A medical care monitoring system

Abstract

The application provides a medical care monitoring system, comprising: a data acquisition module for acquiring health data of a user from multiple heterogeneous data sources; a data preprocessing and fusion module for cleaning, standardizing and fusing the health data to generate a unified health data file; an intelligent analysis engine module for analyzing the health data file, the analysis including identifying health patterns and calculating a dynamic health risk index through a machine learning model; an early warning and intervention module for generating early warning information and personalized health intervention recommendations based on the dynamic health risk index; a visual interactive platform for displaying health data, risk index and intervention recommendations. The user compliance and health outcomes are effectively improved, and a complete closed loop of "monitoring-analysis-early warning-intervention-feedback" is formed.

Description

Technical Field

[0001] This invention relates to the medical field, specifically to a medical care monitoring system. Background Technology

[0002] With an aging population and rising incidence of chronic diseases, personal health monitoring and chronic disease management are becoming increasingly important. Existing medical and nursing monitoring technologies suffer from the following limitations: Data silos: While various health monitoring devices (such as smart bracelets, blood glucose meters, and blood pressure monitors) and health management apps exist, their data is often scattered across different platforms, making effective integration difficult and hindering the formation of a comprehensive and coherent view of a user's health status; Insufficient early warning capabilities: Most systems only provide simple threshold alarms (such as alerts when blood pressure exceeds a fixed value), lacking intelligent analysis and proactive early warning based on multi-dimensional data, historical trends, and individual differences. High false alarm and false negative rates prevent effective support for medical decision-making; Passive monitoring: Existing systems typically require users to actively upload data or passively receive simple reminders, lacking proactive, interactive health guidance and intervention mechanisms, resulting in low user engagement. Summary of the Invention

[0003] The problem this invention aims to solve is to provide a medical care monitoring system that addresses the aforementioned shortcomings in the existing technology, enabling comprehensive, continuous, and accurate monitoring and early warning of an individual's health status.

[0004] To achieve the above objectives, the present invention adopts the following technical solution: a medical care monitoring system, comprising: a data acquisition module for acquiring user health data from multiple heterogeneous data sources; a data preprocessing and fusion module for cleaning, standardizing, and fusion of the health data to generate a unified health data archive; an intelligent analysis engine module for analyzing the health data archive, the analysis including identifying health patterns and calculating a dynamic health risk index through a machine learning model; an early warning and intervention module for generating early warning information and personalized health intervention suggestions based on the dynamic health risk index; and a visualization and interactive platform for displaying health data, risk index, and intervention suggestions.

[0005] Furthermore, the heterogeneous data sources include at least two of the following: smart wearable devices, home medical devices, user manual input terminals, and third-party medical information systems; the data preprocessing and fusion module performs timestamp alignment operations to generate time-series health data archives; the machine learning model is a personalized model that is continuously adaptively optimized based on individual user historical data; the dynamic health risk index is a multi-dimensional indicator calculated by comprehensively considering real-time data, historical trend data, and prediction model results; and the early warning information is sent to the user, the user's family members, or designated medical personnel according to the risk level.

[0006] Furthermore, the health intervention recommendations include one or more of the following: medication reminders, dietary advice, exercise programs, or medical guidance; the visual interactive platform provides health data trend charts, risk radar charts, and an intervention program tracking interface.

[0007] Furthermore, the heterogeneous data sources acquired by the data acquisition module include: a) smart wearable devices for collecting heart rate, blood oxygen, body temperature, activity level, and sleep data; b) home medical devices for collecting blood pressure, blood glucose, and electrocardiogram data; c) user terminals for receiving manually input data on symptoms, medication records, dietary information, and subjective feelings from users; and d) third-party medical information systems for acquiring users' electronic medical records, historical diagnoses, and laboratory data through authorized interfaces.

[0008] Furthermore, the data acquisition module also includes environmental sensors for collecting temperature, humidity, and air quality data of the user's environment; the intelligent analysis engine module uses environmental data as an analysis dimension when calculating the dynamic health risk index; the data preprocessing and fusion module performs the following operations: interpolating or removing missing data; identifying and smoothing abnormal data; aligning and resampling data from different frequencies and formats using timestamps; and binding the processed data with the user's identity identifier and storing it in a unified health data archive.

[0009] Furthermore, the intelligent analysis engine module includes: a short-term anomaly detection unit, used to detect in real time whether physiological parameters exceed the dynamic threshold range calculated based on personal historical data using statistical process control methods; a long-term trend prediction unit, used to predict the development trend of specific health indicators in the future time period using a time-series prediction model; and a multimodal fusion analysis unit, used to integrate physiological data, behavioral data, and environmental data, and calculate the comprehensive dynamic health risk index through a fusion algorithm. The machine learning model is an ensemble learning model, whose input features include current real-time data, statistical features of historical data, time-series features, and user profile data. The model can perform incremental learning based on newly added user data to continuously optimize prediction accuracy. The intelligent analysis engine module also includes a model interpretability unit, used to generate a report of the main contributing factors of the dynamic health risk index and describe the key reasons for triggering the warning in natural language. The operations performed by the warning and intervention module include: presetting multiple risk level thresholds to map the dynamic health risk index to different warning levels; and defining different notification strategies for each warning level, including notification recipients, notification channels, and notification urgency. The personalized health intervention recommendations are generated by a knowledge graph-based inference engine, which matches the user's current health status with a medical knowledge base and outputs a structured intervention plan. For intermediate alerts, the notification strategy targets the user and one emergency contact person, and the notification channels are APP push and SMS. For advanced alerts, the notification targets additionally include designated medical personnel, and the notification channel includes an automated voice call.

[0010] Furthermore, the intervention plan includes one or more of the following: medication adjustment suggestions, lifestyle guidance, rehabilitation training plans, or emergency medical treatment guidance. It is synchronized with the user's calendar application to generate specific execution reminders. The visual interactive platform provides different view ports: Patient View: Displays concise health scores, trend charts, and pending intervention tasks in the form of daily or weekly reports; Healthcare View: Simultaneously monitors risk index rankings, warning lists, and health record details for multiple patients in a dashboard format, and provides data export functionality. The Healthcare View allows healthcare professionals to directly send customized health guidance information or modify intervention plans to the patient view of a specific patient through the platform. The system also includes a privacy and security module for: end-to-end encrypted transmission and storage of all health data; implementing role-based access control to finely manage the viewing and operation permissions of different roles such as users, family members, and healthcare professionals; and providing data anonymization processing functionality for use in medical research.

[0011] The data acquisition module includes a basic data acquisition submodule and a risk-triggered data acquisition submodule. The basic data acquisition submodule continuously collects the user's routine health data at a preset first acquisition frequency. The risk-triggered data acquisition submodule is activated by the early warning and intervention module when the intelligent analysis engine module detects that the dynamic health risk index reaches or exceeds a preset risk concern threshold, and collects specific health data related to the risk type at a preset second acquisition frequency. The second acquisition frequency is higher than the first acquisition frequency. The early warning and intervention module matches and executes a corresponding target alarm strategy from multiple preset alarm strategies based on the risk level and risk type of the dynamic health risk index. Different alarm strategies include at least one of different alarm receiving objects, alarm channel combinations, or alarm information detail levels. The early warning and intervention module also includes a strategy dynamic adjustment unit, which adjusts the alarm channel combination and alarm information detail level in the target alarm strategy based on the user's feedback behavior data on historical intervention suggestions.

[0012] The early warning and intervention module categorizes the health intervention suggestions into three levels: primary, intermediate, and advanced. Primary intervention suggestions are non-mandatory suggestions containing only health tips, passively displayed through the visual interactive platform when the dynamic health risk index is at a low risk level. Intermediate intervention suggestions are mandatory suggestions containing specific action guidelines, actively pushed through the target alarm strategy when the dynamic health risk index is at a medium risk level. Advanced intervention suggestions are emergency suggestions containing emergency contact information and first aid guidance, actively pushed through the target alarm strategy when the dynamic health risk index is at a high risk level, while simultaneously interrupting the current display interface of the visual interactive platform and displaying the advanced intervention suggestion at the top.

[0013] The basic data acquisition submodule and the risk-triggered data acquisition submodule have differentiated configurations in terms of data acquisition objects; the basic data acquisition submodule is connected to a first heterogeneous data source group, which includes smart wearable devices and user manual input terminals; the risk-triggered data acquisition submodule is connected to a second heterogeneous data source group, which includes home medical devices and third-party medical information systems; when the risk-triggered data acquisition submodule is activated, the intelligent analysis engine module acquires refined data from the second heterogeneous data source group and combines it with data from the first heterogeneous data source group to jointly update the dynamic health risk index; the early warning and intervention module dynamically adjusts the classification level of the executed target alarm strategy and the health intervention recommendation based on the updated dynamic health risk index.

[0014] Compared with existing technologies, this invention has the following beneficial effects: It effectively solves the problem of data silos, constructs a multi-dimensional personal health digital twin, lays the foundation for accurate analysis, and utilizes a multi-level AI analysis model (anomaly detection, trend prediction, multimodal fusion) to achieve a leap from passive alarm to proactive early warning. The accuracy and clinical value of early warning are significantly improved. The system learns each user's health baseline and patterns, and the risk assessment and intervention plans provided are highly targeted, effectively improving user compliance and health outcomes. It forms a complete closed loop of "monitoring-analysis-early warning-intervention-feedback", combining technology with humanistic care, transforming passive into proactive care, and truly realizing full-process and extended medical care.

[0015] Other advantages, objectives and features of the present invention will become apparent in part from the following description, and in part from those skilled in the art through study and practice of the invention. Attached Figure Description

[0016] Figure 1 This is a system architecture diagram of a medical care monitoring system.

[0017] Figure 2 Flowchart for a health data processing and early warning system.

[0018] Figure 3 This is a flowchart of the core modules of the health management system.

[0019] Figure 4 This is a second flowchart of a medical care monitoring system. Detailed Implementation

[0020] To make the technical means, creative features, objectives, and effects of this invention clearer and easier to understand, the invention will be further described below in conjunction with the accompanying drawings and specific embodiments:

[0021] This invention proposes a medical care monitoring system, comprising: an interconnected data acquisition module 100, a data preprocessing and fusion module 200, an intelligent analysis engine module 300, an early warning and intervention module 400, and a visualization and interaction platform 500; the data acquisition module 100 is used to acquire user health data from multiple heterogeneous data sources; the data preprocessing and fusion module 200 is used to clean, standardize, and fuse the health data to generate a unified health data archive; the intelligent analysis engine module 300 is used to analyze the health data archive, the analysis including identifying health patterns and calculating a dynamic health risk index through a machine learning model 310; the early warning and intervention module 400 is used to generate early warning information and personalized health intervention suggestions based on the dynamic health risk index; and the visualization and interaction platform 500 is used to display health data, risk indices, and intervention suggestions.

[0022] The data acquisition module 100 is used to acquire the user's health data from multiple heterogeneous data sources. These data sources include smart wearable devices (collecting heart rate, blood oxygen, body temperature, activity level, and sleep data), home medical devices (collecting blood pressure, blood sugar, and ECG data), user terminals (for manually inputting symptoms, medication records, dietary information, and subjective feelings), and third-party medical information systems (obtaining electronic medical records, historical diagnoses, and laboratory data through authorized interfaces). The data acquisition module 100 can also integrate environmental sensors to collect temperature, humidity, and air quality data of the user's environment.

[0023] The data preprocessing and fusion module 200, connected to the data acquisition module 100, is used to clean (handle missing values, identify and correct outliers), standardize the format, align timestamps, and resample the acquired multi-source heterogeneous data, and fuse it to generate a unified health data archive based on the user and time sequence. The data preprocessing and fusion module aligns, cleans, and integrates all the above data according to the timeline. For example, it generates a fused record: {Date: 2023-10-27, Time: 08:30, Blood glucose: 8.5 mmol / L, Blood pressure: 150 / 95 mmHg, Medication: Metformin (taken), Diet: High carbohydrates, Time: 14:00, Heart rate (resting): 90 bpm, Ambient temperature: 26°C}.

[0024] The intelligent analysis engine kicks in: the short-term anomaly detection unit detects that the patient's resting heart rate has been consistently higher than their personal baseline (75 bpm) by more than 15% recently, triggering attention; the long-term trend prediction unit analyzes the patient's blood glucose data over the past 3 months, predicting a continued upward trend in fasting blood glucose over the next week; the multimodal fusion analysis unit integrates multiple factors such as "currently high blood glucose and blood pressure values," "recently elevated HbA1c," "abnormally high resting heart rate," "high-carbohydrate diet records," and "potential cardiovascular stress from higher ambient temperatures," and calculates a "high-risk" comprehensive risk index for the patient to experience a hyperglycemic crisis or cardiovascular discomfort within the next 72 hours using a machine learning model; the model interpretability unit generates a report indicating that "abnormally high heart rate, poor blood glucose control, and dietary factors are the main risk contributing factors."

[0025] The intelligent analysis engine module 300, connected to the data preprocessing and fusion module 200, is used for in-depth analysis of the fused health data archive. It includes: a short-term anomaly detection unit, which uses statistical process control (SPC) to detect in real-time whether physiological parameters exceed dynamic threshold ranges calculated based on personal historical data; a long-term trend prediction unit, which uses time-series prediction models such as ARIMA and LSTM to predict the future trend of specific health indicators (such as blood sugar and blood pressure); and a multimodal fusion analysis unit, which integrates physiological, behavioral, and environmental data and calculates a comprehensive dynamic health risk index using machine learning fusion algorithms (such as random forest and gradient boosting tree). The machine learning model can incrementally learn based on newly added user data to continuously optimize prediction accuracy. This module also includes a model interpretability unit, which generates a report of the main contributing factors of the risk index and describes the key reasons triggering the warning in natural language.

[0026] The early warning and intervention module 400 is connected to the intelligent analysis engine module 300 and is used to generate early warning information of different levels based on the dynamic health risk index. The early warning information is sent to the user, family members or designated medical staff through user terminals, SMS, telephone and other means according to the preset strategy (notification object, notification channel, urgency level). Meanwhile, this module automatically generates and pushes personalized health intervention suggestions (such as medication adjustments, lifestyle guidance, rehabilitation training, and medical appointment guidance) through a knowledge graph-based reasoning engine, and can synchronize them to the user's calendar application. The early warning and intervention module activates preset notification strategies based on the "high-risk" level: immediately sending a strong reminder pop-up alert to the patient's mobile app, simultaneously sending an alert SMS to the patient's son's mobile phone, and sending an alert work order to the nurses' workstation at the community hospital's diabetes management center. The system calls the knowledge graph to generate personalized intervention suggestions: "It is recommended to immediately retest blood glucose and blood pressure." "Remind you to have a follow-up appointment tomorrow; the system has reserved an appointment for you." "A low-GI (glycemic index) diet plan for the next three days has been generated; please check it." "Today's suggestion is to avoid activities in high-temperature environments and engage in light exercise during cooler periods." These suggestions are pushed to the patient's "Today's Tasks" section in the app and synchronized to their calendar.

[0027] The visualization and interactive platform 500 provides a graphical health data display interface for users (patient view) and medical staff (medical staff view). This interface can display health trend charts, risk radar charts, warning history, and intervention plan implementation status. The medical staff view also supports monitoring multiple patients, sending customized guidance information, and data export. Patients can see clear overlay charts of blood glucose, blood pressure, and heart rate trends on their app, and can intuitively see high-risk warnings and specific action lists. The community nurse saw the patient's warning information at the top of the list on the computer-based medical staff view. She clicked to view the patient details and saw the risk contribution report generated by the system and recent complete data records. She acknowledged the system's judgment and sent a customized message to the patient's app through the platform: "Mr. Wang, the system has detected abnormal indicators in your recent indicators. Please be sure to follow the advice for a follow-up visit. If you feel unwell, please call emergency services immediately."

[0028] The heterogeneous data sources include at least two of the following: smart wearable devices 110, home medical devices 120, user manual input terminals 130, and third-party medical information systems 140; the data preprocessing and fusion module 200 performs timestamp alignment to generate time-series health data archives; the machine learning model 310 is a personalized model that continuously adapts and optimizes based on individual user historical data; the dynamic health risk index is a multi-dimensional indicator calculated by integrating real-time data, historical trend data, and the results of the prediction model 320; the warning information is sent to the user, the user's family members, or designated medical personnel according to the risk level.

[0029] The health intervention recommendations include one or more of the following: medication reminders 410, dietary recommendations 420, exercise programs 430, or medical guidance 440; the visualization and interaction platform 500 provides a health data trend chart 510, a risk radar chart 520, and an intervention program tracking interface 530; the heterogeneous data sources acquired by the data acquisition module 100 include: a) a smart wearable device 110, used to collect heart rate, blood oxygen, body temperature, activity level, and sleep data; b) a home medical device 120, used to collect blood pressure, blood glucose, and electrocardiogram data; c) a user terminal 130, used to receive manually input symptom, medication records, dietary information, and subjective feeling data from the user; d) a third-party medical information system 140, used to obtain the user's electronic medical records, historical diagnoses, and laboratory data through an authorized interface 141; the smartwatch worn by the patient continuously collects their daytime heart rate, nighttime sleep quality, and daily steps. After breakfast each day, patients use a smart blood glucose meter and a smart blood pressure monitor to measure and automatically upload their blood glucose and blood pressure data. Patients also use a mobile app to record their dietary overview for each meal (such as "high carbohydrate" or "greasy") and their daily medication. With the patient's authorization, the system regularly retrieves the latest blood lipid and glycated hemoglobin (HbA1c) test results from the community hospital's HIS system. Environmental sensors in the smart home report the temperature and humidity data of the patient's room.

[0030] The data acquisition module also includes environmental sensors for collecting temperature, humidity, and air quality data of the user's environment; the intelligent analysis engine module uses environmental data as an analysis dimension when calculating the dynamic health risk index; the data preprocessing and fusion module performs the following operations: interpolating or removing missing data; identifying and smoothing abnormal data; aligning and resampling data from different frequencies and formats using timestamps; and binding the processed data with the user's identity identifier and storing it in a unified health data archive.

[0031] The intelligent analysis engine module 300 includes: a short-term anomaly detection unit 330, used to detect in real time whether physiological parameters exceed the dynamic threshold range calculated based on personal historical data using statistical process control methods; a long-term trend prediction unit 340, used to predict the development trend of specific health indicators in the future using a time-series prediction model; a multimodal fusion analysis unit 350, used to integrate physiological data, behavioral data, and environmental data, and calculate the comprehensive dynamic health risk index through a fusion algorithm; the machine learning model 310 is an ensemble learning model, whose input features include current real-time data, statistical features of historical data, time-series features, and user profile data 360; the model can perform incremental learning 370 based on newly added user data to continuously optimize prediction accuracy; the intelligent analysis engine module 300 also includes a model interpretability unit 380, used to generate a report of the main contributing factors of the dynamic health risk index and describe the key reasons for triggering the warning in natural language; the operations performed by the warning and intervention module 400 include: preset multiple risk level thresholds 450, mapping the dynamic health risk index to different warning levels; defining different notification strategies 460 for each warning level, including notification recipients, notification channels, and notification urgency; the personalized health intervention suggestions are generated by a knowledge graph-based inference engine 470, which matches the user's current health status with a medical knowledge base 480 and outputs a structured intervention plan; for intermediate warnings, the notification strategy 460 includes the user and one emergency contact person as the notification recipients, and the notification channels are APP push 461 and SMS 462; for advanced warnings, the notification recipients are additionally designated medical personnel, and the notification channels are additional automated voice calls 463; the short-term abnormality detection unit 330 uses the statistical process control (SPC) method to calculate the dynamic threshold range based on the user's personal historical data and detect whether physiological parameters are abnormal in real time. For example, the recent mean and standard deviation are calculated using a sliding window, and the upper and lower limits of the threshold are dynamically adjusted; the long-term trend prediction unit 340 uses a time-series prediction model (such as ARIMA, LSTM, or Prophet) to predict specific health indicators (such as blood glucose trends, weight changes) for a future period (such as 7 days or 30 days), and outputs trend charts and confidence intervals; the multimodal fusion analysis unit 350 integrates physiological, behavioral, and environmental data, and calculates a comprehensive dynamic health risk index (DHRI) through feature-level and decision-level fusion algorithms (such as weighted average, neural network, or Bayesian inference).The index ranges from 0 to 100, with higher values ​​indicating greater risk. Machine learning model 310 employs an ensemble learning model (such as XGBoost or Random Forest), with input features including real-time data, statistical features of historical data (such as mean, variance, and extreme values), time-series features (such as sliding window trends and periodic indicators), and user profile data 360 (such as age, gender, and medical history). The model supports incremental learning 370, enabling online parameter updates based on new data to continuously optimize prediction accuracy. Model interpretability unit 380 uses interpretability technologies such as SHAP or LIME to generate a report on the main contributing factors of DHRI (e.g., "This increase in the risk index is mainly due to abnormal nighttime heart rate and insufficient sleep"), and describes the key reasons triggering the warning in natural language.

[0032] The intervention plan includes one or more of the following: medication adjustment suggestions 411, lifestyle guidance 421, rehabilitation training plan 431, or emergency medical treatment guidance 441, and is synchronized with the user's calendar application 490 to generate specific execution reminders; the visualization interaction platform 500 provides different view ports: patient view 540: displays concise health scores, trend charts, and pending intervention tasks in the form of daily reports, weekly reports, etc.; medical staff view 550: monitors the risk index ranking, warning list, and health record details of multiple patients simultaneously in the form of a dashboard, and provides a data export function 551; the medical staff view 550 allows medical staff to directly send customized health guidance information or modify intervention plans to the patient view 540 of a specific patient through the platform; the system also includes a privacy and security module 600, used for: end-to-end encrypted transmission and storage of all health data 610; implementing role-based access control 620, finely managing the viewing and operation permissions of different roles such as users, family members, and medical staff; and providing data anonymization processing function 630 for use in medical research.

[0033] Risk Level Mapping 450: Presets multiple risk level thresholds (e.g., low risk: 0-30, medium risk: 30-70, high risk: 70-100), mapping DHRI to different warning levels. Notification Strategy 460: Defines differentiated notification mechanisms for each warning level: Intermediate Warning: Notifies the user and one emergency contact via APP push notification 461 and SMS 462; Advanced Warning: Additionally notifies designated medical personnel via automated voice call 463; Intervention Plan Generation: Through a knowledge graph-based inference engine 470, matches the user's current health status with a medical knowledge base 480 (including clinical guidelines, drug databases, rehabilitation plans, etc.), outputting structured intervention plans, including: Medication adjustment suggestions 411 (e.g., "increase the dose of antihypertensive medication once"); Lifestyle guidance 421 (e.g., "reduce sodium intake and walk 30 minutes daily"); Rehabilitation training plan 431 (e.g., "implement the third stage of the cardiopulmonary rehabilitation training plan"); Emergency medical treatment guidance 441 (e.g., "go to the cardiovascular emergency department immediately"). The intervention plan is synchronized with the user's calendar application 490, generating specific execution reminders (such as medication time, exercise plan); a visual interactive platform 500; a patient view 540: displaying health scores, trend charts, and pending intervention tasks in the form of daily and weekly reports, supporting one-click confirmation or postponement; a healthcare view 550: providing multi-patient monitoring dashboards, including: a risk index ranking (sorted by DHRI); an alert list (updated in real time for medium- and high-risk patients); health record details (viewing historical data, contribution factor reports, and intervention records); a data export function 551 (supporting CSV / PDF format export); supporting healthcare staff to send customized health guidance or modify intervention plans to specific patients; a privacy and security module 600; end-to-end encryption 610: performing AES-256 encryption on all health data transmission and storage; role-based access control (RBAC) 620: defining role permissions for users, family members, healthcare staff, etc. (e.g., only healthcare staff can modify intervention plans). Data Anonymization 630: After desensitization, it is used for medical research (such as removing identifiers like names and ID numbers); through dynamic thresholds, multimodal fusion, and knowledge graph reasoning, it achieves accurate quantification and tracing of health risks; through tiered early warning and personalized intervention, it improves user compliance and medical response efficiency; at the same time, through interpretable output and privacy protection mechanisms, it enhances the credibility and security of the system.

[0034] By dividing the data acquisition modules into basic and risk-triggered categories, the system only needs to collect routine data with low power consumption and low frequency during daily monitoring, greatly saving power and network bandwidth resources of terminal devices. High-frequency, high-precision specialized data acquisition is only initiated when a risk occurs, achieving "precise deployment" of monitoring resources and extending the battery life of wearable devices and other terminals. The system also offers precise and personalized alarms: different alarm strategies are matched according to risk level and type, avoiding information overload or omission of important information caused by a "one-size-fits-all" approach. In particular, the introduction of a dynamic strategy adjustment unit allows the alarm system to learn and optimize alarm methods based on past user feedback (e.g., users often ignore app push notifications but answer phone calls), significantly improving the effective reach rate of alarms and user compliance.

[0035] Enhancing User Experience and Psychological Comfort: Intervention suggestions are categorized into basic, intermediate, and advanced levels, each with different display and push methods. For low-risk situations, only gentle, non-instructional prompts are provided to avoid excessive interference with users, reducing anxiety and resistance to the system, making the system more "human-centered." Ensuring Timely Response in Emergency Situations: For high-risk emergency suggestions, the system features a "forced interruption" interface that is prominently displayed. This highly interactive approach ensures that in the most critical situations, users (or on-site personnel) can see the most crucial first-aid guidance and contact information immediately, unaffected by other interface elements, thus buying precious time for life-saving efforts.

[0036] Data tracing and refined analysis: It clarifies that the data source groups invoked differ under different risk levels. At low risk, the system relies solely on wearable devices and user input for lightweight judgment; however, when the risk increases, the system automatically activates and invokes more diagnostically valuable professional medical devices (such as blood glucose meters and blood pressure monitors) and historical medical record data. This makes the risk index updates more rigorous and clinically based, reducing the false alarm rate.

[0037] A dynamic closed-loop adjustment mechanism is formed: through a complete chain of risk-triggered data collection -> refined data updates to the risk index -> risk index-driven adjustments to strategies and intervention levels, the system forms a self-reinforcing dynamic closed loop. This makes the system not a static rule machine, but an intelligent agent that can "adaptively" adjust monitoring intensity and response strategies according to the user's real-time status, achieving a qualitative leap from "passive response" to "proactive, layered, and precise prevention and control."

[0038] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A medical care monitoring system, characterized in that, include: The system comprises an interconnected data acquisition module (100), a data preprocessing and fusion module (200), an intelligent analysis engine module (300), an early warning and intervention module (400), and a visualization and interaction platform (500); the data acquisition module (100) is used to acquire user health data from multiple heterogeneous data sources. The data preprocessing and fusion module (200) is used to clean, standardize and fuse the health data to generate a unified health data archive; the intelligent analysis engine module (300) is used to analyze the health data archive, the analysis including identifying health patterns and calculating dynamic health risk index through a machine learning model (310); The early warning and intervention module (400) is used to generate early warning information and personalized health intervention suggestions based on the dynamic health risk index; A visual interactive platform (500) is used to display health data, risk indices, and intervention recommendations.

2. The medical care monitoring system according to claim 1, characterized in that, The heterogeneous data sources include at least two of the following: smart wearable devices (110), home medical devices (120), user manual input terminals (130), and third-party medical information systems (140); the data preprocessing and fusion module (200) performs timestamp alignment operations to generate time-series health data archives; the machine learning model (310) is a personalized model that is continuously adaptively optimized based on the user's individual historical data; the dynamic health risk index is a multi-dimensional indicator calculated by integrating real-time data, historical trend data, and the results of the prediction model (320); the warning information is sent to the user, the user's family members, or designated medical personnel according to the risk level.

3. The medical care monitoring system according to claim 1, characterized in that, The health intervention recommendations include one or more of the following: medication reminders (410), dietary recommendations (420), exercise programs (430), or medical guidance (440); the visualization and interactive platform (500) provides a health data trend chart (510), a risk radar chart (520), and an intervention program tracking interface (530).

4. The medical care monitoring system according to claim 1, characterized in that: The heterogeneous data sources acquired by the data acquisition module (100) include: a) a smart wearable device (110) for collecting heart rate, blood oxygen, body temperature, activity level and sleep data; b) a home medical device (120) for collecting blood pressure, blood glucose and electrocardiogram data; c) a user terminal (130) for receiving symptoms, medication records, dietary information and subjective feelings data manually entered by the user; d) a third-party medical information system (140) for acquiring the user's electronic medical records, historical diagnoses and laboratory data through an authorized interface (141).

5. The medical care monitoring system according to claim 1, characterized in that, The data acquisition module also includes an environmental sensor for collecting temperature, humidity, and air quality data of the user's environment; the intelligent analysis engine module uses environmental data as an analysis dimension when calculating the dynamic health risk index. The data preprocessing and fusion module performs the following operations: interpolating or removing missing data; identifying and smoothing abnormal data; aligning and resampling data from different frequencies and formats using timestamps; and binding the processed data with user identifiers and storing it in a unified health data archive.

6. The medical care monitoring system according to claim 1, characterized in that, The intelligent analysis engine module (300) includes: a short-term anomaly detection unit (330), used to detect in real time whether physiological parameters exceed the dynamic threshold range calculated based on personal historical data using statistical process control methods; a long-term trend prediction unit (340), used to predict the development trend of specific health indicators in the future time period using a time-series prediction model; a multimodal fusion analysis unit (350), used to integrate physiological data, behavioral data and environmental data, and calculate the comprehensive dynamic health risk index through a fusion algorithm; the machine learning model (310) is an ensemble learning model, whose input features include the statistical features of current real-time data and historical data, time-series features and user profile data (360); the model can perform incremental learning (370) based on newly added user data to continuously optimize prediction accuracy; the intelligent analysis engine module ( 300) also includes a model interpretability unit (380) for generating a report of the main contributing factors of the dynamic health risk index and describing the key reasons for triggering the warning in natural language; the operation performed by the warning and intervention module (400) includes: preset multiple risk level thresholds (450) to map the dynamic health risk index to different warning levels; define different notification strategies (460) for each warning level, including notification recipients, notification channels and notification urgency; the personalized health intervention suggestions are generated by a knowledge graph-based inference engine (470), which matches the user's current health status with a medical knowledge base (480) and outputs a structured intervention plan; for intermediate warnings, the notification recipients (460) are the user and one emergency contact person, and the notification channels are APP push (461) and SMS (462); for advanced warnings, the notification recipients are additional designated medical personnel, and the notification channels are additional automated voice calls (463).

7. The medical care monitoring system according to claim 1, characterized in that, The intervention plan includes one or more of the following: medication adjustment recommendations (411), lifestyle guidance (421), rehabilitation training plan (431), or emergency medical treatment guidance (441), and is synchronized with the user's calendar application (490) to generate specific execution reminders; the visualization interaction platform (500) provides different view ports: patient view (540): displays concise health scores, trend charts, and to-do intervention tasks in the form of daily reports, weekly reports, etc. The healthcare view (550) simultaneously monitors the risk index ranking, warning list, and health record details of multiple patients in the form of a dashboard, and provides a data export function (551); the healthcare view (550) allows healthcare personnel to send customized health guidance information or modify intervention plans directly to the patient view (540) of a specific patient through the platform; the system also includes a privacy and security module (600) for: end-to-end encrypted transmission and storage of all health data (610); implementing role-based access control (620) to finely manage the viewing and operation permissions of different roles such as users, family members, and healthcare personnel; and providing data anonymization processing function (630) for use in medical research.

8. The medical care monitoring system according to claim 1, characterized in that, The data acquisition module (100) includes a basic data acquisition submodule and a risk-triggered data acquisition submodule. The basic data acquisition submodule (101) is used to continuously collect the user's routine health data at a preset first acquisition frequency. The risk-triggered data acquisition submodule is activated by the early warning and intervention module (400) when the intelligent analysis engine module (300) detects that the dynamic health risk index reaches or exceeds a preset risk concern threshold, and collects specific health data related to the risk type at a preset second acquisition frequency. The second acquisition frequency is higher than the first acquisition frequency. The early warning and intervention module (400) matches and executes a corresponding target alarm strategy from multiple preset alarm strategies according to the risk level and risk type of the dynamic health risk index. Different alarm strategies include at least one of different alarm receiving objects, alarm channel combinations, or alarm information detail levels. The early warning and intervention module (400) also includes a strategy dynamic adjustment unit, which is used to adjust the alarm channel combination and alarm information detail level in the target alarm strategy according to the user's feedback behavior data on historical intervention suggestions.

9. The medical care monitoring system according to claim 8, characterized in that, The early warning and intervention module (400) categorizes the health intervention recommendations, including primary intervention recommendations, intermediate intervention recommendations, and advanced intervention recommendations. The primary intervention recommendations are non-instructional recommendations containing only health tips and are passively displayed through the visualization interaction platform (500) when the dynamic health risk index is at a low risk level. The intermediate intervention recommendations are instructive recommendations containing specific action guidelines and are actively pushed through the target alarm strategy when the dynamic health risk index is at a medium risk level. The advanced intervention recommendations are emergency recommendations containing emergency contact information and first aid guidelines and are used to forcibly interrupt the current display interface of the visualization interaction platform (500) and display the content of the advanced intervention recommendations at the top while actively pushing through the target alarm strategy when the dynamic health risk index is at a high risk level.

10. The medical care monitoring system according to claim 9, characterized in that, The basic data acquisition submodule (101) and the risk-triggered data acquisition submodule have different configurations in terms of data acquisition objects; the basic data acquisition submodule (101) is connected to a first heterogeneous data source group, which includes a smart wearable device (110) and a user manual input terminal (130); the risk-triggered data acquisition submodule (102) is connected to a second heterogeneous data source group, which includes a home medical device (120) and a third-party medical information system (140); when the risk-triggered data acquisition submodule (102) is activated, the intelligent analysis engine module (300) acquires refined data from the second heterogeneous data source group and combines it with data from the first heterogeneous data source group to jointly update the dynamic health risk index; The early warning and intervention module (400) dynamically adjusts the classification level of the target alarm strategy and the health intervention recommendations based on the updated dynamic health risk index.

Images