correlated, especially when multiple sensors measure the same physiological indicator, information redundancy is likely to occur. For example, the data provided by multiple heart rate sensors are likely to be highly consistent, which contains a lot of redundant information.

[0025] To solve this problem, the present invention adopts a method of dynamically reducing the sensor evaluation weights corresponding to the redundant information, which can effectively avoid over-reliance on a single information source and thus improve the efficiency of data fusion.

[0026] In this architecture, the adjustment of sensor weights is not only subject to the performance parameters of the sensor itself, but also needs to take into account the health status of the elderly individual or specific disease states (such as cardiovascular disease, diabetes, etc.). For example, for an elderly person with heart disease, during the health monitoring process, the weight of the heart rate sensor may be automatically increased, while the weight of the blood pressure sensor may be relatively low.

[0027] In summary, according to the specific conditions of elderly individuals, it is necessary to dynamically adjust the weights of different types of monitoring data in order to obtain the data representation that best reflects the current health status of the elderly.

[0028] Furthermore, based on the current output state of the digital twin model, the health risk assessment results of the elderly are obtained, including: After removing the abnormal values ​​of the vital sign data and the environmental data, principal component analysis (PCA) is used to perform feature extraction to obtain feature data, and a corresponding weight is dynamically assigned to each monitoring data in the feature data; Based on the current output state of the digital twin model, the monitoring data is fused and calculated according to the weights to obtain state data that can represent the current physical condition of the elderly; Based on the status data, a health risk assessment result for the elderly is obtained.

[0029] Based on the above technical solution, further, the monitoring data is fused and calculated according to the weights to obtain status data that can represent the current physical condition of the elderly, including: Constructing a reinforcement learning model, and inputting the real-time monitoring data of the elderly, the historical performance data of the sensor that collects the monitoring data, and the simulation output results of the digital twin model as system state data into the reinforcement learning model; The monitoring data is processed using a neural network model, and the neural network model is expressed as , where X is the input monitoring data, F is the health status score output by the neural network model, are neural network model parameters; Based on the health status score, the evaluation value of the historical performance data and the impact evaluation value of the environmental data, a reward function R is constructed: , in, 、 、 are weight coefficients, H is the evaluation value of historical performance data, and P is the impact evaluation value of environmental data; specifically, the weight 、 、 The value of is not fixed, but is dynamically adjusted according to real-time monitoring data and feedback results. to represent a function that dynamically adjusts weights based on feedback, i Take 1 or 2 or 3. Through this dynamic adjustment mechanism, the reward function can more accurately reflect the current system state. For example, the importance of each factor may change in different time periods or under different environmental conditions. If in a sports scenario, the weight of the health status score F It may need to be increased; in the sleep scenario, the weight of the environmental factor P This dynamic adjustment mechanism enables the reward function to better adapt to various changing situations.

[0030] When training the reinforcement learning model, the goal is to maximize the value of the reward function, perform iterative training, and continuously adjust the weight of each monitoring data to obtain a trained reinforcement learning model; Based on the trained reinforcement learning model, the current system state data is input to obtain the action with the maximum reward value, i.e., the weight adjustment strategy; According to the weight adjustment strategy, the monitoring data is weighted and fused to obtain status data that can represent the current physical condition of the elderly. : , in, For sensor i The monitoring data after preprocessing at time t, n is the total number of sensors collecting vital signs data and environmental data; Expressed as The weight assigned.

[0031] Furthermore, with the goal of maximizing the value of the reward function, iterative training is performed, and the weight of each monitoring data is continuously adjusted to obtain a trained reinforcement learning model, including: Define the state space S and the weight-adjusted action space A, where Indicates any time i The system status data obtained includes the monitoring data of the elderly, the historical performance data of the sensors that collect the monitoring data, and the simulation output results of the digital twin model. Indicates any time i The calculated weight adjustment strategy includes the weights assigned to the monitoring data; Initialize the parameters of the current Q network and use the current Q network to calculate the current state The Q-values ​​of all possible actions , and select actions according to the ε-greedy strategy , perform actions in the environment , the reward is calculated by the reward function and select the next state , according to the reward and the next state , using the target Q network to calculate the selected action The target Q value y is expressed as follows: , in, is the discount factor, According to the status , execute the action with the maximum Q value, represents the target Q network; Use the target Q value y and the Q value predicted by the current Q network Calculate the loss value, and update the parameters of the current Q network by back-propagating by minimizing the loss value until the network converges or reaches the maximum iteration round, and obtain the trained reinforcement learning model; The current Q network is a neural network structure of multiple convolutional layers and multiple fully connected layers, and the structure of the target Q network is the same as that of the current Q network.

[0032] Based on the above trained reinforcement learning model, the current system state data is input to obtain the action with the maximum reward value, that is, the weight adjustment strategy. The monitoring data in the weight adjustment strategy Weight The expression is as follows: , in, is the historical performance data of sensor i; Represents the simulation output of the digital twin model, i.e., the sensor i Simulated performance in current health state, , is the health status of the elderly at time t, indicating the acquired vital signs data and environmental data, Is with sensor i Performance data at specified health states; is sensor i at time t The redundant information evaluation value.

[0033] Specifically, the redundant information evaluation value The acquisition method includes: calculating the correlation coefficient between any monitoring data of the elderly, and obtaining the redundant information evaluation value of the monitoring data through the correlation coefficient ,as follows: , in, For sensor j The monitoring data after preprocessing at time t, Is to seek and The covariance of is to find the standard deviation of x.

[0034] In order to ensure the safety of the elderly, the environmental model predicts environmental data, evaluates the potential risk outcomes of the elderly in the current environment, and adjusts the output state of the digital twin model based on the potential risk outcomes, specifically including: Accelerometers and gyroscopes are integrated into smart devices worn daily by the elderly (such as bracelets and watches). The accelerometers are capable of monitoring the elderly's movement status in real time. When abnormal acceleration changes are detected, especially sudden free fall accelerations, the system will preliminarily determine that a fall may have occurred. At the same time, combined with the gyroscope data, by analyzing the elderly's posture changes, if it is detected that the posture changes rapidly from upright to horizontal accompanied by rapid movement changes, the system's judgment on the possibility of a fall will be further confirmed.

[0035] In the environment where the elderly live, slippery floor sensors and light sensors are installed. The slippery floor sensor monitors the ground humidity in real time. Once the humidity exceeds the set threshold, the system will record this environmental factor and list it as one of the potential causes of falls. The light sensor continuously monitors the ambient light intensity. When the light intensity is lower than the standard value of 300 lux, the system recognizes that this may affect the elderly's vision, thereby increasing the risk of falling.

[0036] In order to more accurately judge fall events, the system uses a large amount of historical data obtained from sensors to train the machine learning model so that it can accurately identify the pattern of fall events. The model makes comprehensive judgments based on multi-sensor data such as acceleration, posture, and environmental conditions. The system will efficiently fuse and deeply analyze the data of the acceleration sensor, gyroscope, ground slippery sensor (humidity sensor), and light sensor, and judge whether the current environmental conditions of the elderly meet the triggering conditions of a fall event based on pre-set thresholds and rules, such as when the acceleration changes exceed a specific value and the posture becomes horizontal. Once the system determines that a fall has occurred based on the data fusion analysis results, an alarm will be triggered immediately. The alarm information includes the following key contents: (1) The specific location where the fall occurred: With the help of positioning technologies such as GPS or indoor positioning systems, the current geographic location information of the elderly can be accurately obtained so that rescue personnel can quickly locate the scene of the incident.

[0037] (2) Time when the fall occurred: Record the specific time of the incident in detail to provide an important time node for subsequent medical rescue and event tracing.

[0038] (3) Possible causes of falls: The system combines environmental sensor data, such as excessive humidity detected by the slippery floor sensor and insufficient light detected by the light sensor, to clearly provide possible causes of falls, such as slippery floor, insufficient light, etc.

[0039] (4) The current status of the elderly: Through sound sensors or other monitoring methods, it can be determined whether the elderly are awake and whether they are calling for help, so that caregivers can understand the elderly’s condition in advance and make full preparations for rescue work.

[0040] In addition, the system can automatically update the simulated output status of the digital twin, which fully reflects the latest health status and environmental conditions of the elderly. Based on these data, the system can accurately predict the health risks that the elderly may face, such as cardiovascular disease attacks, tendency to fall, etc., and automatically generate personalized health intervention recommendations (such as diet adjustments, exercise recommendations, medication reminders, etc.). At the same time, the system quantitatively assesses the level of health risks through the hidden danger index to help medical staff develop targeted intervention measures. Once a potential health crisis or fall risk is detected, the system will immediately trigger the early warning mechanism and quickly notify family members or professional caregivers through preset communication channels, so as to ensure that intervention measures can be taken in a timely and accurate manner to maximize the life safety and health of the elderly.

[0041] The present invention solves the problems of data isolation, insufficient personalized management and lack of long-term tracking in existing elderly care service technologies through comprehensive data management, personalized health intervention and long-term trend analysis, provides comprehensive and accurate health management services for the elderly, and significantly improves the quality and efficiency of elderly care services. Embodiment 2

[0042] Based on the same inventive concept as Example 1, the present invention also provides an elderly health status monitoring and early warning device based on digital twins, which includes the system described in Example 1, a user interaction module, a server side and multiple sensors communicatively connected to the server side; wherein, the system is installed in the server side, and the system uses the computing resources in the server side to calculate the acquired sensor data to obtain the health risk assessment results of the elderly, and presents the health risk assessment results to the caregivers through the user interaction module.

[0043] Furthermore, the device also includes a wireless communication module, and the user interaction module and the server end perform data transmission through the wireless communication module. The wireless communication module includes but is not limited to a WIFI module and a BLE Bluetooth module, and the physiological and environmental data of the elderly collected by various sensors are transmitted to the server end through the wireless communication module.

[0044] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0045] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0046] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0047] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0048] Obviously, the above embodiments are merely examples for clear explanation and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived from these are still within the protection scope of the invention.

Claims

1. A digital twin-based elderly health status monitoring and early warning system, characterized in that: include: The data collection module is used to collect monitoring data of the elderly in real time, including their vital signs data and environmental data; A digital twin module, for constructing a digital twin model and an environmental model that can respectively reflect the changes in the physical state of the elderly in real time based on the vital sign data and the environmental data, wherein the environmental model evaluates the potential risk results of the elderly in the current environment by predicting the environmental data, and adjusts the output state of the digital twin model based on the potential risk results; A risk prediction module, used to obtain health risk assessment results for the elderly based on the current output state of the digital twin model; And an early warning module is used to provide customized health management plans for the elderly or send risk warning information to caregivers based on the health risk assessment results.

2. The elderly health situation monitoring and early warning system based on digital twins according to claim 1 is characterized in that: The risk prediction module obtains the health risk assessment results of the elderly based on the current output state of the digital twin model, including: Perform feature extraction on the vital sign data and the environmental data after preprocessing to obtain feature data, and dynamically assign corresponding weights to each monitoring data in the feature data; Based on the current output state of the digital twin model, the monitoring data is fused and calculated according to the weights to obtain state data that can represent the current physical condition of the elderly; Based on the status data, a health risk assessment result for the elderly is obtained.

3. The elderly health situation monitoring and early warning system based on digital twins according to claim 2 is characterized in that: The monitoring data is fused and calculated according to the weights to obtain status data that can represent the current physical condition of the elderly, including: Constructing a reinforcement learning model, and inputting the real-time monitoring data of the elderly, the historical performance data of the sensor that collects the monitoring data, and the simulation output results of the digital twin model as system state data into the reinforcement learning model; The monitoring data is processed using a neural network model, and the neural network model is expressed as , where X is the input monitoring data, F is the health status score output by the neural network model, are neural network model parameters; Based on the health status score, the evaluation value of the historical performance data and the impact evaluation value of the environmental data, a reward function R is constructed: ， in, 、 、 They are all weight coefficients, H is the evaluation value of historical performance data, and P is the impact evaluation value of environmental data; When training the reinforcement learning model, the goal is to maximize the value of the reward function, perform iterative training, and continuously adjust the weight of each monitoring data to obtain a trained reinforcement learning model; Based on the trained reinforcement learning model, the current system state data is input to obtain the action with the maximum reward value, i.e., the weight adjustment strategy; According to the weight adjustment strategy, the monitoring data is weighted and fused to obtain status data that can represent the current physical condition of the elderly. .

4. The elderly health situation monitoring and early warning system based on digital twins according to claim 3 is characterized in that: With the goal of maximizing the value of the reward function, iterative training is performed, and the weight of each monitoring data is continuously adjusted to obtain a trained reinforcement learning model, including: Define the state space S and the weight-adjusted action space A, where Indicates any time i The system status data obtained includes the monitoring data of the elderly, the historical performance data of the sensors that collect the monitoring data, and the simulation output results of the digital twin model. Indicates any time i The calculated weight adjustment strategy includes the weights assigned to the monitoring data; Initialize the parameters of the current Q network and use the current Q network to calculate the current state The Q-values ​​of all possible actions , and select actions according to the ε-greedy strategy , perform actions in the environment , the reward is calculated by the reward function and select the next state , according to the reward and the next state , using the target Q network to calculate the selected action The target Q value y; Use the target Q value y and the Q value predicted by the current Q network Calculate the loss value, and update the parameters of the current Q network by back-propagating by minimizing the loss value until the network converges or reaches the maximum iteration round, and obtain the trained reinforcement learning model; The current Q network is a neural network structure of multiple convolutional layers and multiple fully connected layers, and the structure of the target Q network is the same as that of the current Q network.

5. The digital twin-based elderly health status monitoring and early warning system according to claim 4 is characterized in that: The expression of the target Q value y is as follows: ， in, is the discount factor, According to the status , execute the action with the maximum Q value, represents the target Q network.

6. The elderly health situation monitoring and early warning system based on digital twins according to claim 3 is characterized in that: According to the weight adjustment strategy, the monitoring data is weighted and fused to obtain status data that can represent the current physical condition of the elderly. ,as follows: ， in, For sensor i The monitoring data after preprocessing at time t, n is the total number of sensors collecting vital signs data and environmental data; Expressed as The weight assigned.

7. The elderly health situation monitoring and early warning system based on digital twins according to claim 6 is characterized in that: The weight as follows: ， in, is the historical performance data of sensor i; Represents the simulation output of the digital twin model, i.e., the sensor i Simulated performance in current health state, , is the health status of the elderly at time t, indicating the acquired vital signs data and environmental data, Is with sensor i Performance data at specified health states; is sensor i at time t The redundant information evaluation value.

8. The digital twin-based elderly health status monitoring and early warning system according to claim 7 is characterized in that: The redundant information evaluation value The acquisition method includes: calculating the correlation coefficient between any monitoring data of the elderly, and obtaining the redundant information evaluation value of the monitoring data through the correlation coefficient ,as follows: ， in, For sensor j The monitoring data after preprocessing at time t, Is to seek and The covariance of is to find the standard deviation of x.

9. A digital twin-based elderly health status monitoring and early warning device, characterized in that: It includes a system as described in any one of claims 1 to 8, a user interaction module, a server side and a plurality of sensors communicatively connected to the server side; wherein the system is installed in the server side, and the system uses the computing resources in the server side to calculate the acquired sensor data to obtain the health risk assessment results of the elderly, and presents the health risk assessment results to the caregiver through the user interaction module.

10. The digital twin-based elderly health status monitoring and early warning device according to claim 9 is characterized in that: The device also includes a wireless communication module, and the user interaction module and the server end perform data transmission through the wireless communication module.

Images