Health state intelligent monitoring and guidance system for the elderly and rehabilitation persons

Abstract

The present application relates to the technical field of health monitoring, in particular to a health state intelligent monitoring and guiding system for the elderly and rehabilitation persons, comprising: a data acquisition layer, used for acquiring non-contact physical sign data, environmental data, equipment use frequency and measurement data, flexible wearable device data, voice interaction data and individual basic disease and rehabilitation target data through a non-contact physical sign acquisition module, an environmental and equipment data acquisition module and a multi-modal physiological and basic data acquisition module. The non-contact physical sign acquisition module of the data acquisition layer can collect physical sign and behavior data without active operation of the user, the environmental and equipment data acquisition module is used to collect environmental and medical equipment use data, physical examination reports, the multi-modal physiological and basic data acquisition module is used to integrate physiological emotion and basic disease information, and the problem of incomplete data collection caused by the difficulty of the user to cooperate in the existing system is solved.

Description

Technical Field

[0001] This invention relates to the field of health monitoring technology, specifically to an intelligent monitoring and guidance system for the health status of the elderly and those undergoing rehabilitation. Background Technology

[0002] With the increasing aging population and growing demand for rehabilitation medicine, health management for the elderly and those undergoing rehabilitation is gradually becoming a focus. Health status monitoring and guidance systems typically collect two types of key data: environmental data, including temperature, air pressure, and humidity in living or activity environments, which directly affects human comfort and the stability of physiological indicators; for example, sudden temperature changes may cause blood pressure fluctuations in the elderly. Personal physical condition data, encompassing age, sleep, weight, heart health, emotional health, blood oxygen saturation, and body temperature, is the core basis for assessing health status. Simultaneously, video monitoring equipment can assist in observing behavioral states, and the frequency of use of monitoring devices (such as blood pressure monitors and blood glucose meters) can indirectly reflect the user's level of attention to their own health and their ability to act. All of the above data collectively constitute the basic information source for health assessment.

[0003] While existing health monitoring systems can collect some environmental and personal health data, they often fail to adequately support elderly individuals and those in rehabilitation due to declining physical function or limited mobility. This results in reduced device usage frequency and incomplete data collection. Existing systems neither employ contactless technology to compensate for the lack of proactive data collection nor deeply integrate abnormal changes in usage frequency with behavioral characteristics and environmental parameters, making it difficult to identify potential health risks in advance. Furthermore, existing systems can only provide text or voice guidance, failing to link hospital treatment and rehabilitation guidance with devices in daily life for immediate intervention, leading to significant intervention delays. More importantly, existing systems do not consider individual differences within this population, using uniform standards for assessment and failing to track long-term rehabilitation effects and optimize plans. This results in a lack of continuity and personalization in health management, making it difficult to meet the full-cycle health needs of specific populations. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an intelligent health monitoring and guidance system for the elderly and those undergoing rehabilitation. This system solves the problems of incomplete data collection, delayed intervention, and lack of personalization and continuity caused by the difficulty of users actively cooperating in existing technologies.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an intelligent health monitoring and guidance system for the elderly and those undergoing rehabilitation, comprising:

[0006] The data acquisition layer is used to acquire non-contact vital sign data, environmental data, equipment usage frequency and measurement data, flexible wearable device data, voice interaction data, and individual underlying disease and rehabilitation goal data through non-contact vital sign acquisition module, environmental and equipment data acquisition module, and multimodal physiological and basic data acquisition module.

[0007] The intelligent processing layer is used for:

[0008] The acquired data is then merged and cleaned.

[0009] Through the individual basic parameter adaptation module, the data weights and evaluation criteria are dynamically adjusted according to the individual's underlying diseases and rehabilitation goals;

[0010] By using the risk prediction model module, combined with changes in equipment usage frequency, fluctuations in environmental parameters, time-series data of physiological indicators, and types of underlying diseases, health risks and risk triggers can be predicted.

[0011] The dynamic weighting and correlation analysis module adjusts data weights based on the rehabilitation stage and individual underlying diseases, and analyzes the causes of abnormal equipment usage frequency.

[0012] Assessment guidance layer, used for:

[0013] The tiered assessment module combines risk prediction results, real-time data, physical examination reports, individual underlying diseases and rehabilitation goals to generate a health score and trigger tiered early warnings.

[0014] Through the personalized guidance and rehabilitation program generation module, a combination of environmental adjustment, physiological intervention and rehabilitation training is generated and presented in multiple forms such as voice broadcast, smart screen animation demonstration and paper graphic manual;

[0015] The equipment linkage layer is used for:

[0016] Through the intelligent device control module, intelligent devices such as dehumidifiers, thermostatic mattresses, smart meal boxes, rehabilitation equipment, and home service robots can automatically perform operations according to the guidance plan;

[0017] Through the multi-scenario emergency response submodule, when abnormal situations such as falls or sudden drops in heart rate are detected, emergency responses such as indoor sound and light alarms, linkage with service robots, automatic opening of room doors, or push of medication records are triggered.

[0018] The intervention effect feedback module continuously collects environmental and physiological data after the smart device performs the operation, and adjusts the device parameters or marks the intervention plan based on the data feedback.

[0019] Effect backtracking layer, used for:

[0020] The rehabilitation effectiveness tracking module generates rehabilitation reports to evaluate the rehabilitation plan;

[0021] Through the solution iteration and optimization module, data weights and intervention measures are adjusted based on effect tracking data and user feedback, and the individual basic parameter adaptation library and risk prediction model are updated.

[0022] Furthermore, in the data acquisition layer:

[0023] The non-contact vital sign acquisition module is used to deploy millimeter-wave radar and far-infrared thermal imaging equipment to collect vital sign data such as respiratory rate, heart rate variability, body posture, and gait characteristics without contact. At the same time, it can capture behaviors such as falls, prolonged sitting, and abnormal wandering, and upload the data in real time after encryption.

[0024] The environmental and equipment data acquisition module is used to collect data on temperature, air pressure, humidity, and PM2.5 concentration through distributed sensors. At the same time, it connects to medical devices such as blood pressure monitors, body fat scales, and blood glucose meters through an IoT module to record the usage frequency, measurement values, and online status of the medical devices, forming an equipment operation log.

[0025] Furthermore, in the data acquisition layer, the multimodal physiological and basic data acquisition module is used to collect sleep structure, blood oxygen saturation, body temperature, and blood glucose trend data through flexible wearable devices. At the same time, it combines with the voice interaction module to obtain emotional state, and allows guardians or medical staff to enter individual underlying diseases and rehabilitation goals to form a fusion dataset of physiological, emotional, and underlying diseases.

[0026] Furthermore, in the intelligent processing layer, the individual basic parameter adaptation module is used to construct a mapping library between basic diseases and data weights, and adjust the weights of various types of data according to the individual's basic diseases; at the same time, it sets phased data thresholds according to rehabilitation goals and dynamically adjusts the evaluation criteria of each indicator.

[0027] Furthermore, in the intelligent processing layer, the risk prediction model module is used to construct a prediction model based on long short-term memory network and attention mechanism. The prediction model receives the frequency change rate, environmental parameter fluctuation value, physiological indicator time series data and basic disease type as input, and outputs the health risk probability and risk triggers. The prediction model is continuously iteratively optimized through user historical data.

[0028] Furthermore, in the intelligent processing layer, the dynamic weight and correlation analysis module is used to adjust the data weights in a dual manner, combining the rehabilitation stage and the individual's underlying disease. When the device usage frequency is abnormal, it is linked to analyze the number of abnormal postures, heart rate variability, and emotional state during the same period to locate the causes such as decreased operational ability, forgetting to use, or resistance to use.

[0029] Furthermore, in the aforementioned assessment guidance layer:

[0030] The grading and assessment module is used to combine risk prediction results, real-time data, individual underlying diseases and rehabilitation goals, and generate a health score using basic standards and individual correction values, and trigger graded early warnings based on the health score;

[0031] The multi-role early warning module is used to push reminder information or real-time data to at least one of the following based on the level of early warning: guardian, rehabilitation therapist, nursing home care station, community medical station, and emergency system. The information includes at least one of the following: risk trigger, real-time vital signs data, on-site video clips, indoor positioning information, and history of underlying diseases or recent medication records.

[0032] Furthermore, in the assessment guidance layer, the personalized guidance and rehabilitation plan generation module is used to generate a combination plan of environmental adjustment, physiological intervention and rehabilitation training based on the assessment results and rehabilitation goals; the plan is presented through at least one of voice broadcast, smart screen animation demonstration or paper-based graphic manual.

[0033] Furthermore, in the device linkage layer:

[0034] The intelligent device control module is used to link at least one of the following via IoT protocol: dehumidifier, constant temperature mattress, smart meal box, rehabilitation equipment or home service robot, and automatically execute operations according to the guidance plan;

[0035] The multi-scenario emergency response submodule is used to automatically trigger indoor sound and light alarms and link service robots to check when a fall is detected, while sending location information and video clips to the guardian; when a sudden drop in heart rate is detected, it automatically opens the smart door lock and pushes medication records to the emergency system.

[0036] Furthermore, in the effect backtracking layer, the scheme iteration optimization module is used to adjust the data weights and intervention measures for intervention schemes marked as to be optimized, in combination with rehabilitation effect tracking data and user feedback, and update the individual basic parameter adaptation library and risk prediction model according to long-term data trends.

[0037] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0038] This invention utilizes a non-contact vital sign acquisition module in the data acquisition layer to collect vital sign and behavioral data without user intervention. This is combined with an environmental and equipment data acquisition module to collect environmental and medical equipment usage data, and a multimodal physiological and basic data acquisition module to integrate physiological, emotional, and underlying disease information, addressing the problem of incomplete data collection caused by user cooperation difficulties in existing systems. The intelligent processing layer adjusts data weights and assessment standards according to underlying diseases and rehabilitation goals through an individual basic parameter adaptation module, and combines this with a risk prediction model module to identify health risks in advance, avoiding the drawbacks of standardized approaches in existing systems. The assessment guidance layer generates personalized solutions in various forms, including voice broadcasts. The device linkage layer links intelligent devices to perform real-time intervention, addressing the problem of delayed intervention in existing systems. The effect tracking layer achieves long-term personalized health management through rehabilitation effect tracking and iterative optimization of the plan, compensating for the lack of continuity in existing systems and comprehensively improving the overall health management of the elderly and those undergoing rehabilitation. Attached Figure Description

[0039] Figure 1 This is a system structure diagram of the present invention. Detailed Implementation

[0040] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0041] Please see Figure 1 This invention provides an intelligent health monitoring and guidance system for the elderly and those undergoing rehabilitation, comprising:

[0042] The data acquisition layer is used to acquire non-contact vital sign data, environmental data, equipment usage frequency and measurement data, flexible wearable device data, voice interaction data, and individual underlying disease and rehabilitation goal data through non-contact vital sign acquisition module, environmental and equipment data acquisition module, and multimodal physiological and basic data acquisition module.

[0043] The intelligent processing layer is used for:

[0044] The acquired data is then merged and cleaned.

[0045] Through the individual basic parameter adaptation module, the data weights and evaluation criteria are dynamically adjusted according to the individual's underlying diseases and rehabilitation goals;

[0046] By using the risk prediction model module, combined with changes in equipment usage frequency, fluctuations in environmental parameters, time-series data of physiological indicators, and types of underlying diseases, health risks and risk triggers can be predicted.

[0047] The dynamic weighting and correlation analysis module adjusts data weights based on the rehabilitation stage and individual underlying diseases, and analyzes the causes of abnormal equipment usage frequency.

[0048] Assessment guidance layer, used for:

[0049] The tiered assessment module combines risk prediction results, real-time data, physical examination reports, individual underlying diseases and rehabilitation goals to generate a health score and trigger tiered early warnings.

[0050] Through the personalized guidance and rehabilitation program generation module, a combination of environmental adjustment, physiological intervention and rehabilitation training is generated and presented in multiple forms such as voice broadcast, smart screen animation demonstration and paper graphic manual;

[0051] The equipment linkage layer is used for:

[0052] Through the intelligent device control module, intelligent devices such as dehumidifiers, thermostatic mattresses, smart meal boxes, rehabilitation equipment, and home service robots can automatically perform operations according to the guidance plan;

[0053] Through the multi-scenario emergency response submodule, when abnormal situations such as falls or sudden drops in heart rate are detected, emergency responses such as indoor sound and light alarms, linkage with service robots, automatic opening of room doors, or push of medication records are triggered.

[0054] The intervention effect feedback module continuously collects environmental and physiological data after the smart device performs the operation, and adjusts the device parameters or marks the intervention plan based on the data feedback.

[0055] Effect backtracking layer, used for:

[0056] The rehabilitation effectiveness tracking module generates rehabilitation reports to evaluate the rehabilitation plan;

[0057] Through the solution iteration and optimization module, data weights and intervention measures are adjusted based on effect tracking data and user feedback, and the individual basic parameter adaptation library and risk prediction model are updated.

[0058] Specifically, in practical applications of this system, the data acquisition layer deploys millimeter-wave radar and far-infrared thermal imaging equipment through a non-contact vital sign acquisition module. These devices are placed in frequently used areas such as bedrooms and living rooms for elderly individuals and those undergoing rehabilitation. This allows for the collection of vital sign data such as respiratory rate and heart rate variability without user intervention, while also capturing behaviors such as falls and prolonged sitting. The system is further enhanced by distributed sensors in the environmental and equipment data acquisition module, which collects indoor temperature and humidity data and connects to medical devices such as blood pressure monitors to record usage frequency and measurement data. Combined with a flexible wearable device in the multimodal physiological and basic data acquisition module, the system acquires data such as sleep structure and obtains emotional states through voice interaction. Caregivers or medical staff input basic diseases and rehabilitation goals. Simultaneously, the system interfaces with hospital systems to obtain recent medical examination reports, integrating these data into the fused dataset of the multimodal physiological and basic data acquisition module. This addresses the problem of incomplete data collection in existing systems due to user resistance. The intelligent processing layer first fuses, cleans, and removes redundant and interfering information from various types of data. Then, through the individual basic parameter adaptation module, it adjusts the weights of different data based on the user's underlying diseases, such as hypertension and diabetes. Addressing the complexity of mapping library management due to the large number of disease types, it constructs a primary classification library based on human body systems, such as cardiovascular, metabolic, and musculoskeletal systems. It then refines this into secondary sub-libraries based on specific diseases, such as hypertension grade 1 / 2 and type 1 / 2 diabetes. Simultaneously, it supports medical staff in real-time supplementation of disease weights based on clinical experience, forming a dynamic update mechanism. Based on rehabilitation goals, such as postoperative limb function recovery, it sets phased data thresholds and dynamically adjusts assessment standards, avoiding the drawbacks of the uniform standards in existing systems. The risk prediction model module combines multi-dimensional data to calculate the probability of health risks using formulas. ,in This represents the probability of health risk, with a value ranging from 0 to 1. , , , These represent the weights corresponding to the rate of change in equipment usage frequency, fluctuation values ​​of environmental parameters, deviation of time-series physiological indicators, and the influence coefficient of underlying diseases, respectively. ; This represents the rate of change in equipment usage frequency, which is the ratio of the difference between the usage frequency in the current period and the usage frequency in the previous period to the usage frequency in the previous period. It needs to be normalized to the range of 0 to 1. This represents the fluctuation value of environmental parameters, that is, the proportion of the difference between the actual value of the environmental parameter and the upper or lower limit of the suitable range to the suitable range, normalized to the interval of 0 to 1. It represents the deviation of physiological indicator time series data, that is, the proportion of the difference between the current physiological indicator and the historical average level to the historical fluctuation range, normalized to the range of 0 to 1. The underlying disease impact coefficient is set based on the type and severity of the underlying disease, with a value ranging from 0.1 to 0.5. To address the issue of insufficient model training data, a hybrid training set can be constructed by integrating publicly available medical datasets and system-tested data. An incremental learning algorithm is employed, fine-tuning the model every 30 days of new data accumulation to avoid overfitting. The initial values ​​of α, β, γ, and δ in the formula are set with reference to clinical guidelines, such as the weights of hypertension risk factors, and then optimized through model iteration. This formula can accurately output the probability of health risks and risk triggers, enabling early identification of potential risks. The dynamic weight and correlation analysis module, combined with the rehabilitation stage (such as the early postoperative period) and underlying disease, adjusts data weights. When abnormal equipment usage frequency occurs, it analyzes the number of abnormal postures, heart rate variability, and emotional state during the same period to pinpoint the cause. The assessment guidance layer generates a health score through a tiered assessment module, using a formula... ,in Indicates health score; The basic standard score is derived from general health standards. Key indicators from the physical examination report, such as blood routine, liver and kidney function, and imaging examination results, are included in the calculation to make the basic standard score more consistent with medical diagnostic criteria. This represents a weighting coefficient adjusted based on the individual's underlying disease and rehabilitation stage, with a value ranging from 0.8 to 1.2. This represents an individual correction value that combines risk prediction results with real-time data and abnormal items in the physical examination report (such as high blood sugar, abnormal blood lipids, etc.). The value ranges from -10 to 10. Ensure that before calculation... and With consistent dimensions, any discrepancies are normalized first, making the health score more closely reflect individual circumstances and triggering tiered early warnings. The personalized guidance and rehabilitation plan generation module combines environmental adjustment, physiological intervention, and rehabilitation training to generate combined plans. Plan development references medical advice from physical examination reports, such as avoiding weight-bearing after orthopedic surgery and dietary restrictions for diabetes, bridging the gap between lifestyle guidance and hospital diagnosis scenarios, strengthening the health guidance attribute, and presenting it in multiple formats for easy understanding by users with different cognitive abilities. The device linkage layer uses an intelligent device control module to link various intelligent devices according to the plan, enabling real-time intervention and solving the problem of delayed intervention in existing systems. The multi-scenario emergency response submodule triggers emergency responses when anomalies are detected, ensuring user safety. The intervention effect feedback module continuously collects data, adjusting device parameters or marking the plan. The effect tracking layer generates rehabilitation reports through the rehabilitation effect tracking module. The reports simultaneously compare the changing trends of key indicators from previous physical examination reports, intuitively presenting the improvement effect of the rehabilitation plan on health status. The plan iteration and optimization module updates the system based on data and feedback, achieving long-term personalized health management and improving the system's applicability and effectiveness.

[0059] Furthermore, to address device interface compatibility issues, the system adds a "protocol adaptation layer" to support parsing mainstream IoT protocols such as MQTT and CoAP. Devices using non-standard protocols can manually enter their data formats through a configuration wizard. To reduce data latency, edge computing technology is adopted, deploying core logic such as fall detection and abnormal heart rate determination on the local gateway. The emergency response submodule is configured with priority scheduling, prioritizing computing power for emergency events such as falls and sudden drops in heart rate to ensure a response time of ≤10 seconds.

[0060] In this embodiment, in the data acquisition layer:

[0061] The non-contact vital sign acquisition module is used to deploy millimeter-wave radar and far-infrared thermal imaging equipment to collect vital sign data such as respiratory rate, heart rate variability, body posture, and gait characteristics without contact. At the same time, it can capture behaviors such as falls, prolonged sitting, and abnormal wandering, and upload the data in real time after encryption.

[0062] The environmental and equipment data acquisition module is used to collect data on temperature, air pressure, humidity, and PM2.5 concentration through distributed sensors. At the same time, it connects to medical devices such as blood pressure monitors, body fat scales, and blood glucose meters through an IoT module to record the usage frequency, measurement values, and online status of the medical devices, forming an equipment operation log.

[0063] Specifically, when deploying the non-contact vital sign acquisition module, millimeter-wave radar is installed in the center of the bedroom ceiling and above the sofa in the living room, covering the main activity areas. It acquires respiratory rate and heart rate variability by detecting subtle human movements, and analyzes body posture and gait characteristics using radar echoes. It also integrates simple gesture recognition, such as waving to confirm and clenching a fist to cancel, supporting quick responses to system prompts, such as "Start rehabilitation training?". A "one-click evaluation" button is added to the smart screen and bedside terminal, with options for "Satisfied," "Needs Adjustment," and "Uncomfortable," facilitating feedback on the solution experience. Far-infrared thermal imaging equipment is installed at the bedroom door and on the side of the living room passageway. It captures behaviors such as falls, prolonged sitting, and abnormal wandering through thermal images. The collected vital sign data and behavioral information are processed by an encryption algorithm and uploaded to the system backend in real time. The encryption algorithm uses existing symmetric encryption technology, which will not be elaborated here, ensuring data transmission security and preventing data loss due to user inconvenience or forgotten operations. The distributed sensors of the environmental and equipment data acquisition module are located in different indoor locations. Temperature, air pressure, and humidity sensors are placed on the living room balcony and near bedroom windows, while the PM2.5 concentration sensor is near the outdoor ventilation opening to ensure comprehensive environmental data collection. The module connects to medical devices such as blood pressure monitors, body fat scales, and blood glucose meters via an IoT module, using the existing MQTT IoT protocol for data exchange (details omitted here). This allows for real-time acquisition of device usage frequency, measurement values, and online status, automatically generating device operation logs. In Provide data support to improve the comprehensiveness and continuity of data collection.

[0064] Furthermore, to address the issues of high sensor dependence and low fault tolerance, a multi-sensor complementary verification method is adopted: for example, temperature data is combined with air sensor and flexible wristband skin temperature for cross-verification. When a sensor fails, an approximate value can be calculated through the season and indoor-outdoor temperature difference. At the same time, a fault detection module is added to monitor the rationality of sensor data in real time. If the temperature changes suddenly beyond the physical range, a local alarm is triggered and the backup acquisition logic is automatically switched when a fault occurs.

[0065] Furthermore, the system adds a family member collaboration portal, allowing family members to remotely view simplified health reports such as daily scores and training completion status, and send voice reminders such as "Remember to measure your blood pressure." Monthly progress reports on rehabilitation are generated, including indicator trends, effect assessments, and adjustment suggestions, which are simultaneously pushed to users, family members, and medical staff to assist in adjusting data weights.

[0066] In this embodiment, the multimodal physiological and basic data acquisition module in the data acquisition layer is used to collect sleep structure, blood oxygen saturation, body temperature, and blood glucose trend data through a flexible wearable device. At the same time, it combines the voice interaction module to obtain emotional state, and allows guardians or medical staff to enter individual underlying diseases and rehabilitation goals to form a fusion dataset of physiological, emotional, and underlying diseases.

[0067] Specifically, the multimodal physiological and basic data acquisition module uses flexible wearable devices. A flexible wristband is worn on the user's wrist to collect sleep structure data, including deep sleep and light sleep duration, blood oxygen saturation, and body temperature. A flexible blood glucose monitoring vest conforms to the torso to collect blood glucose trend data. The devices are made of soft and breathable materials, so they do not interfere with the user's daily activities or rest. The voice interaction module is integrated into an indoor smart speaker or smart screen, using existing voice recognition and natural language processing technologies (details omitted here). It obtains the user's emotional state through daily conversation and supports users in actively providing feedback on their physical sensations. Individual underlying diseases and rehabilitation goals are entered through a mobile or desktop terminal connected to the system. After logging into the system, guardians or medical staff select the corresponding user and enter relevant information item by item, forming a fusion dataset. This dataset is a formula. In and formula In It provides basic parameters, offering a comprehensive data foundation for subsequent personalized assessments and solution generation, making the system more tailored to individual user needs.

[0068] In this embodiment, the individual basic parameter adaptation module in the intelligent processing layer is used to construct a mapping library between basic diseases and data weights, and adjust the weights of various types of data according to the individual's basic diseases; at the same time, it sets phased data thresholds according to rehabilitation goals and dynamically adjusts the evaluation criteria of each indicator.

[0069] Specifically, the individual baseline parameter adaptation module first constructs a mapping library between baseline diseases and data weights. Weights are assigned according to the baseline disease category: for cardiovascular diseases, heart rate variability is weighted at 0.3 and blood pressure at 0.25; for metabolic diseases, blood glucose is weighted at 0.3 and body fat at 0.25; and the total weight for other data is 0.45. After the system accesses user information, it retrieves the corresponding weight configuration from the mapping library based on the user's baseline disease and adjusts the evaluation weights of each data point. These weights directly affect the formulas. In Values ​​are assigned. Simultaneously, phased data thresholds are set based on the user's rehabilitation goals. For example, in the recovery of language function after stroke, the threshold for the percentage of clearly expressed sentences in the first stage of speech interaction is set at 40%, the second stage at 60%, and the third stage at 80%. The evaluation standards for each indicator are dynamically adjusted as the rehabilitation progresses to avoid evaluation bias caused by uniform standards and improve the accuracy and relevance of health assessments.

[0070] In this embodiment, the risk prediction model module in the intelligent processing layer is used to construct a prediction model based on long short-term memory network and attention mechanism. The prediction model receives the frequency change rate, environmental parameter fluctuation value, physiological indicator time series data and basic disease type as input, and outputs the health risk probability and risk triggers. The prediction model is continuously iterated and optimized through user historical data.

[0071] Specifically, the risk prediction model module constructs a prediction model based on a long short-term memory network and an attention mechanism. The long short-term memory network is an existing deep learning model, which will not be elaborated on here, and is used to process time-series data such as the rate of change in device usage frequency, fluctuation values ​​of environmental parameters, and time-series data of physiological indicators. The attention mechanism is used to dynamically adjust the formula. In , , , When the model runs, first... , , , Standardization is performed using the existing Min-Max normalization method, which will not be elaborated on here. The goal is to ensure that the parameters are within the range of 0 to 1. The processed data is then input into a Long Short-Term Memory (LSTM) network to extract temporal features, such as the three-day trend of physiological indicators. Subsequently, an attention mechanism is used to adjust the data based on the user's recent health status and underlying disease type. , , , For example, when a diabetic patient experiences significant recent blood sugar fluctuations, corresponding Increase to 0.4. , , Set the values ​​to 0.2, 0.2, and 0.2 respectively, and then substitute them into the formula. Calculate the risk probability and locate the triggers by combining the contribution of parameters, such as If the percentage exceeds 40%, the underlying cause may be abnormal device usage frequency. The model is periodically iterated and optimized using historical user data. , , , The initial value range improves the accuracy and timeliness of risk prediction.

[0072] In this embodiment, the intelligent processing layer includes a dynamic weighting and correlation analysis module, which is used to adjust data weights in a dual manner, combining the rehabilitation stage with the individual's underlying disease. When the device usage frequency is abnormal, it is linked to analyze the number of abnormal postures, heart rate variability, and emotional state during the same period to pinpoint the causes such as decreased operational ability, forgetfulness of use, or resistance to use.

[0073] Specifically, when the dynamic weighting and correlation analysis module is running, it adjusts data weights based on both the user's rehabilitation stage and underlying disease. In the early stages of post-fracture rehabilitation, gait characteristic data is weighted at 0.3 and body posture data at 0.25. In the mid-stage of rehabilitation, rehabilitation equipment usage data is weighted at 0.3 and gait characteristic data at 0.25. If the user has osteoporosis, an additional weight of 0.1 is added to the skeletal-related physiological indicators. When abnormal device usage frequency is detected, such as a blood pressure monitor not being used for three consecutive days, the module retrieves data on the number of abnormal postures, heart rate variability, and emotional state during the same period. Correlation analysis is performed using the formula... middle The values ​​and corresponding Weighting, if the number of abnormal poses increases by more than 50% compared to previous periods and A score of 0.8 could indicate decreased operational ability leading to reduced equipment usage; if the emotional state is characterized by irritability and resistance... A value of 0.7 may be considered an indication of conflicting use; if there are no obvious abnormalities and A score of 0.6 may indicate forgotten use. Accurately pinpointing the cause provides a basis for adjusting guidance or reminder methods, improving the system's sensitivity to changes in user behavior.

[0074] In this embodiment, in the evaluation guidance layer:

[0075] The grading and assessment module is used to combine risk prediction results, real-time data, individual underlying diseases and rehabilitation goals, and generate a health score using basic standards and individual correction values, and trigger graded early warnings based on the health score;

[0076] The multi-role early warning module is used to push reminder information or real-time data to at least one of the following based on the level of early warning: guardian, rehabilitation therapist, nursing home care station, community medical station, and emergency system. The information includes at least one of the following: risk trigger, real-time vital signs data, on-site video clips, indoor positioning information, and history of underlying diseases or recent medication records.

[0077] Specifically, the grading and assessment module uses the same formula when generating health scores. First, determine according to the industry's generally accepted health assessment standards. For example, 20 points are awarded for normal sleep structure, 15 points for adequate blood oxygen saturation, and 10 points for emotional stability. The score is 45; then adjusted according to the underlying medical conditions. Hypertension patients Adjust to 1.1 for diabetic patients and 1.05 for diabetic patients; combined with Sure , If it exceeds 0.6 Take -8, -3 for values ​​between 0.3 and 0.6, and 2 for values ​​less than 0.3. According to... Values ​​trigger tiered early warnings: values ​​above 80 indicate low risk, 60-80 indicate medium risk, and values ​​below 60 indicate high risk. A multi-role early warning module pushes information according to the warning level: low risk sends risk triggers and real-time vital signs data to guardians; medium risk sends on-site video and location information to rehabilitation therapists and nursing home care stations; and high risk sends underlying medical history and medication records to the emergency medical system, ensuring timely risk management.

[0078] In this embodiment, the personalized guidance and rehabilitation plan generation module in the assessment guidance layer is used to generate a combination plan of environmental adjustment, physiological intervention and rehabilitation training based on the assessment results and rehabilitation goals; the plan is presented through at least one of voice broadcast, smart screen animation demonstration or paper graphic manual.

[0079] Specifically, the personalized guidance and rehabilitation plan generation module generates a combination plan based on the assessment results and rehabilitation goals, referring to the formula. As a result, if If environmental parameters are below standard or too low, an environmental adjustment plan will be prioritized. Environmental adjustments will be based on current data; for example, if humidity is 65%, which is 40% to 60% above the suitable range, a dehumidifier will be run for 4 hours daily. Physiological interventions will be tailored to physiological indicators; for example, if blood oxygen saturation is 92%, which is below the standard of 95%, 10 minutes of abdominal breathing exercises will be implemented each morning and evening. Rehabilitation training will be tailored to the rehabilitation stage; for example, post-operative knee rehabilitation will include 3 sets of knee flexion and extension exercises daily. The plan will be presented through voice broadcasts, smart screen animations, and printed illustrated manuals to accommodate different user habits, ensure accurate understanding and implementation, and improve intervention effectiveness.

[0080] In this embodiment, in the device linkage layer:

[0081] The intelligent device control module is used to link at least one of the following via IoT protocol: dehumidifier, constant temperature mattress, smart meal box, rehabilitation equipment or home service robot, and automatically execute operations according to the guidance plan;

[0082] The multi-scenario emergency response submodule is used to automatically trigger indoor sound and light alarms and link service robots to check when a fall is detected, while sending location information and video clips to the guardian; when a sudden drop in heart rate is detected, it automatically opens the smart door lock and pushes medication records to the emergency system.

[0083] Specifically, the smart device control module connects to various smart devices via existing IoT protocols such as WiFi and Bluetooth, selecting the appropriate protocol based on the device type (e.g., WiFi for dehumidifiers, Bluetooth for rehabilitation equipment). After receiving the solution, it automatically sends control commands, generating a reference formula for the commands. The contribution of environmental and physiological parameters, such as If the environmental humidity deduction factor is high, the system will prioritize issuing a dehumidifier command to adjust the humidity to 50%. If rehabilitation training is the core focus, the system will issue a command to set the rehabilitation equipment to a medium speed for 15 minutes of training, simultaneously issuing a command to heat the meal in the smart meal box 30 minutes after training, thus automating equipment operation and improving convenience. The multi-scenario emergency response submodule detects falls, triggering an audible and visual alarm and linking with a service robot to check, pushing the location and video to the caregiver; when a sudden drop in heart rate is detected, the system automatically unlocks the smart door and pushes medication records (including formulas) to the emergency medical system. middle The corresponding disease types provide a reference for emergency response and improve emergency efficiency.

[0084] In this embodiment, the effect backtracking layer includes a scheme iteration optimization module, which is used to adjust the data weights and intervention measures for intervention schemes marked as to be optimized, by combining rehabilitation effect tracking data and user feedback, and to update the individual basic parameter adaptation library and risk prediction model according to long-term data trends.

[0085] Specifically, when the program iteration and optimization module selects intervention programs to be optimized, it does so based on the rehabilitation report. Trends, such as the implementation of the plan after 1 month If the improvement is less than 5%, it is marked as needing optimization. This should be combined with rehabilitation follow-up data, such as... Monthly changes and user feedback indicate high training difficulty; after analyzing the problem, adjust the plan accordingly. If improvement is slow and feedback is difficult, adjust the intensity of rehabilitation training and the formula accordingly. Weighting of rehabilitation-related parameters, such as the frequency of use of rehabilitation equipment. Adjust the formula by changing it from 0.2 to 0.25. Traditional Chinese medicine rehabilitation related indicators For example, the score might be increased from 15 to 20 points. This is based on long-term data trends, such as the impact of environmental parameters in different seasons. To mitigate the impact of changes in health and rehabilitation needs, the system is updated with individual basic parameter adaptation library weights and risk prediction model parameters to ensure it aligns with users' health changes and rehabilitation requirements, thereby enhancing its long-term applicability.

[0086] In summary, this invention addresses the problem of incomplete data collection caused by user resistance in existing systems through a non-contact vital sign acquisition module at the data acquisition layer, which collects vital sign and behavioral data without user intervention. This is combined with an environmental and equipment data acquisition module to collect environmental and medical equipment usage data, and a multimodal physiological and basic data acquisition module to integrate physiological, emotional, and underlying disease information. The intelligent processing layer adjusts data weights and assessment criteria according to underlying diseases and rehabilitation goals through an individual basic parameter adaptation module, and identifies health risks in advance using a risk prediction model module, avoiding the drawbacks of standardized approaches in existing systems. The assessment guidance layer generates personalized solutions in various forms, including voice broadcasts. The device linkage layer links intelligent devices to perform real-time intervention, addressing the problem of delayed intervention in existing systems. The effect tracking layer achieves long-term personalized health management through rehabilitation effect tracking and iterative optimization of solutions, compensating for the lack of continuity in existing systems and comprehensively improving the overall health management of the elderly and those undergoing rehabilitation.

[0087] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0088] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A smart health monitoring and guidance system for the elderly and those undergoing rehabilitation, characterized in that: include: The data acquisition layer is used to acquire non-contact vital sign data, environmental data, equipment usage frequency and measurement data, flexible wearable device data, voice interaction data, and individual underlying disease and rehabilitation goal data through non-contact vital sign acquisition module, environmental and equipment data acquisition module, and multimodal physiological and basic data acquisition module. The intelligent processing layer is used for: The acquired data is then merged and cleaned. Through the individual basic parameter adaptation module, the data weights and evaluation criteria are dynamically adjusted according to the individual's underlying diseases and rehabilitation goals; By using the risk prediction model module, combined with changes in equipment usage frequency, fluctuations in environmental parameters, time-series data of physiological indicators, and types of underlying diseases, health risks and risk triggers can be predicted. The dynamic weighting and correlation analysis module adjusts data weights based on the rehabilitation stage and individual underlying diseases, and analyzes the causes of abnormal equipment usage frequency. Assessment guidance layer, used for: The tiered assessment module combines risk prediction results, real-time data, physical examination reports, individual underlying diseases, and rehabilitation goals to generate a health score and trigger tiered early warnings. Through the personalized guidance and rehabilitation program generation module, a combination of environmental adjustment, physiological intervention and rehabilitation training is generated and presented in multiple forms such as voice broadcast, smart screen animation demonstration and paper-based graphic manual; The equipment linkage layer is used for: Through the intelligent device control module, intelligent devices such as dehumidifiers, thermostatic mattresses, smart meal boxes, rehabilitation equipment, and home service robots can automatically perform operations according to the guidance plan; Through the multi-scenario emergency response submodule, when abnormal situations such as falls or sudden drops in heart rate are detected, emergency responses such as indoor sound and light alarms, linkage with service robots, automatic opening of room doors, or push of medication records are triggered. The intervention effect feedback module continuously collects environmental and physiological data after the smart device performs the operation, and adjusts the device parameters or marks the intervention plan based on the data feedback. Effect backtracking layer, used for: The rehabilitation effectiveness tracking module generates rehabilitation reports to evaluate the rehabilitation plan; Through the solution iteration and optimization module, data weights and intervention measures are adjusted based on effect tracking data and user feedback, and the individual basic parameter adaptation library and risk prediction model are updated.

2. The intelligent health monitoring and guidance system for the elderly and those undergoing rehabilitation as described in claim 1, characterized in that, In the data acquisition layer: The non-contact vital sign acquisition module is used to deploy millimeter-wave radar and far-infrared thermal imaging equipment to collect vital sign data such as respiratory rate, heart rate variability, body posture, and gait characteristics without contact. At the same time, it can capture behaviors such as falls, prolonged sitting, and abnormal wandering, and upload the data in real time after encryption. The environmental and equipment data acquisition module is used to collect data on temperature, air pressure, humidity, and PM2.5 concentration through distributed sensors. At the same time, it connects to medical devices such as blood pressure monitors, body fat scales, and blood glucose meters through an IoT module to record the usage frequency, measurement values, and online status of the medical devices, forming an equipment operation log.

3. The intelligent health monitoring and guidance system for the elderly and those undergoing rehabilitation as described in claim 1, characterized in that, In the data acquisition layer, the multimodal physiological and basic data acquisition module is used to collect sleep structure, blood oxygen saturation, body temperature, and blood glucose trend data through flexible wearable devices. At the same time, it combines the voice interaction module to obtain emotional state, and allows guardians or medical staff to enter individual underlying diseases and rehabilitation goals to form a fusion dataset of physiological, emotional, and underlying diseases.

4. The intelligent health monitoring and guidance system for the elderly and those undergoing rehabilitation according to claim 1, characterized in that, In the intelligent processing layer, the individual basic parameter adaptation module is used to construct a mapping library between basic diseases and data weights, and adjust the weights of various types of data according to the individual's basic diseases; at the same time, it sets phased data thresholds according to rehabilitation goals and dynamically adjusts the evaluation criteria of each indicator.

5. The intelligent health monitoring and guidance system for the elderly and those undergoing rehabilitation as described in claim 1, characterized in that, In the intelligent processing layer, the risk prediction model module is used to construct a prediction model based on long short-term memory network and attention mechanism. The prediction model receiving device uses frequency change rate, environmental parameter fluctuation value, physiological indicator time series data and basic disease type as input, and outputs health risk probability and risk trigger. The predictive model is continuously optimized through iterative iterations using historical user data.

6. The intelligent health monitoring and guidance system for the elderly and those undergoing rehabilitation according to claim 1, characterized in that, In the intelligent processing layer, the dynamic weight and correlation analysis module is used to adjust the data weights in a dual manner, combining the rehabilitation stage and the individual's underlying disease. When the device usage frequency is abnormal, it is linked to analyze the number of abnormal postures, heart rate variability, and emotional state during the same period to locate the causes such as decreased operational ability, forgetting to use, or resistance to use.

7. The intelligent health monitoring and guidance system for the elderly and those undergoing rehabilitation according to claim 1, characterized in that, In the aforementioned assessment guidance layer: The grading and assessment module is used to combine risk prediction results, real-time data, individual underlying diseases and rehabilitation goals, and generate a health score using basic standards and individual correction values, and trigger graded early warnings based on the health score; The multi-role early warning module is used to push reminder information or real-time data to at least one of the following based on the level of early warning: guardian, rehabilitation therapist, nursing home care station, community medical station, and emergency system. The information includes at least one of the following: risk trigger, real-time vital signs data, on-site video clips, indoor positioning information, and history of underlying diseases or recent medication records.

8. The intelligent health monitoring and guidance system for the elderly and those undergoing rehabilitation according to claim 1, characterized in that, In the assessment and guidance layer, the personalized guidance and rehabilitation plan generation module is used to generate a combination plan of environmental adjustment, physiological intervention and rehabilitation training based on the assessment results and rehabilitation goals; the plan is presented through at least one of voice broadcast, smart screen animation demonstration or paper graphic manual.

9. The intelligent health monitoring and guidance system for the elderly and those undergoing rehabilitation according to claim 1, characterized in that, In the device linkage layer: The intelligent device control module is used to link at least one of the following via IoT protocol: dehumidifier, constant temperature mattress, smart meal box, rehabilitation equipment or home service robot, and automatically execute operations according to the guidance plan; The multi-scenario emergency response submodule is used to automatically trigger indoor sound and light alarms and link service robots to check when a fall is detected, while sending location information and video clips to the guardian; when a sudden drop in heart rate is detected, it automatically opens the smart door lock and pushes medication records to the emergency system.

10. The intelligent health monitoring and guidance system for the elderly and those undergoing rehabilitation according to claim 1, characterized in that, In the effect backtracking layer, the scheme iteration and optimization module is used to adjust the data weights and intervention measures for intervention schemes marked as to be optimized, in combination with rehabilitation effect tracking data and user feedback, and update the individual basic parameter adaptation library and risk prediction model according to long-term data trends.

Images