Health risk assessment and intervention reminder method and device, electronic equipment and medium

Abstract

This application relates to the field of health monitoring technology, providing a method, device, electronic device, and medium for health risk assessment and intervention reminders. It generates a health status sequence by preprocessing physiological parameter data and personal health information, identifies health risk events within the health status sequence, and retrieves relevant medical knowledge entries based on the health status sequence and health risk events. According to user requests, it analyzes and evaluates the health status sequence, health risk events, and relevant medical knowledge entries to obtain initial health management information. It then performs security verification on the initial health management information to obtain verification management information, and performs multi-level intervention interactions based on the verification management information. This application solves the technical problems of existing health data analysis schemes, such as poor risk assessment accuracy and the inability to complete graded and differentiated interactive processing.

Description

Technical Field

[0001] This application relates to the field of health monitoring technology, and in particular to a method, device, electronic device, and medium for health risk assessment and intervention reminders. Background Technology

[0002] Various smart wearable devices continuously collect human physiological indicators, combining them with personal basic, lifestyle, and medical history information entered on mobile devices to generate massive amounts of heterogeneous health data. Existing health analysis solutions typically only use fixed thresholds to determine anomalies in indicators, lacking a mechanism for fusing and processing multi-source heterogeneous data. This results in low accuracy in health risk assessment, making it prone to misjudgments and omissions. Furthermore, using a uniform push notification method for all identified health risks fails to differentiate intervention levels according to risk severity, leading to a flood of alarm messages and an inability to achieve differentiated interactive processing based on risk levels. Summary of the Invention

[0003] In view of this, this application provides a health risk assessment and intervention reminder method, device, electronic device, and medium to solve the technical problems of poor risk assessment accuracy and inability to complete graded and differentiated interactive processing in existing health data analysis solutions.

[0004] The first aspect of this application provides a method for health risk assessment and intervention reminders, the method comprising:

[0005] The received physiological parameter data and personal health information are preprocessed to generate a health status sequence; Identify health risk events in the health status sequence, and perform matching retrieval based on the health status sequence and the health risk events to obtain relevant medical knowledge entries; In response to user requests, the health status sequence, the health risk events, and the relevant medical knowledge items are analyzed and evaluated to obtain initial health management information; The initial health management information is security verified to obtain verification management information, and multi-level intervention interactions are performed based on the verification management information.

[0006] In an optional implementation, the preprocessing of the received physiological parameter data and personal health information to generate a health status sequence includes: Real-time collection of physiological detection data from target users, followed by data cleaning and standardization to obtain a physiological parameter dataset; The user status features are extracted from the received health description information, and the user status features are structured to obtain individual health information data. By using a preset time window, the physiological parameter dataset and the individual health information data are time-series correlated and fused to generate a health status sequence.

[0007] In an optional implementation, identifying health risk events in the health status sequence and performing a matching search based on the health status sequence and the health risk events to obtain relevant medical knowledge entries includes: The health status sequence is subjected to time-series feature extraction processing to obtain an individualized health baseline; Based on a pre-defined medical rule base and the individualized health baseline, risk analysis is performed on the health status sequence to identify health risk events. A retrieval feature vector is constructed based on the health risk events and the health status sequence, and semantic matching retrieval is performed on a preset medical encyclopedia knowledge base based on the retrieval feature vector to obtain relevant medical knowledge entries.

[0008] In an optional implementation, the step of performing risk analysis on the health status sequence based on a preset medical rule base and the individualized health baseline to identify health risk events includes: Read the physiological parameter type corresponding to the individualized health baseline, and match the corresponding target medical rule base and target statistical threshold from the preset medical rule base according to the physiological parameter type; The current physiological parameter values ​​in the health status sequence are compared with the individualized health baseline to obtain the baseline deviation, and the baseline deviation is compared with the target statistical threshold to obtain the first risk feature; The physiological parameter change trends in the health status sequence are matched with the target medical rule base to obtain the second risk feature; Based on the first risk feature and the second risk feature, health risk events are identified from the health status sequence.

[0009] In an optional implementation, the step of analyzing and evaluating the health status sequence, the health risk events, and the relevant medical knowledge items to obtain initial health management information includes: The health status sequence, the health risk events, and the relevant medical knowledge items are fused according to a preset data format to generate data to be analyzed. The data to be analyzed is evaluated based on the received user request in order to filter out initial health management information.

[0010] In an optional implementation, the step of performing security verification on the initial health management information to obtain verification management information, and performing multi-level intervention interactions based on the verification management information, includes: The initial health management information is security verified according to preset multi-level security rules to obtain the first verification sub-data. The security conflict sub-data is corrected by a preset correction mechanism to obtain the second verification sub-data; Identify the intervention type and intervention level corresponding to the first verification sub-data and / or the second verification sub-data, and perform graded intervention interaction according to the intervention type and the intervention level through a preset graded intervention interaction method.

[0011] In an optional implementation, the step of correcting the security conflict sub-data through a preset correction mechanism to obtain the second verification sub-data includes: The conflict type is identified for the security conflict sub-data, and the conflict level corresponding to the security conflict sub-data is determined. Based on the conflict level, a target correction strategy corresponding to the conflict level is selected from the preset correction mechanisms; When the target correction strategy is the preset first strategy, the security conflict sub-data is subjected to content correction processing to obtain the second verification sub-data; When the target correction strategy is the preset second strategy, the security conflict sub-data is intervened and downgraded to obtain the second verification sub-data.

[0012] A second aspect of this application provides a health risk assessment and intervention reminder device, the device comprising: The data optimization module is used to preprocess the received physiological parameter data and personal health information to generate a health status sequence; The knowledge matching module is used to identify health risk events in the health status sequence, and to perform matching retrieval based on the health status sequence and the health risk events to obtain relevant medical knowledge entries; The analysis and evaluation module is used to respond to user requests, analyze and evaluate the health status sequence, the health risk events, and the relevant medical knowledge items, and obtain initial health management information. The hierarchical interaction module is used to perform security verification on the initial health management information to obtain verification management information, and to perform multi-level intervention interactions based on the verification management information.

[0013] A third aspect of this application provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the health risk assessment and intervention reminder method as described above.

[0014] A fourth aspect of this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the health risk assessment and intervention reminder method described above.

[0015] In summary, this application includes at least the following beneficial technical effects: This application preprocesses received multi-source physiological parameter data and heterogeneous personal health information to generate a continuous, structured time-series sequence of health states. This achieves effective fusion and aggregation of multi-source heterogeneous data, providing a standardized and high-quality data foundation for subsequent accurate risk assessment and improving the accuracy of health data analysis from the source. Based on the health state time-series sequence, it identifies health risk events and performs matching retrieval based on the sequence and events, retrieving relevant medical knowledge entries as supporting data for assessment. This enriches the reference dimensions for risk assessment and solves the problem of misjudgment and omission that easily occurs with fixed threshold judgments. Simultaneously, by binding professional medical benchmark data through a standardized retrieval mechanism, risk assessment no longer relies on a single indicator threshold, thus improving... This system ensures the comprehensiveness and accuracy of health risk identification; responds to user-triggered requests, integrates the time sequence of health status, identified health risk events, and relevant medical knowledge entries, and outputs initial health risk management information; it performs security verification on the initial health risk management information to obtain compliant and valid verification management information, avoiding invalid reminders and erroneous interventions caused by erroneous health information output, ensuring the accuracy and reliability of subsequent risk grading and interactive push notifications, and optimizing the overall data processing fault tolerance; finally, it performs multi-level intervention interactions based on the verification management information, realizing layered intervention, accurately matching different levels of interaction methods, reducing invalid alarm push notifications, improving the targeting and interactive processing efficiency of health risk intervention, and achieving intelligent, graded, and differentiated intervention. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of the operating environment of a health risk assessment and intervention reminder method provided in an embodiment of this application; Figure 2 This is a flowchart of a health risk assessment and intervention reminder method provided in an embodiment of this application; Figure 3 This is a functional block diagram of a health risk assessment and intervention reminder device provided in an embodiment of this application; Figure 4 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0018] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0019] like Figure 3 The diagram shown illustrates the operating environment of a health risk assessment and intervention reminder method provided in this embodiment. This embodiment operates in a three-layer distributed computing environment comprised of wearable devices, mobile devices, and a backend intelligent system.

[0020] Wearable devices include at least one of the following: smartwatches, smart rings, smart bracelets, continuous glucose monitoring devices, electronic blood pressure monitors, ECG patches, and sleep monitoring devices. These devices are worn on the target user's body and collect at least one physiological parameter data in real time using various built-in micro-sensors (such as photoelectric sensors, electrode pads, accelerometers, and gyroscopes). This data includes at least one of the following: blood pressure, blood glucose, heart rate, blood oxygen, body temperature, respiratory rate, ECG data, sleep duration, sleep quality, steps taken, exercise intensity, and energy expenditure. The wearable device's built-in microprocessor converts the raw physiological signals captured by the sensors into digitized physiological parameter data. This data is then wirelessly connected to the mobile device via its communication module (supporting at least one of Bluetooth, Wi-Fi, cellular network, or near-field communication), transmitting the collected physiological parameter data to a mobile application on the mobile device in real time.

[0021] The mobile device layer consists of smartphones, tablets, or other mobile terminals with mobile applications installed and running. This mobile application acts as a data aggregation node, receiving physiological parameter data uploaded from wearable devices via Bluetooth, Wi-Fi, cellular networks, or near-field communication. Simultaneously, it serves as a user interface, acquiring personal health information (including at least one of age, gender, height, weight, past medical history, family medical history, allergy history, current symptoms, medication information, lifestyle habits, dietary information, and exercise habits) entered by the target user through text input, voice input, image recognition, prescription scanning, or import from external medical systems. The mobile application sends the received physiological parameter data and acquired personal health information to the backend intelligent system for processing via a network (such as the internet). Upon receiving the processing results from the backend system, it provides the target user with at least one of the following through in-app pop-ups, message notifications, voice reminders, SMS reminders, or contact reminders: health advice, auxiliary health assessment, risk warnings, medication reminders, retest reminders, medical advice, or medical encyclopedia recommendations.

[0022] The backend intelligent system layer includes a processor, a memory, and a target generation model deployed therein. This backend system can be at least one of a single server, multiple servers, a cloud computing platform, or a virtualization center. The processor executes the computer program stored in the memory to implement the health risk assessment and intervention reminder method described in this application, specifically including: preprocessing received physiological parameter data and personal health information to generate structured health data; constructing a health status sequence based on the structured health data; identifying health risk events; retrieving relevant medical knowledge entries from a doctor-verified medical encyclopedia knowledge base; generating input content for the target generation model; calling the target generation model to obtain health management results; and performing security verification on the health management results. The memory stores the computer program required for processor operation and intermediate data generated during operation. The target generation model includes a general large language model or a shared intelligence body (the shared intelligence body includes a large language model and a doctor-customized personalized knowledge base), used to generate health management results based on the input content of the target generation model. Furthermore, the backend intelligent system also deploys a doctor-verified medical encyclopedia knowledge base module to store doctor-verified medical knowledge entries and provide medical encyclopedia content to the mobile application.

[0023] A complete data link is formed between wearable devices, mobile devices, and the backend intelligent system: wearable devices collect physiological parameter data and transmit it to the mobile application on the mobile device via short-range wireless communication; the mobile application simultaneously obtains the personal health information input by the user, summarizes the two types of data, and uploads it to the backend intelligent system via the Internet; after completing data processing, risk identification, knowledge retrieval, model reasoning, and security verification, the backend intelligent system transmits the health management results back to the mobile application via the network; the mobile application finally outputs intervention reminder information to the target user in various interactive ways.

[0024] The health risk assessment and intervention reminder method provided in this application is executed by an electronic device, and correspondingly, the health risk assessment and intervention reminder device operates within the electronic device. The health risk assessment and intervention reminder method provided in this application will be described below from the perspective of an electronic device integrating a backend intelligent system.

[0025] like Figure 2 The diagram shown is a flowchart of a health risk assessment and intervention reminder method provided in an embodiment of this application. The health risk assessment and intervention reminder method provided in this embodiment includes the following steps.

[0026] Step S1: Preprocess the received physiological parameter data and personal health information to generate a health status sequence.

[0027] It should be understood that one or more wearable devices worn by the target user continuously and in real time capture multidimensional signals from the user's body at a preset sampling frequency. Wearable devices include at least one of the following: smart rings, smartwatches, smart bracelets, continuous glucose monitoring devices, electronic blood pressure monitors, ECG patches, and sleep monitoring devices. Wearable devices integrate various miniature sensors: photoelectric sensors for detecting heart rate and blood oxygen, electrode pads for collecting ECG data, tiny accelerometers and gyroscopes for recording steps and sleep patterns, and dedicated sensing units for measuring body temperature and blood pressure. These sensors, at specific sampling rates (e.g., a 5-minute sampling interval for heart rate, a 15-minute sampling interval for continuous glucose, and daily cumulative step counts), convert the captured raw physiological signals into digital electrical signals via a built-in analog-to-digital converter. The wearable device's built-in microprocessor then performs preliminary encapsulation processing on these digital signals, adding a sampling timestamp and device identification information to form raw format physiological detection data. Subsequently, the wearable device establishes a wireless connection with the target user's mobile device via its communication module (supporting at least one of Bluetooth, Wi-Fi, cellular network, or near-field communication), transmitting the packaged physiological detection data in real time to a mobile application running on the mobile device. The mobile application runs a persistent data receiving service in the background, responsible for listening to and receiving data streams from the bound wearable device, and temporarily storing all received raw physiological detection data in the mobile terminal's local cache, forming a continuously updated but unprocessed set of physiological detection data.

[0028] Meanwhile, the mobile application acquires health description information actively input or imported by the target user through its user interface. Target users can provide health description information in several ways: by inputting text via a virtual keyboard, filling in descriptive content such as current symptoms, medication history, and dietary records; by speaking naturally, such as "I haven't been sleeping well for the past three nights, and I feel a bit palpitations during the day," with the mobile application's built-in speech recognition module converting the audio into text; or by taking a picture of and uploading a paper prescription or medical examination report using the mobile device's camera, with the mobile application extracting the text from the image using optical character recognition technology. After acquiring the health description information in these various forms, the mobile application immediately performs preliminary text collection and processing, forming a data set containing the user's original health description text (i.e., health description information). The reason why the above-mentioned complete collection and transmission process of physiological test data is necessary is that the raw signals collected by wearable devices are mixed with noise data caused by loose device wearing, motion artifacts or environmental interference. At the same time, the data sampling frequency and timestamp format of different devices are different. This raw data cannot be directly used for subsequent health status analysis. It must be preprocessed by the system to be transformed into a structured form that can be efficiently recognized and processed by computer programs.

[0029] The mobile device uploads the received physiological data to the backend intelligent system via the internet, whereby the system performs data cleaning. Data cleaning first involves thoroughly checking and deleting completely duplicate records caused by network transmission errors or device restarts. Then, it identifies data points that clearly exceed the device's reasonable measurement range—for example, a human resting heart rate is typically between 40 and 120 beats per minute; if a heart rate record shows 0 or exceeds 300, that data point is marked as a suspected anomaly. Such anomalies are usually caused by temporary device loosening, momentary sensor malfunction, or extreme motion interference. After data cleaning, the backend intelligent system performs standardization processing on the physiological data. Standardization includes two aspects: first, unit conversion, unifying the different units of measurement that different wearable devices may use into standard medical units—for example, converting height from centimeters to meters, and weight from pounds to kilograms; second, numerical standardization, employing appropriate standardization methods for different physiological parameter types, enabling data from different devices and with different dimensions to be compared and analyzed on the same numerical scale. The physiological parameter dataset obtained after the above data cleaning and standardization processes is a clean dataset that has removed duplicate records, marked outliers, and standardized units of measurement and numerical scales.

[0030] After cleaning and standardizing the physiological test data, the backend intelligent system simultaneously performs structured processing on the health description information uploaded from mobile devices to extract user status features and obtain individual health information data. This structured processing first segments the text content in the health description information into word segments, dividing continuous text into the smallest linguistic units with independent semantics. Then, entity recognition is performed on the segmentation results to identify entities with specific medical meanings from the text—for example, from the description "I have high blood pressure and am taking nifedipine, once a day, 30 mg each time," the entities "high blood pressure," "nifedipine," "once a day," and "30 mg" can be identified. After entity recognition, the backend intelligent system further performs relation extraction processing to identify the semantic relationships between entities—for example, associating "nifedipine" with "once a day" as a medication regimen relationship, and associating "high blood pressure" with "nifedipine" as a treatment relationship. For health descriptions input via voice, the backend intelligent system first converts the audio into text using speech recognition technology before entity recognition and relation extraction. Then, it performs the same natural language processing (NLP) process on the text input. For health descriptions obtained via image recognition (such as scanned prescriptions), the backend intelligent system first extracts the text content from the image into editable text using optical character recognition (OCR) technology before NLP. Then, it performs word segmentation, entity recognition, and relation extraction on this text. After this complete NLP process, all user status features extracted from the original health descriptions (including the target user's age, gender, height, weight, past medical history, family medical history, allergy history, current symptoms, current medication information (including drug name, usage, dosage, and frequency), dietary information, exercise habits, lifestyle habits, and doctor's diagnosis information) are organized and stored according to a pre-defined standardized data format, forming structured individual health information data. Because health descriptions are fragmented, unstructured self-descriptions of health provided by users in natural language, they are transformed into standardized data fields that computer programs can quickly read, accurately match, and efficiently calculate.

[0031] After obtaining cleaned and standardized physiological parameter datasets and structured individual health information data, the backend intelligent system performs time-series correlation processing on these two types of data through a preset time window to generate a health status sequence. To address the inconsistency in the time dimension between the two data sources—the physiological parameter dataset is automatically generated by wearable devices at a fixed collection frequency (e.g., heart rate measurement every 5 minutes, blood glucose recording every hour), with regular time intervals and clear timestamps; while most of the content in the individual health information data (e.g., past medical history, allergy history, height, weight, etc.) is relatively static user attribute information without time-varying characteristics, and a small portion (e.g., the onset time of current symptoms, medication administration time), although possessing time attributes, is usually not perfectly aligned with the collection time of the physiological parameter data—the backend intelligent system first sets at least one preset time window (e.g., a time window length of 5 minutes, 10 minutes, 1 hour, or 1 day), using this preset time window as a unified time-series analysis benchmark. For the physiological parameter dataset, the system resamples the data according to this preset time window. For physiological parameters with multiple sampling points within a time window, their mean is calculated as the representative value for that window. For physiological parameters without sampling points within a window, time completion is performed using forward padding (i.e., using the value from the previous valid window) or linear interpolation. For individual health information data, the system treats static attribute information (such as age, gender, and past medical history) as constant features across all time windows; and maps dynamic event information (such as symptom onset time and medication administration time) to corresponding preset time windows according to their occurrence time. After completing the above time alignment processing, the system associates and combines the representative value of the physiological parameter and the corresponding individual health information feature value within each preset time window to form a multidimensional data vector describing the user's complete health status within that time window. Arranging all preset time windows in chronological order, these multidimensional data vectors together constitute a temporal data structure describing the evolution of the user's health status from a certain point in the past to the present moment, i.e., a health status sequence. By performing time-series correlation processing on physiological parameter datasets and individual health information data through preset time windows, multi-source heterogeneous data from different devices, time granularities, and update frequencies are unified onto the same time axis for joint analysis. This enables the risk identification step to simultaneously acquire users' objective physiological measurement data and subjective health description data within any time window, achieving health status assessment based on a complete time context.

[0032] Step S2: Identify health risk events in the health status sequence, and perform matching retrieval based on the health status sequence and the health risk events to obtain relevant medical knowledge entries.

[0033] After obtaining a health status sequence describing the evolution of a target user's health status, the backend intelligent system first performs time-series feature extraction processing on the health status sequence to obtain an individualized health baseline. The health status sequence is a multi-dimensional data vector set composed of multiple preset time windows arranged chronologically. Each time window's data vector contains representative values ​​of various physiological parameters of the target user within that window, along with corresponding individual health information feature values. The time-series feature extraction processing is performed separately for each type of physiological parameter in the health status sequence. For a specific physiological parameter (e.g., resting heart rate), the backend intelligent system extracts all historical values ​​of that physiological parameter from the health status sequence within a long-term time window (e.g., the past 30 days or the past 90 days) and performs statistical analysis on these historical values: calculating the arithmetic mean of these historical values ​​as the individualized central baseline value of the physiological parameter, and calculating the standard deviation of these historical values ​​as the individualized fluctuation range of the physiological parameter. Through this method, the backend intelligent system establishes a corresponding individualized health baseline for each physiological parameter (including at least one of blood pressure, blood glucose, heart rate, blood oxygen, body temperature, respiratory rate, sleep duration, and step count). This individualized health baseline differs from the universally accepted normal ranges found in medical textbooks. For example, a young athlete who engages in endurance exercise regularly might have an individualized resting heart rate baseline of 50 beats / min, while a typical sedentary adult might have an individualized resting heart rate baseline of 78 beats / min. The reason for establishing individualized health baselines for each physiological parameter, rather than directly using universal medical reference ranges, is that due to the combined influence of various factors such as age, gender, physical condition, exercise habits, and underlying diseases, there are significant individual differences in the normal levels of various physiological parameters among different target users. Using a uniform universal threshold for risk assessment would produce a large number of false positive or false negative results. Only by establishing individualized baselines based on the user's own historical data can truly clinically significant deviations be accurately identified for that user.

[0034] After obtaining an individualized health baseline, the backend intelligent system performs risk analysis on the health status sequence based on a preset medical rule base and the individualized health baseline to identify health risk events. This risk analysis process is further refined into four sequentially executed sub-processing steps. The first sub-processing step is the matching of rules and thresholds. The backend intelligent system first reads the physiological parameter type corresponding to the individualized health baseline—for example, if the current risk analysis is related to heart rate, the physiological parameter type corresponding to the heart rate baseline in the individualized health baseline is read as "heart rate." Based on this physiological parameter type, the backend intelligent system matches the target medical rule base corresponding to this physiological parameter type from the preset medical rule base, and simultaneously matches the target statistical threshold corresponding to this physiological parameter type from the preset statistical threshold base. The pre-defined medical rule base stores multiple logical judgment rules pre-defined by medical experts. For example, for the physiological parameter type of heart rate, it may include a rule that "a sustained increase in resting heart rate for more than a certain number of hours suggests possible cardiovascular stress." The pre-defined statistical threshold base stores various statistical discrimination thresholds. For example, for the degree of heart rate deviation from the individualized baseline, it may include a statistical threshold that "the deviation exceeds twice the standard deviation of the individualized baseline." The second sub-processing step is baseline deviation analysis. The backend intelligent system compares the physiological parameter value (i.e., the current physiological parameter value) within the current time window in the health status sequence with the corresponding baseline value in the individualized health baseline to calculate the baseline deviation. The baseline deviation can be calculated as follows: Let the current physiological parameter value be X, the central baseline value of the individualized health baseline be μ, and the standard deviation of the individualized health baseline be σ. Then, the baseline deviation can be obtained by calculating the ratio of the difference between the current physiological parameter value and the central baseline value to the standard deviation: D=(X-μ) / σ. After obtaining the baseline deviation, the backend intelligent system compares this baseline deviation with the target statistical threshold matched in the first sub-processing step: when the baseline deviation exceeds the upper limit specified by the target statistical threshold, it is determined that the current physiological parameter value has a significant deviation from the individualized health baseline; when the baseline deviation does not exceed the target statistical threshold, it is determined that the current physiological parameter value is within the individual's normal fluctuation range. After the above comparison processing, the backend intelligent system obtains the first risk feature, which is used to characterize the degree of deviation of the current physiological parameter value from the target user's own historical normal level. The third sub-processing step is the trend matching processing. The backend intelligent system extracts the change trajectory of the current physiological parameter from the health status sequence within a short time window (e.g., the past 24 hours or the past 48 hours), including the direction of change of the physiological parameter (continuous increase, continuous decrease, stable fluctuation) and the rate of change (the magnitude of increase or decrease per hour).The backend intelligent system compares the extracted physiological parameter change trends with the target medical rule base matched in the first sub-processing stage. For example, if the target medical rule base contains a rule that "a sustained increase in heart rate exceeding ten beats / min over six hours warrants attention," the backend intelligent system determines whether the currently extracted heart rate change trend meets the conditions described by that rule. After the above matching process, the backend intelligent system obtains a second risk feature, which characterizes whether the current physiological parameter change trend conforms to the risk pattern defined in the preset medical rule base. The fourth sub-processing stage is a comprehensive judgment process. The backend intelligent system jointly analyzes the first risk feature obtained in the second sub-processing stage and the second risk feature obtained in the third sub-processing stage to identify health risk events from the health status sequence. The joint analysis specifically involves the following steps: When the first risk feature indicates that the current physiological parameter value significantly deviates from the individualized health baseline, and the second risk feature indicates that the current physiological parameter change trend conforms to the risk pattern defined by one or more target medical rule bases, the backend intelligent system identifies a health risk event within the current time window. When only the first risk feature or only a single feature from the second risk feature indicates an abnormality, the backend intelligent system determines whether it is a low-level risk event or not to be classified as a risk event based on preset confidence rules. The health risk events identified after the above complete risk analysis process include at least one of the following: abnormal blood pressure events, abnormal blood sugar events, abnormal heart rate events, sleep deprivation events, insufficient exercise events, medication omission events, symptom exacerbation events, and suspected acute risk events. For example, a target user with hypertension has an individualized baseline systolic blood pressure of 130 mmHg and a standard deviation of 5 mmHg. The target statistical threshold is set to trigger a risk warning if the deviation exceeds twice the standard deviation. When the user's current systolic blood pressure reaches 142 mmHg, the baseline deviation is calculated as (142-130) / 5=2.4, which exceeds the target statistical threshold of 2.0. At the same time, the user's systolic blood pressure has been rising continuously over the past 6 hours, which matches the rule in the preset medical rule base that "systolic blood pressure rising continuously for more than two hours requires attention". Then, the backend intelligent system integrates the above first risk feature and second risk feature to identify the health risk event "abnormal blood pressure event" from the health status sequence.

[0035] After identifying a health risk event, the backend intelligent system constructs a retrieval feature vector based on the event and the health status sequence. This feature vector is then used to perform semantic matching searches on a pre-defined medical encyclopedia to obtain relevant medical knowledge entries. The construction of the retrieval feature vector begins with the identified health risk event as core information (including the type of the event, such as "abnormal blood pressure event"), risk level (such as high, medium, or low risk), occurrence time, and key data features that triggered the event. This is combined with the demographic characteristics of the target user in the health status sequence (such as age and gender), pre-existing medical history (such as a history of hypertension), and the current health status context (such as information on medications being taken). This multi-dimensional information is fused and encoded to form a high-dimensional numerical vector, the retrieval feature vector. The purpose of constructing this retrieval feature vector is to transform the health risk event and health status information described in natural language into a vector form that can be mathematically calculated. This allows the retrieval operation to move beyond rigid keyword matching and instead rely on the semantic similarity between vectors for retrieval. The pre-built and continuously maintained medical encyclopedia knowledge base stores a vast number of medical knowledge entries that have been reviewed and approved by doctors or medical experts. Each medical knowledge entry is converted into a corresponding knowledge entry vector using the same vectorization model and stored in the knowledge base upon entry. The backend intelligent system calculates the similarity between the constructed retrieval feature vector and all knowledge entry vectors stored in the medical encyclopedia knowledge base (e.g., calculating the cosine similarity between the retrieval feature vector and each knowledge entry vector) to measure the semantic proximity between the retrieval feature vector and each knowledge entry vector. The backend intelligent system sets a preset similarity threshold and selects several medical knowledge entries with similarity scores higher than the threshold or the highest similarity ranking as the search results. For example, when the identified health risk event is an "abnormal blood pressure event", the retrieval feature vector constructed by the backend intelligent system may retrieve a series of related medical knowledge items from the medical encyclopedia knowledge base through semantic matching, such as "diagnostic criteria and classification of hypertension", "correct methods of home blood pressure monitoring", "lifestyle intervention recommendations for hypertensive patients (including low-salt diet, regular exercise, etc.)", "mechanism of action and precautions of commonly used antihypertensive drugs", and "when hypertensive patients experience certain symptoms, they should seek medical attention promptly".

[0036] Step S3: In response to the user request, analyze and evaluate the health status sequence, the health risk events, and the relevant medical knowledge items to obtain initial health management information.

[0037] After identifying health risk events and acquiring corresponding medical knowledge entries, the backend intelligent system needs to perform structured fusion processing on the aforementioned multi-source heterogeneous data in order to comprehensively analyze the overall health status of the target user and generate targeted preliminary responses. This fusion processing organizes and arranges the health status sequence, health risk events, and related medical knowledge entries according to a preset data format to generate data for analysis. The health status sequence includes information such as the current value, historical average, trend, fluctuation range, and symptom description of various physiological parameters of the target user within multiple preset time windows. This sequence exists in the form of time-series multi-dimensional data vectors, where each data vector contains the values ​​of various physiological parameters within that time window and the corresponding individual health information feature values. Health risk events include the type of risk event (e.g., abnormal blood pressure, abnormal blood sugar, abnormal heart rate, insufficient sleep, etc.), risk level (high risk, medium risk, or low risk), occurrence time, and key data features that triggered the risk event. The relevant medical knowledge entries are medical knowledge content reviewed by doctors and obtained through semantic matching retrieval from a medical encyclopedia knowledge base. This includes information such as disease names, risk factors, common symptoms, monitoring indicators, lifestyle recommendations, medication precautions, retesting suggestions, medical consultation tips, applicable populations, and contraindications. These three types of data differ significantly in data structure, format, and expression: health status sequences are numerical data organized as time series; health risk events are structured event records with type labels and level annotations; and relevant medical knowledge entries are knowledge-based content in natural language text format. If these three types of data are directly input into the analysis and evaluation stage in their raw formats, the inconsistent data formats and information organization methods will prevent the analysis and evaluation stage from efficiently utilizing all three types of information for comprehensive reasoning. Therefore, the backend intelligent system performs unified fusion processing on the three types of data according to a preset data format.

[0038] The fusion process first extracts content and generates summaries from the health status sequence. The backend intelligent system extracts the current values ​​of various physiological parameters within the current time window from the health status sequence, forming a "current physiological parameter summary"; it extracts the trends and fluctuations of various physiological parameters within short-term time windows (e.g., the past 24 hours or the past 7 days) from the historical portion of the health status sequence, forming a "historical trend summary"; and it extracts tagged abnormal events and user-entered symptom descriptions from the health status sequence, forming an "abnormal event and symptom summary." Secondly, the backend intelligent system formats the descriptions of health risk events, converting the type, level, occurrence time, and triggering characteristics of risk events into natural language descriptions according to preset text templates. For example, a "blood pressure abnormality event" is converted into a standardized description such as, "The system has identified an abnormal blood pressure event in you, with a risk level of medium risk, occurring this morning, triggered by a systolic blood pressure of 142 mmHg, deviating from your individualized health baseline of 130 mmHg by approximately 9.2%." Secondly, the backend intelligent system categorizes and organizes relevant medical knowledge items according to their respective medical themes. For example, items belonging to the "disease knowledge" category, the "lifestyle advice" category, and the "medical treatment guidance" category are grouped separately, and the core content summary of each item is extracted. After completing the content extraction and format conversion of the above three types of data, the backend intelligent system arranges and assembles all the above information according to a preset data organization template in the order of "user basic information, recent physiological parameter summary, historical trend summary, risk event summary, current symptoms and medication information, relevant medical knowledge items, and output constraints," generating a complete and orderly dataset to be analyzed. This dataset integrates information that was originally scattered across different data structures and storage locations into a unified and logically clear data whole, enabling the analysis and evaluation process to simultaneously obtain a complete health profile of the target user, identified risk events, and authoritative medical knowledge for reference based on a single input.

[0039] After generating the data to be analyzed, the backend intelligent system analyzes and evaluates the data based on received user requests to filter out initial health management information. Target users submit requests through the mobile application's interface via text input, voice input, or interface selection. These requests can take various forms, including questions about current health status (e.g., "My blood pressure has been high lately, what should I be careful about?"), inquiries about specific symptoms (e.g., "I've been experiencing heart palpitations these past few days, what could be the cause?"), confirmation requests for medication plans (e.g., "Do I need to take my blood pressure medication today?"), or inquiries about health trends (e.g., "Please check my sleep patterns for the past week"). Upon receiving the user request, the mobile application uploads it to the backend intelligent system via the internet. The backend intelligent system then correlates the received user request with the generated data to be analyzed, using the user request as the trigger and focus of the analysis, and the data to be analyzed as the information basis for the analysis, to perform a comprehensive analysis and evaluation.

[0040] The analysis and evaluation process begins with semantic understanding and contextual association of the content in the data to be analyzed. The backend intelligent system parses the intent and focus contained in the user request—for example, if the user request is "My blood pressure has been high recently, what should I pay attention to?", the backend intelligent system identifies the core intent of the request as "to obtain health management advice related to blood pressure," with the focus on the physiological parameter "blood pressure" and the abnormal state of "high." Based on the identified intent and focus, the backend intelligent system filters all blood pressure-related information from the data to be analyzed, including the current value, historical trend, and fluctuation range of blood pressure parameters in the health status sequence, details of blood pressure-related risk events in health risk events, and knowledge content related to hypertension management in relevant medical knowledge entries. Subsequently, the backend intelligent system jointly analyzes the filtered targeted information with the user request to generate preliminary response content directly corresponding to the user request. This analysis and evaluation process is not simply extracting existing content from the data to be analyzed, but rather performing comprehensive reasoning and judgment based on the information provided by the data to be analyzed. For example, the severity of the abnormality can be inferred from the deviation of the current blood pressure value from the individualized health baseline, the direction of risk development can be inferred from the blood pressure change trend, and then combined with the suggestions on hypertension management in relevant medical knowledge entries to form a personalized response for the user's current specific situation.

[0041] For example, a target user with a history of hypertension submits a user request: "My blood pressure has been high recently, what should I pay attention to?" Upon receiving this request, the backend intelligent system extracts the following information from the data to be analyzed: the user's current systolic blood pressure is 142 mmHg, the individualized health baseline is 130 mmHg, the standard deviation is 5 mmHg, the systolic blood pressure trend over the past six hours shows a continuous upward trend, the identified health risk event is "abnormal blood pressure event (medium risk)," and relevant medical knowledge items include "lifestyle intervention recommendations for hypertension patients" and "medical consultation tips for hypertension patients." The backend intelligent system comprehensively analyzes the extracted information with the user request, determining that the user's current blood pressure value has significantly deviated from the individualized health baseline and shows a continuous upward trend. Therefore, it determines that a preliminary response containing an explanation of the risk's cause, specific lifestyle adjustment recommendations, and advice on when to seek medical attention is needed. The initial response includes explanations of possible causes for the current elevated blood pressure (e.g., "Recent sleep deprivation may have caused fluctuations in blood pressure"), specific actionable recommendations (e.g., "It is recommended to measure and record blood pressure twice a day, once in the morning and once in the evening, for three consecutive days," "Reduce sodium intake, keeping daily salt intake below 5 grams," "Increase moderate aerobic exercise, such as brisk walking for 30 minutes daily"), and clear medical guidance (e.g., "If blood pressure is higher than 135 / 85 mmHg for three consecutive mornings, it is recommended to consult a cardiologist promptly"). The initial health management information obtained after the above analysis and evaluation is a comprehensive analysis result combining the target user's real-time physiological data, personalized health baseline, identified risk events, authoritative medical knowledge, and the user's specific requests.

[0042] Step S4: Perform security verification on the initial health management information to obtain verification management information, and perform multi-level intervention interaction based on the verification management information.

[0043] After obtaining the initial health management information, the backend intelligent system performs a security verification process on this information to ensure that all content ultimately output to the target user complies with medical safety standards. The initial health management information is generated by the analysis and evaluation phase based on a comprehensive analysis of health status sequences, health risk events, and relevant medical knowledge items. Although the analysis and evaluation process references medical knowledge verified by doctors, it is important to consider that the target generation model used in the analysis and evaluation may suffer from model illusion risks due to limitations in training data, or may generate recommendations that contradict medical safety principles due to inference biases about the individual circumstances of the target user. Therefore, an independent, rule-based, and rigorous security review must be conducted before the content is output to the user. The security verification operation reviews each item of the initial health management information according to preset multi-level security rules. The pre-defined multi-level safety rules include several rule entries marked as mandatory prohibitions or mandatory requirements. Specifically, these include: prohibiting the output content from suggesting the target user start using any new prescription drugs; prohibiting the output content from suggesting the target user increase, decrease, or adjust existing medication dosages; prohibiting the output content from suggesting the target user discontinue any long-term prescription drugs; requiring that output content involving high-risk symptoms such as chest pain, difficulty breathing, severe headache, or confusion must also include clear medical advice; and requiring that for judgments with uncertainty in the analysis and evaluation results, an additional suggestion to the target user to consult a doctor must be added. After obtaining the complete text content of the initial health management information, the backend intelligent system performs sentence-by-sentence scanning and semantic analysis, comparing each sentence with each rule in the aforementioned multi-level safety rules. For prohibition rules, it determines whether the current sentence contains a semantically equivalent expression to the content prohibited by the rule; for mandatory rules, it determines whether the entire initial health management information already contains the content required by the rule. After the above item-by-item comparison process, the backend intelligent system classifies the content in the initial health management information that does not trigger any security rules and meets all mandatory rule requirements as the first verification sub-data, and classifies the content in the initial health management information that triggers at least one security rule as the security conflict sub-data. For example, if a target user's initial health management information contains the statement "Given your recent large fluctuations in blood pressure, you can try increasing the nifedipine dose to 40 mg per day," this statement triggers the security rule "It is forbidden to modify the recommended drug dosage without authorization," so this sentence is classified as the security conflict sub-data; while other parts of the initial health management information that do not contain any violation content, such as "It is recommended to monitor morning blood pressure for three consecutive days" and "It is recommended to reduce sodium intake," are classified as the first verification sub-data.Since different parts of the initial health management information may belong to trusted content and conflicting content respectively, the first verification sub-data and the security conflict sub-data obtained by the backend intelligent system have an "and / or" relationship. That is, it may only obtain the first verification sub-data (when all the initial health management information passes verification), it may only obtain the security conflict sub-data (when all the initial health management information is conflicting), or it may obtain both the first verification sub-data and the security conflict sub-data (when some content in the initial health management information passes verification and some content is conflicting).

[0044] After completing security verification and obtaining security conflict sub-data, the backend intelligent system corrects the security conflict sub-data through a preset correction mechanism to obtain the second verification sub-data. This correction mechanism first identifies the conflict type of the security conflict sub-data to determine its corresponding conflict level. The conflict type identification process analyzes the type of security rules triggered in the security conflict sub-data, distinguishing whether it belongs to one of the following categories: "conflict involving new prescription drug recommendations," "conflict involving unauthorized modification of dosage recommendations," "conflict involving unauthorized discontinuation of medication recommendations," "conflict involving lack of mandatory medical consultation reminders," or "conflict involving conflict with already verified medical knowledge." Based on this, the backend intelligent system determines the conflict level corresponding to the security conflict sub-data according to the triggered conflict type and its severity. The rules for determining conflict levels are as follows: conflicts involving new prescription drug recommendations or lack of medical advice for high-risk symptoms are classified as high-risk conflicts; conflicts involving unauthorized modification of drug dosage recommendations, unauthorized discontinuation of medication recommendations, or serious conflicts with already-corrected medical knowledge entries are classified as medium-risk conflicts; conflicts involving minor inconsistencies with already-corrected medical knowledge entries or lack of non-mandatory recommendations are classified as low-risk conflicts. For example, if the safety conflict sub-data includes a new prescription drug recommendation such as "You can try taking aspirin," the conflict is identified as a "new prescription drug recommendation conflict," and the corresponding conflict level is classified as high-risk; if the safety conflict sub-data includes a dosage modification recommendation such as "It is recommended that you reduce your blood pressure medication dosage yourself," the conflict is identified as a "unauthorized dosage modification conflict," and the corresponding conflict level is classified as medium-risk.

[0045] After determining the conflict level corresponding to the security conflict sub-data, the backend intelligent system selects a target correction strategy corresponding to the conflict level from a preset correction mechanism. The preset correction mechanism stores multiple different correction strategies, each corresponding to a specific conflict level and processing method. When the target correction strategy is the preset first strategy, the backend intelligent system performs content correction processing on the security conflict sub-data to obtain the second verification sub-data. This content correction processing semantically rewrites the specific conflict content in the security conflict sub-data: if the security conflict sub-data contains the statement "you can increase the drug dosage yourself," the backend intelligent system rewrites this statement as "Do not adjust the drug dosage yourself; if blood pressure remains abnormal, please consult a doctor"; if the security conflict sub-data contains the statement "it is recommended to stop taking this drug," the backend intelligent system rewrites this statement as "Do not stop taking the medication without authorization; if you need to stop taking the medication, please consult your attending physician." The first strategy is mainly applicable to scenarios where the conflict level is low-risk or medium-risk and the conflict can be eliminated through rewriting. When the target correction strategy is the preset second strategy, the backend intelligent system performs intervention downgrade processing on the security conflict sub-data to obtain the second verification sub-data. This intervention downgrade process does not retain or rewrite the content of the security conflict sub-data. Instead, it directly removes that part of the content from the health management information to be output, or replaces it with a preset standard safety prompt statement. For example, it replaces the entire sentence that triggered the "add prescription drug suggestion" conflict with "Regarding medication use, please follow your doctor's instructions. It is not recommended to add any medications on your own." The second strategy is mainly applicable to scenarios where the conflict level is high-risk or where rewriting the content cannot effectively eliminate the security risk. The second verification sub-data obtained after the above correction process is a secure version of the security conflict sub-data after correction or downgrade. It, together with the first verification sub-data, constitutes the complete verification management information.

[0046] After obtaining the verification management information, the backend intelligent system identifies the corresponding intervention type and level, and then performs tiered intervention interactions based on the intervention type and level using a preset tiered intervention interaction method. The identification of the intervention type is accomplished by analyzing the content of the verification management information: if the verification management information contains explicit reminders related to medication time, dosage, and frequency, the intervention type is identified as "medication reminder"; if the verification management information contains risk warning statements, the intervention type is identified as "risk warning"; if the verification management information contains a suggestion requiring the target user to remeasure a certain physiological parameter within a preset time, the intervention type is identified as "retest reminder"; if the verification management information contains a suggestion for the target user to visit a medical institution, the intervention type is identified as "medical treatment suggestion"; if the verification management information contains general explanations of health status, lifestyle suggestions, or medical encyclopedia recommendations, the intervention type is identified as "health advice". Intervention levels are identified by analyzing the risk level markers and urgency levels in the verification management information: if the verification management information contains a high-risk marker (such as "suspected acute myocardial infarction risk") or a mandatory emergency medical attention prompt, the intervention level is identified as "high-level"; if the verification management information contains a medium-risk marker or content suggesting medical attention but not urgently, the intervention level is identified as "medium-level"; if the verification management information contains a low-risk marker or only general health advice, the intervention level is identified as "low-level". After the backend intelligent system completes the identification of the intervention type and level, it executes the tiered intervention interaction according to the identified intervention type and level through a preset tiered intervention interaction method.The specific implementation method of this tiered intervention interaction is as follows: When the intervention type is "medication reminder" and the intervention level is medium, the backend intelligent system outputs medication reminder information through the mobile application via in-app notifications and message reminders, instructing the target user to take medication on time; when the intervention type is "risk warning" and the intervention level is high, the backend intelligent system forcibly pops up a full-screen warning pop-up window through the mobile application, accompanied by continuous vibration and prompt sound. At the same time, the backend intelligent system automatically sends an SMS or in-app message containing the target user's identification and risk summary to the emergency contact information preset by the target user, in order to seek timely external assistance. When the intervention type is "retest reminder" and the intervention level is low, the backend intelligent system reminds the target user to remeasure relevant physiological parameters within a specified time through a lightweight message notification via the mobile application. When the intervention type is "medical advice" and the intervention level is medium, the backend intelligent system displays information about nearby medical institutions and medical guidance to the target user through a combination of in-app pop-ups and message notifications via the mobile application. When the intervention type is "health advice" and the intervention level is low, the backend intelligent system displays health advice content and entry points to relevant medical encyclopedia entries in the form of information cards on the main interface of the mobile application. For example, a target user taking antihypertensive medication might have a medication reminder in their verification management information that reads, "Take nifedipine controlled-release tablets once today, 30 mg each time, at 8:00 AM," and a retest reminder that "Systolic blood pressure has significantly increased compared to the individualized health baseline; it is recommended to monitor blood pressure for three consecutive days." The backend intelligent system identifies the intervention type as "medication reminder" and "retest reminder," and the intervention level as "medium level." Subsequently, at 7:30 AM, it sends an in-app notification to the target user via mobile application, containing the medication time, drug name, and dosage, with an additional prompt: "Please measure your blood pressure two hours after taking this medication and record the result." After the above complete security verification, correction processing, and tiered intervention interaction, the backend intelligent system completes the entire closed loop from the generation of initial health management information to the safe and compliant output of intervention reminders, ensuring that all health management information ultimately delivered to the target user meets the system design requirements in terms of medical safety, individual targeting, and interactive effectiveness.

[0047] like Figure 3 The diagram shown is a functional block diagram of a health risk assessment and intervention reminder device provided in an embodiment of this application.

[0048] In some embodiments, the health risk assessment and intervention reminder device 2 may include multiple functional modules composed of computer program segments. The computer programs for each program segment in the health risk assessment and intervention reminder device 2 may be stored in the server's memory and executed by at least one processor to perform (see details). Figure 2 (Description) Functions of health risk assessment and intervention reminder methods.

[0049] In this embodiment, the health risk assessment and intervention reminder device 2 can be divided into multiple functional modules according to its functions. These functional modules may include: a data optimization module 21, a knowledge matching module 22, an analysis and evaluation module 23, and a hierarchical interaction module 24. The module referred to in this invention is a series of computer program segments that can be executed by at least one processor and perform a fixed function, stored in memory. In this embodiment, the functions of each module will be detailed in subsequent embodiments.

[0050] Data optimization module 21 is used to preprocess the received physiological parameter data and personal health information to generate a health status sequence; The knowledge matching module 22 is used to identify health risk events in the health status sequence, and to perform matching retrieval based on the health status sequence and the health risk events to obtain relevant medical knowledge entries; The analysis and evaluation module 23 is used to respond to user requests, analyze and evaluate the health status sequence, the health risk events and the relevant medical knowledge items, and obtain initial health management information. The hierarchical interaction module 24 is used to perform security verification on the initial health management information to obtain verification management information, and to perform multi-level intervention interaction based on the verification management information.

[0051] It should be understood that the various variations and specific embodiments of the methods provided in the above embodiments are also applicable to the health risk assessment and intervention reminder device of this embodiment. Through the foregoing detailed description of the health risk assessment and intervention reminder method, those skilled in the art can clearly understand the implementation method of the health risk assessment and intervention reminder device of this embodiment. For the sake of brevity, it will not be described in detail here.

[0052] like Figure 4 The diagram shown is a structural schematic of an electronic device provided in an embodiment of this application.

[0053] In a preferred embodiment of the present invention, the electronic device 3 may include, but is not limited to, a memory 31, at least one processor 32, and at least one communication bus 33.

[0054] Those skilled in the art should understand that Figure 4 The structure of the electronic device 3 shown does not constitute a limitation of the embodiments of the present invention. The electronic device 3 may also include more or fewer other hardware or software than shown, or different component arrangements.

[0055] In some embodiments, the electronic device 3 is a device capable of automatically performing numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes, but is not limited to, microprocessors, application-specific integrated circuits, programmable gate arrays, digital processors, and embedded devices.

[0056] It should be noted that the electronic device 3 is merely an example. Other existing or future electronic products that are suitable for this application should also be included within the scope of protection of this application and are incorporated herein by reference.

[0057] In some embodiments, the memory 31 stores a computer program that, when executed by the at least one processor 32, implements all or part of the steps in the health risk assessment and intervention reminder method described above. The memory 31 includes read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data. Further, the computer-readable storage medium may primarily include a program storage area and a data storage area, wherein the program storage area may store an operating system, at least one application program required for a function, etc.

[0058] In some embodiments, the at least one processor 32 is the control unit of the electronic device 3, connecting various components of the electronic device 3 via various interfaces and lines. It executes programs or modules stored in the memory 31 and calls data stored in the memory 31 to perform various functions and process data. For example, when the at least one processor 32 executes a computer program stored in the memory 31, it implements all or part of the steps of the health risk assessment and intervention reminder method described in this application embodiment; or it implements all or part of the functions of the health risk assessment and intervention reminder device. The at least one processor 32 may be composed of integrated circuits, such as a single-packaged integrated circuit or multiple integrated circuits with the same or different functions, including combinations of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips.

[0059] In some embodiments, the at least one communication bus 33 is configured to enable communication between the memory 31 and the at least one processor 32, etc. Although not shown, the electronic device 3 may also include a power supply (e.g., a battery) to power the various components. Preferably, the power supply can be logically connected to the at least one processor 32 via a power management device, thereby enabling functions such as charging, discharging, and power consumption management. The power supply may also include one or more DC or AC power supplies, recharging devices, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components. The electronic device 3 may also include various sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be described in detail here.

[0060] The integrated unit implemented as a software functional module described above can be stored in a computer-readable storage medium. This software functional module, stored in a storage medium, includes several instructions to cause an electronic device (which may be a personal computer, electronic device, or network device, etc.) or processor to execute portions of the methods described in the various embodiments of this application.

[0061] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0062] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0063] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A method for health risk assessment and intervention reminders, characterized in that, The method includes: The received physiological parameter data and personal health information are preprocessed to generate a health status sequence; Identify health risk events in the health status sequence, and perform matching retrieval based on the health status sequence and the health risk events to obtain relevant medical knowledge entries; In response to user requests, the system integrates the health status sequence, the health risk events, and the relevant medical knowledge entries for analysis and evaluation to obtain initial health management information. The initial health management information is security verified to obtain verification management information, and multi-level intervention interactions are performed based on the verification management information.

2. The health risk assessment and intervention reminder method according to claim 1, characterized in that, The step of preprocessing the received physiological parameter data and personal health information to generate a health status sequence includes: Real-time collection of physiological detection data from target users, followed by data cleaning and standardization to obtain a physiological parameter dataset; The user status features are extracted from the received health description information, and the user status features are structured to obtain individual health information data. By using a preset time window, the physiological parameter dataset and the individual health information data are time-series correlated and fused to generate a health status sequence.

3. The health risk assessment and intervention reminder method according to claim 1, characterized in that, The process of identifying health risk events in the health status sequence and performing matching retrieval based on the health status sequence and the health risk events to obtain relevant medical knowledge entries includes: The health status sequence is subjected to time-series feature extraction processing to obtain an individualized health baseline; Based on a pre-defined medical rule base and the individualized health baseline, risk analysis is performed on the health status sequence to identify health risk events. A retrieval feature vector is constructed based on the health risk events and the health status sequence, and semantic matching retrieval is performed on a preset medical encyclopedia knowledge base based on the retrieval feature vector to obtain relevant medical knowledge entries.

4. The health risk assessment and intervention reminder method according to claim 3, characterized in that, The step of performing risk analysis on the health status sequence based on a preset medical rule base and the individualized health baseline to identify health risk events includes: Read the physiological parameter type corresponding to the individualized health baseline, and match the corresponding target medical rule base and target statistical threshold from the preset medical rule base according to the physiological parameter type; The current physiological parameter values ​​in the health status sequence are compared with the individualized health baseline to obtain the baseline deviation, and the baseline deviation is compared with the target statistical threshold to obtain the first risk feature; The physiological parameter change trends in the health status sequence are matched with the target medical rule base to obtain the second risk feature; Based on the first risk feature and the second risk feature, health risk events are identified from the health status sequence.

5. The health risk assessment and intervention reminder method according to claim 1, characterized in that, The fusion process analyzes and evaluates the health status sequence, the health risk events, and the relevant medical knowledge items to obtain initial health management information, including: The health status sequence, the health risk events, and the relevant medical knowledge items are fused according to a preset data format to generate data to be analyzed. The data to be analyzed is evaluated based on the received user request in order to filter out initial health management information.

6. The health risk assessment and intervention reminder method according to claim 1, characterized in that, The step of performing security verification on the initial health management information to obtain verification management information, and performing multi-level intervention interactions based on the verification management information, includes: The initial health management information is security verified according to preset multi-level security rules to obtain the first verification sub-data. The security conflict sub-data is corrected by a preset correction mechanism to obtain the second verification sub-data; Identify the intervention type and intervention level corresponding to the first verification sub-data and / or the second verification sub-data, and perform graded intervention interaction according to the intervention type and the intervention level through a preset graded intervention interaction method.

7. The health risk assessment and intervention reminder method according to claim 6, characterized in that, The step of correcting the security conflict sub-data through a preset correction mechanism to obtain the second verification sub-data includes: The conflict type is identified for the security conflict sub-data, and the conflict level corresponding to the security conflict sub-data is determined. Based on the conflict level, a target correction strategy corresponding to the conflict level is selected from the preset correction mechanisms; When the target correction strategy is the preset first strategy, the security conflict sub-data is subjected to content correction processing to obtain the second verification sub-data; When the target correction strategy is the preset second strategy, the security conflict sub-data is intervened and downgraded to obtain the second verification sub-data.

8. A health risk assessment and intervention reminder device, applied to the health risk assessment and intervention reminder method of claim 1, characterized in that, The device includes: The data optimization module is used to preprocess the received physiological parameter data and personal health information to generate a health status sequence; The knowledge matching module is used to identify health risk events in the health status sequence, and to perform matching retrieval based on the health status sequence and the health risk events to obtain relevant medical knowledge entries; The analysis and evaluation module is used to respond to user requests, analyze and evaluate the health status sequence, the health risk events, and the relevant medical knowledge items, and obtain initial health management information. The hierarchical interaction module is used to perform security verification on the initial health management information to obtain verification management information, and to perform multi-level intervention interactions based on the verification management information.

9. An electronic device, characterized in that, The electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the health risk assessment and intervention reminder method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the health risk assessment and intervention reminder method according to any one of claims 1 to 7.

Images