Intelligent medical rehabilitation equipment management and monitoring system based on edge cloud collaboration

Abstract

The invention is suitable for the technical field of management and monitoring of intelligent medical rehabilitation equipment, and provides an intelligent medical rehabilitation equipment management and monitoring system based on edge cloud collaboration, and the system comprises a terminal equipment layer which is used for collecting health monitoring data of a user, and the terminal equipment comprises an intelligent wheelchair or a nursing bed, and a family health terminal; the edge computing layer is deployed at a local node of a user side and is used for receiving the data of the terminal equipment, performing localization processing through a lightweight artificial intelligence algorithm, judging the abnormity of a health state index and pushing alarm information to a cloud according to the risk level of an event; the cloud platform layer is constructed based on a micro-service architecture, supports large-scale equipment access and resource dynamic scheduling, has a data intercommunication and sharing function and supports multi-role collaborative access and management, and roles include medical staff, family members and nursing workers; and the AI intervention module is used for executing equipment scheduling and remote intervention based on the analysis result.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent medical rehabilitation equipment management and monitoring, more specifically, it relates to an intelligent medical rehabilitation equipment management and monitoring system based on edge cloud cooperation. BACKGROUND

[0002] With the acceleration of the aging society and the increasing demand for chronic disease management, intelligent medical and rehabilitation equipment is increasingly popular in families, communities and medical institutions. In the prior art, intelligent wheelchairs, nursing beds, wearable bracelets and other devices can continuously collect health parameters such as heart rate, body movement and blood pressure, and upload them to a remote cloud server through Wi-Fi, Bluetooth and other means for data analysis and alarm processing. However, this highly cloud-dependent data processing architecture has the following significant problems:

[0003] Firstly, in the cloud centralized processing model, the system response time is high, especially in scenarios involving falls, arrhythmia and other situations that require rapid response, there is a risk of alarm delay and feedback lag, which is not conducive to real-time intervention in emergencies. Secondly, the cloud-deployed artificial intelligence model cannot be personalized according to individual user behavior patterns, disease types and functional use preferences, limiting the accuracy and adaptability of the model. Especially in scenarios where edge devices have limited functionality, there is currently a lack of a mechanism to sink frequently used AI functions to the local device while recycling low-frequency functions to the cloud, thus failing to balance resource utilization efficiency while ensuring real-time performance.

[0004] Furthermore, there is currently a lack of effective mechanisms to implement dynamic scheduling and updating of edge AI functions, so that the AI algorithms deployed on user local devices cannot be automatically optimized according to environmental, state or behavior changes, making it difficult to achieve personalized health management that varies from person to person and from disease to disease. In addition, in traditional system architecture, data protocols are not unified, and there are information silos between hospitals, families and communities, hindering the realization of cross-scenario collaborative management and data sharing.

[0005] Therefore, based on the above problems, the present application proposes an intelligent medical rehabilitation equipment management and monitoring system based on edge cloud cooperation. SUMMARY

[0006] In view of the deficiencies in the prior art, the purpose of the present application is to provide an intelligent medical rehabilitation equipment management and monitoring system based on edge cloud cooperation.

[0007] To achieve the above-mentioned purpose, the present application provides the following technical solutions:

[0008] The intelligent medical rehabilitation equipment management and monitoring system based on edge cloud cooperation comprises:

[0009] Terminal device layer, used for collecting health monitoring data of users, the terminal device comprising:

[0010] Intelligent wheelchair or nursing bed, integrated with multi-modal sensors, the multi-modal sensors comprising pressure sensors, acceleration sensors and bioelectrode sensors, used for realizing heart rate monitoring, body movement detection and abnormal behavior recognition of users;

[0011] Home health terminal, used for accessing blood pressure meters, smart bracelets and other third-party health devices, and performing standardized processing on collected data;

[0012] Edge computing layer, deployed at a local node of users, used for receiving data of the terminal device, and performing localized processing through a lightweight artificial intelligence algorithm, judging abnormality of health status indicators, and pushing alarm information to the cloud according to a risk level of an event, the risk level comprising high risk, medium risk and low risk;

[0013] The lightweight artificial intelligence algorithm is synchronized through the cloud platform layer, and is synchronized based on a heat value of a user;

[0014] Cloud platform layer, constructed based on a micro-service architecture, supporting large-scale device access and resource dynamic scheduling, having a data intercommunication and sharing function, supporting multi-role collaborative access and management, the roles comprising medical staff, family members and caregivers; and further comprising an AI intervention module, used for performing device scheduling and remote intervention based on analysis results.

[0015] The application further provides that the cloud platform maintains a heat value of each function module of an edge node If the heat value of a function module exceeds a set migration threshold value, the cloud platform automatically deploys the function module to the edge node to reduce response delay; if the heat value is lower than a recovery threshold value, the function is unloaded from the edge node and processed by the cloud.

[0016] The application further provides that the heat value is calculated based on three parameters of function call frequency, average response time and local hit rate, and a weighted calculation, wherein the calculation formula is:

[0017]

[0018] wherein,

[0019] Heat value of function module j on edge node i at time t;

[0020] Function call frequency, i.e., the number of calls per unit time;

[0021] Average response latency, reflecting the response efficiency of the module;

[0022] Local cache hit rate, indicating whether to be called locally first;

[0023] Adjustable weight coefficient, satisfying alpha+beta+gamma=1.

[0024] The application is further provided: in order to avoid frequent migration or shock of the model caused by short-time burst calling, a historical heat decay factor is introduced when the system calculates the heat value of the functional module of the edge node, and the heat value is calculated by using the following recursive formula:

[0025]

[0026] Among them,

[0027] Heat value of functional module j on edge node i at time t;

[0028] Frequency of functional calling, i.e. the number of calls per unit time;

[0029] Average response latency, reflecting the response efficiency of the module;

[0030] Local cache hit rate, indicating whether to be called locally first;

[0031] Adjustable weight coefficient, satisfying alpha+beta+gamma=1.

[0032] Lambda is in (0, 1): decay factor, used to control the influence weight of historical heat on current heat.

[0033] The application is further provided: the frequency of functional calling , average response latency and local cache hit rate are calculated by the following formula:

[0034]

[0035] ;

[0036]

[0037] Among them, is the call count at each time in the time window;

[0038] Delta t is the time window width;

[0039] The number of times of direct function calls from the local;

[0040] The total number of function calls (including cloud callback).

[0041] The application further provides that the data collected by the multi-modal sensors in the terminal device layer is encapsulated and transmitted through a unified data communication protocol, to solve the problem of cross-device and cross-scene data incompatibility, and the protocol supports the standardized description of sensor identification, timestamp, data type and numerical content.

[0042] The application further provides that the lightweight artificial intelligence algorithm in the edge computing layer includes an abnormality recognition model based on heart rate variability (HRV) and a fall detection model based on an acceleration sensor, and the model is deployed through a local computing node and supports dynamic updating of model parameters.

[0043] The application further provides that the cloud platform layer adopts a SpringCloud or an equivalent micro-service architecture, supports concurrent access of at least 100,000 devices, and the architecture supports container orchestration and elastic load scheduling.

[0044] The application further provides that the system supports providing differentiated data access permissions and alarm pushing to different roles through a mobile terminal, the mobile terminal can display user physical sign data, abnormal event records and a remote configuration interface according to an identity authentication result, and the role differentiated data access permissions and alarm pushing can adapt to the edge computing layer, and the edge computing layer is adaptively adjusted according to different roles, including an algorithm saved in the edge computing layer and an offloaded algorithm.

[0045] The application further provides that the risk level is determined based on joint analysis of multiple source indexes, the indexes include heart rate, respiratory rate, body movement frequency, fall event identification and historical health event frequency, and a weighted strategy is adopted for comprehensive evaluation.

[0046] In summary, the present application includes at least one of the following beneficial technical effects:

[0047] 1、System adopts distributed micro-service architecture, supports concurrent access of at least 100,000 devices, and combines container orchestration and load balancing mechanism, has excellent high-concurrency processing capability, guarantees stable operation in large-scale deployment scenarios. At the same time, the light-weight AI algorithm is deployed in the edge computing layer to realize the local real-time processing and response of key sign data such as fall detection, abnormal heart rate, etc., effectively reduce the dependence on the cloud, and improve the alarm response speed. In addition, the system breaks down the data barriers between hospitals, families and communities through a unified communication protocol, realizes the interconnection of multi-terminal data; through the multi-role hierarchical management mechanism, medical staff, family members and caregivers can participate in health monitoring tasks through mobile terminals, improving the interactivity and collaboration of the system. Modular design also facilitates future function extension and new device access, enhancing the long-term evolvability of the system.

[0048] 2、By maintaining the heat value of each edge node function module on the cloud platform, combining function call frequency, average response time and local hit rate, etc. Index, dynamically adjust the deployment state of the module between the edge and the cloud, realize the resource coordination mechanism of sinking common functions to the local and collecting uncommon functions to the cloud. The present application also introduces a heat decay factor to avoid the heat value from fluctuating sharply due to short-term bursts, effectively improving the intelligence and stability of the system's function scheduling, thereby saving edge resource overhead and improving system efficiency and response robustness. BRIEF DESCRIPTION OF DRAWINGS

[0049] Fig. 1 The module function diagram of the intelligent medical rehabilitation equipment management and monitoring system based on edge cloud cooperation of the present application.

[0050] Fig. 2 The flowchart of risk event processing in the present application.

[0051] Fig. 3 The flowchart of the heat value driven function module scheduling in the present application. DETAILED DESCRIPTION

[0052] It should be noted that the embodiments and features in the embodiments in the present application can be combined with each other without conflict. The present application will be described in detail below with reference to the accompanying drawings and in combination with embodiments.

[0053] It should be noted that, unless otherwise specified, all technical and scientific terms used in the present application have the same meaning as generally understood by those skilled in the art to which the present application belongs.

[0054] Please refer to Figs. 1-3 The present application provides the following technical solutions:

[0055] Embodiment one, an intelligent medical rehabilitation equipment management and monitoring system based on edge cloud cooperation, comprising:

[0056] a terminal device layer for collecting health monitoring data of a user, the terminal device comprising: a smart wheelchair or a nursing bed integrating multi-modal sensors including pressure sensors, acceleration sensors and bioelectrode sensors for realizing heart rate monitoring, body movement detection and abnormal behavior recognition of the user;

[0057] a home health terminal for accessing a blood pressure meter, a smart bracelet and other third-party health devices and performing standardized processing on collected data to ensure consistency of data format and cross-platform compatibility. Through a standardized interface, the terminal device layer can seamlessly access medical instruments from different manufacturers, thereby realizing multi-source data fusion and supporting dynamic identification of new devices and automatic configuration.

[0058] an edge computing layer deployed at a local node at a user side for receiving data of the terminal device and performing localized processing through a lightweight artificial intelligence algorithm to judge abnormality of a health status indicator and push alarm information to the cloud according to a risk level of an event, the risk level including high risk, medium risk and low risk;

[0059] The lightweight artificial intelligence algorithm is synchronized through the cloud platform layer based on a user adoption heat value.

[0060] Specifically, the system determines a risk level according to current user physical sign changes and event characteristics, in combination with historical event frequency and behavior patterns, and is divided into:

[0061] high risk: falling, sudden heart rate change, long time without action, which belongs to a state requiring urgent intervention;

[0062] medium risk: mild heart rate abnormality, activity frequency lower than daily baseline, which prompts attention;

[0063] low risk: slight body movement change, uncomfortable temperature and humidity environment, etc., which provides a life prompting suggestion.

[0064] The determination of the risk level is based on joint analysis of multi-source indicators including heart rate, respiratory rate, body movement frequency, fall event identification and historical health event frequency, and a weighted strategy is adopted for comprehensive evaluation.

[0065] After recognizing an event, the edge node can instantly push a local sound and light alarm and upload the event to the cloud in a structured format. The structured data content includes user identification, event type, occurrence time, sensor value and preliminary determination result, etc.

[0066] The cloud platform layer is constructed based on a micro-service architecture, supports large-scale device access and dynamic resource scheduling, has a data intercommunication and sharing function, supports multi-role collaborative access and management, the roles include medical staff, family members and caregivers, and further comprises an AI intervention module for performing device scheduling and remote intervention based on analysis results.

[0067] The system supports providing differentiated data access rights and alarm pushing to different roles through a mobile terminal, the mobile terminal can display user sign data, abnormal event records and remote configuration interfaces according to the identity authentication result, and the role differentiated data access rights and warning pushing can adapt to the edge computing layer, the edge computing layer adaptively adjusts according to different roles, including algorithms saved in the edge computing layer and offloaded algorithms.

[0068] Specifically, according to the different roles, the system supports differentiated access of the mobile terminal:

[0069] Including, the medical staff end: comprehensive access to user data, view historical curves and trends, and can perform diagnosis support and health guidance suggestions;

[0070] The family end: limited access to sign data and alarm records to ensure privacy security while obtaining key information;

[0071] The caregiver end: can perform device debugging and operation and maintenance viewing to improve on-site service efficiency.

[0072] At the same time, the edge computing layer selectively loads different algorithm modules according to the user identity, saves resources and improves processing efficiency, such as default deployment of a fall detection module for home users, and deployment of more complex heart rate analysis and HRV prediction modules in the hospital end.

[0073] Specifically, a SpringCloud or similar micro-service architecture is adopted to support concurrent access of 100,000 devices, and the platform supports the following features:

[0074] Container orchestration: such as Kubernetes to schedule resources and deploy instances;

[0075] Distributed storage: such as Ceph, MinIO to support massive sign data storage, backup and analysis;

[0076] Security mechanism: supports user authentication based on OAuth2.0, TLS data encryption transmission and role permission control based on the principle of least privilege.

[0077] The data fusion and AI intervention includes the following functions: individual and group health situation analysis based on uploaded data, early warning of high-risk users by the system combined with a chronic disease prediction model, personalized intervention suggestions for users with high blood pressure, arrhythmia, and pre-diabetes, and dynamic scheduling of device operation according to system load and user state to form an active intervention mechanism and automatically complete edge / cloud function allocation and update.

[0078] Because too many functional modules are long-term resident in the edge node, it will occupy a large amount of computing and storage resources, resulting in a significant increase in device cost, increased energy consumption, and even response delay when processing high-concurrency data streams, thereby affecting the real-time performance and stability of the system. Therefore, in order to achieve efficient resource allocation and function scheduling between the edge node and the cloud platform, the system designs a function module migration mechanism based on a heat value. The cloud platform maintains the heat value of each functional module in the edge node If the heat value of a functional module exceeds the set migration threshold, the cloud platform automatically deploys the functional module to the edge node to reduce the response delay; if the heat value is below the recovery threshold, the function is unloaded from the edge node and processed by the cloud.

[0079] Specifically, the cloud platform continuously maintains the heat value of each functional module in the edge node to measure the usage activity and local adaptability of the function in the current time period. When the heat value of a functional module exceeds the set migration threshold , the system automatically migrates the functional module from the cloud to the corresponding edge node to complete data processing locally, thereby reducing transmission latency and response delay and enhancing real-time response capability.

[0080] Conversely, if the heat value of a functional module is below the recovery threshold , it indicates that the function has a low frequency of local use and weak dependence on real-time performance. At this time, the system unloads the module from the edge node and processes it centrally on the cloud platform to free up local computing resources on the edge node and provide support for other higher-priority functions. The entire process does not require human intervention, and through the dynamic evolution of the heat value and the automatic execution of the system strategy engine, resources are reasonably allocated between different nodes to form an intelligent scheduling mechanism of “cold and hot function hierarchical management and edge-cloud collaborative processing”. This mechanism can effectively reduce the load of edge devices and improve the overall resource utilization and economy of the system.

[0081] Specifically, the heat value is calculated based on the weighted combination of the function call frequency, average response latency, and local hit rate, and the calculation formula is:

[0082]

[0083] wherein, the heat value of the function module j on the edge node i at time t;

[0084] the function call frequency, i.e., the number of calls per unit time;

[0085] the average response delay, reflecting the response efficiency of the module;

[0086] the local cache hit rate, indicating whether to be called locally first;

[0087] the adjustable weight coefficient, satisfying α+β+γ=1, which can be dynamically adjusted according to the application scenario.

[0088] the function call frequency the average response delay and the local cache hit rate are calculated by the following formula:

[0089]

[0090]

[0091]

[0092] wherein, is the call count at each time in the time window;

[0093] Δt is the time window width;

[0094] is the number of direct local calls of the function;

[0095] is the total number of function calls (including cloud callback).

[0096] The data collected by the multi-modal sensors in the terminal device layer are encapsulated and transmitted through a unified data communication protocol, which is used to solve the problem of incompatible data across devices and across scenes. The protocol supports standardized description of sensor identification, timestamp, data type and value content. The protocol also contains error detection mechanism and data synchronization instruction field, which ensures the stable access and long-term operation of the edge node.

[0097] The lightweight artificial intelligence algorithm in the edge computing layer includes an abnormality recognition model based on heart rate variability (HRV) and a fall detection model based on acceleration sensor. The model is deployed through a local computing node and supports dynamic updating of model parameters. ​

[0098] In actual use, fall recognition as a key function in the monitoring of the elderly and rehabilitation patients is crucial for its accuracy and response time. The fall detection model based on acceleration sensor deployed in the system is usually integrated with the acceleration detection unit of the intelligent wheelchair or nursing bed to realize real-time monitoring and rapid response on the edge side. Specifically, taking the intelligent wheelchair as an example, the wheelchair body integrates multi-modal sensing devices such as three-axis acceleration sensors, bioelectrode (ECG) sensors, and pressure sensors, which can real-time perceive the user's body posture changes, heart rate fluctuations, and body weight distribution during use.

[0099] When the user performs a fall risk action (such as rapid descent or impact acceleration exceeding the threshold), the acceleration sensor captures abnormal data fluctuations in real time and inputs them into the lightweight AI model on the edge node for analysis. The model uses a combination of sliding time window feature extraction, threshold segmentation judgment, and machine learning classification to quickly determine whether it is a real fall event. Once the model determines a "high-risk fall" event, the edge node immediately triggers a local sound and light alarm to remind surrounding personnel, and sends alarm information (such as push notifications or SMS) to family members through the home health terminal or mobile application.

[0100] At the same time, event-related data (including sensor raw data, determination results, trigger time, user ID, etc.) are structured and encapsulated, and uploaded to the cloud platform for subsequent comprehensive evaluation, health report update, and intervention plan adjustment by medical staff or remote monitoring center. The cloud system can also further build a personalized risk model based on historical fall events and user physical sign trends for disease prediction and rehabilitation intervention optimization, forming a closed-loop intelligent health management process.

[0101] In embodiment two, the system introduces a historical heat decay factor when calculating the heat value of the function module on the edge node to avoid frequent migration or shock of the model caused by short-term burst calls. The heat value is calculated using the following recursive formula:

[0102]

[0103] Where,

[0104] Heat value of function module j on edge node i at time t;

[0105] Function call frequency, i.e., the number of calls per unit time;

[0106] Average response latency, reflecting the response efficiency of the module;

[0107] Local cache hit rate, indicating whether to be called locally first;

[0108] Adjustable weight coefficients, satisfying α+β+γ=1;

[0109] λ∈(0,1): decay factor, used to control the influence weight of historical hotness on current hotness.

[0110] The decay factor λ is introduced to avoid the dramatic fluctuation of hotness value caused by the dramatic increase or decrease of call frequency in a short time, so that the system is more stable and robust when making edge function migration or cloud recycling decisions.

[0111] When λ tends to 1, the system tends to retain the historical state, and the hotness value is updated slowly, which is suitable for stable system;

[0112] When λ is small, the system is more responsive to the latest changes, which is suitable for fast adaptation to new environment;

[0113] Usually, λ∈[0.6,0.95], which can be dynamically adjusted to adapt to the response stability requirements in different scenarios.

[0114] This method constructs a hotness evolution model with a first-order memory mechanism, which can effectively buffer the frequent switching risk of the system and improve the intelligence and execution efficiency of the function distribution strategy.

[0115] The device embodiments described above are only illustrative and not all embodiments. For example, the division of the units is only a logical functional division, and in actual implementation, there can be another division method. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection between the units through some communication interface, and can be electrical, mechanical or other forms. All other embodiments obtained by those skilled in the art without creative labor shall belong to the scope of protection of the present application.

Claims

1. An intelligent medical rehabilitation equipment management and monitoring system based on edge-cloud collaboration, characterized in that: include: The terminal device layer is used to collect users' health monitoring data, and the terminal device includes: Intelligent wheelchairs or nursing beds integrate multimodal sensors, including pressure sensors, acceleration sensors, and bioelectrode sensors, to enable users to monitor heart rate, detect body movement, and identify abnormal behavior. A home health terminal is used to connect to third-party health devices such as blood pressure monitors and smart bracelets, and to standardize the collected data. The edge computing layer, deployed on a local node on the user side, is used to receive data from the terminal device, perform localized processing through a lightweight artificial intelligence algorithm, determine abnormalities in health status indicators, and push alarm information to the cloud according to the risk level of the event, including high risk, medium risk, and low risk. Among them, the lightweight artificial intelligence algorithm is synchronized through the cloud platform layer, based on the popularity value adopted by the user; The cloud platform layer is built on a microservice architecture, supports large-scale device access and dynamic resource scheduling, has data interoperability and sharing capabilities, and supports multi-role collaborative access and management, including medical staff, family members and caregivers; it also includes an AI intervention module for performing device scheduling and remote intervention based on analysis results.

2. The intelligent medical rehabilitation equipment management and monitoring system based on edge-cloud collaboration according to claim 1, characterized in that: The cloud platform maintains the heat value of each edge node's functional module. If the popularity value of a certain functional module exceeds the set migration threshold, the cloud platform will automatically deploy the functional module to the edge node to reduce response latency; if the popularity value is lower than the recycling threshold, the function will be unloaded from the edge node and processed by the cloud.

3. The intelligent medical rehabilitation equipment management and monitoring system based on edge-cloud collaboration according to claim 2, characterized in that: The heat value The calculation is based on a weighted average of three parameters: function call frequency, average response latency, and local hit rate. The calculation formula is as follows: in, The heat value of functional module j on edge node i at time t; Function call frequency, i.e., the number of times it is called per unit of time; Average response latency reflects the response efficiency of this module; Local cache hit rate indicates whether local calls are prioritized. The adjustable weighting coefficients satisfy α+β+γ=1.

4. The intelligent medical rehabilitation equipment management and monitoring system based on edge-cloud collaboration according to claim 2, characterized in that: To avoid frequent model migrations or oscillations caused by sudden short-term calls, the system incorporates a historical popularity decay factor when calculating the popularity value of functional modules at edge nodes. This popularity value is calculated using the following recursive formula: in, The heat value of functional module j on edge node i at time t; Function call frequency, i.e., the number of times it is called per unit of time; Average response latency reflects the response efficiency of this module; Local cache hit rate indicates whether local calls are prioritized. Adjustable weighting coefficients satisfy α+β+γ=1; λ∈(0,1): Attenuation factor, used to control the weight of the influence of historical popularity on current popularity.

5. The intelligent medical rehabilitation equipment management and monitoring system based on edge-cloud collaboration according to claim 3 or 4, characterized in that: The frequency of function calls Average response time and local cache hit rate The calculation is performed using the following formula: ； in, It is the call count at each moment within the time window; Δt is the width of the time window; This represents the number of times the function is directly called from the local machine. This represents the total number of function calls (including cloud callbacks).

6. The intelligent medical rehabilitation equipment management and monitoring system based on edge-cloud collaboration according to claim 1, characterized in that: The data collected by the multimodal sensors in the terminal device layer is encapsulated and transmitted through a unified data communication protocol to solve the problem of data incompatibility across devices and scenarios. The protocol supports standardized descriptions of sensor identifiers, timestamps, data types, and numerical content.

7. The intelligent medical rehabilitation equipment management and monitoring system based on edge-cloud collaboration according to claim 1, characterized in that: The lightweight artificial intelligence algorithms in the edge computing layer include an anomaly recognition model based on heart rate variability (HRV) and a fall detection model based on acceleration sensors. These models are deployed through local computing nodes and support dynamic updates of model parameters.

8. The intelligent medical rehabilitation equipment management and monitoring system based on edge-cloud collaboration according to claim 7, characterized in that: The cloud platform layer adopts Spring Cloud or an equivalent microservice architecture, supporting concurrent access from at least 100,000 devices. The architecture supports container orchestration and elastic load balancing.

9. The intelligent medical rehabilitation equipment management and monitoring system based on edge-cloud collaboration according to claim 8, characterized in that: The system supports providing differentiated data access permissions and alarm push notifications to different roles via mobile terminals. The mobile terminals can display user vital signs data, abnormal event records, and remote configuration interfaces based on identity authentication results. Furthermore, the role-based differentiated data access permissions and alarm push notifications can adapt to the edge computing layer. The edge computing layer makes adaptive adjustments based on different roles, including algorithms stored in the edge computing layer and offloading algorithms.

10. The intelligent medical rehabilitation equipment management and monitoring system based on edge-cloud collaboration according to claim 1, characterized in that: The risk level is determined based on a multi-source indicator joint analysis, which includes heart rate, respiratory rate, body movement frequency, fall event markers, and historical health event frequency, and a weighted strategy is used for comprehensive evaluation.

Images