Health portrait generation and dynamic monitoring method and system based on artificial intelligence

Abstract

The invention discloses a health portrait generation and dynamic monitoring method and system based on artificial intelligence, and the method comprises the steps: obtaining multi-source health data of a user, and carrying out the preprocessing of data cleaning, missing value filling, standardization, semantic analysis and the like; feature extraction and label generation are carried out according to the preprocessed user multi-source health data, and a structured health portrait is constructed; based on the health portrait of the user, in combination with a preset medical knowledge base and a risk rule base, performing early warning on an abnormal trend; and generating personalized suggestions for the user according to the health portrait. The system comprises a data acquisition unit, a portrait construction unit and an early warning unit. According to the invention, the health portrait which comprehensively reflects the health condition of the user can be constructed, and health monitoring can be realized. The method can be widely applied to the field of health monitoring.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of health monitoring, in particular to a health portrait generation and dynamic monitoring method and system based on artificial intelligence. BACKGROUND

[0002] With the continuous improvement of enterprise artificial intelligence big data general model technology accumulation, the gradual deepening of the digital construction of hospitals and other health institutions, the outline of AI medical industry development gradually emerges. In recent years, various types of digital health management tools have emerged in the market, such as hospital information systems, personal health monitoring applications and wearable smart devices, and health management products have gradually become popular. However, most of the existing health management service modes are passive response types, and their core functions are concentrated in the diagnosis and treatment links after the occurrence of diseases, lacking the ability of active discovery and early intervention of health risks, and often intervening only after the user has obvious symptoms, thus missing the best health management opportunity. In addition, most of the current health management applications only output charts based on a single data source (such as a sphygmomanometer or a bracelet), and fail to integrate medical records, physical examination reports, wearable devices, and multi-modal data such as inquiry voice, making it difficult to form a comprehensive personal health portrait.

[0003] Therefore, there is an urgent need for a health monitoring method that integrates multi-source data. SUMMARY

[0004] Therefore, in order to solve the technical problem that most of the existing health dynamic monitoring methods are based on data from a single source (such as physical examination reports or wearable devices), which leads to a one-sided health portrait, the present application proposes a health portrait generation and dynamic monitoring method based on artificial intelligence, which includes the following steps: Obtain user multi-source health data and perform data cleaning, missing value filling, standardization and semantic analysis, etc. According to the pre-processed user multi-source health data, perform feature extraction and label generation, and construct a structured health portrait. Based on the user's health portrait, combine the preset medical knowledge base and risk rule base to warn of abnormal trends. Generate personalized recommendations for users according to the health portrait.

[0005] The present application also proposes a health portrait generation and dynamic monitoring system based on artificial intelligence, which is applied to the above method and includes a data acquisition unit, a portrait construction unit and a warning unit.

[0006] Based on the above solutions, this invention provides an artificial intelligence-based method and system for generating and dynamically monitoring health profiles. It maps multimodal information uniformly into a structured profile framework, achieving a panoramic depiction of an individual's health status. Compared to traditional single-indicator or single-dimensional display methods, the health profiles of this invention have higher coverage and representativeness. Furthermore, based on the features of the health profiles and combined with the knowledge reasoning capabilities of large models, targeted intervention plans are generated for different users. Through the entire process of "health profile generation—dynamic monitoring—risk warning—personalized intervention—profile re-updating," a true shift from passive diagnosis and treatment to proactive health management is achieved, improving the continuity and systematic nature of health management. Attached Figure Description

[0007] Figure 1 This is a flowchart of the steps of a health profile generation and dynamic monitoring method based on artificial intelligence according to the present invention; Figure 2 This is a structural block diagram of a health profile generation and dynamic monitoring system based on artificial intelligence, according to the present invention. Detailed Implementation

[0008] 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.

[0009] It should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described in this application can be combined with each other.

[0010] It should be understood that the terms "system," "apparatus," "unit," and / or "module" used in this application are a method of distinguishing different components, elements, parts, sections, or assemblies at different levels. However, if other terms can achieve the same purpose, they may be replaced by other expressions.

[0011] As indicated in this application and claims, unless the context clearly indicates otherwise, the words "a," "an," "a," and / or "the" are not specifically singular and may include the plural. Generally, the terms "comprising" and "including" only indicate the inclusion of expressly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements. An element defined by the phrase "comprising an..." does not exclude the presence of other identical elements in the process, method, product, or apparatus that includes the element.

[0012] In the description of the embodiments of this application, "a plurality of" refers to two or more. The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature.

[0013] Furthermore, flowcharts are used in this application to illustrate the operations performed by the system according to embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed precisely in sequence. Instead, the steps can be processed in reverse order or simultaneously. Additionally, other operations can be added to these processes, or one or more steps can be removed from them.

[0014] Reference Figure 1 This is a flowchart illustrating an optional example of the AI-based health profile generation and dynamic monitoring method proposed in this invention. The method can be applied to computer devices, and the dynamic monitoring method proposed in this embodiment may include, but is not limited to, the following steps: Step S1: Obtain user's multi-source health data; Step S2: Map user's multi-source health data into a structured health profile; Step S3: Based on the structured health profile and combined with preset risk rules, issue early warnings for abnormal trends; Step S4: Personalized suggestions.

[0015] In some feasible embodiments, the sources of the user's multi-source health data in step S1 include electronic medical records, physical examination data, imaging data, physiological indicators (heart rate, blood pressure, blood sugar) collected by wearable devices, daily life behavior data (exercise, sleep, diet), and user self-reports (voice or text consultation).

[0016] In some feasible embodiments, step S1 further includes: The user's multi-source health data is cleaned, standardized, and semantically parsed.

[0017] Data cleaning includes missing value imputation and outlier removal; standardization includes unit unification and time-to-time standardization; semantic parsing involves constructing semantic relationships between data using a large language model.

[0018] For missing values ​​in a user's multi-source health data, the imputation method is selected as follows: For single-point or very small gaps (less than 3 consecutive points): linear interpolation or spline interpolation is a safe choice.

[0019] Medium gap (3-10 points): More advanced methods are needed, such as time series forecasting (ARIMA / Prophet). The natural cycles of the data (such as heart rate and sleep circadian rhythms) must be considered.

[0020] Large missing blocks (more than 10 consecutive points): No interpolation is used. The "missing" status itself is treated as a feature label, which may contain important information (such as device detachment, user illness).

[0021] In some feasible embodiments, step S2 specifically includes: S2.1, Perform feature extraction on user multi-source health data, including: Static characteristics: age, sex, height, genetic history (family history), basal metabolic rate, etc. These are usually relatively stable and slowly changing baseline information.

[0022] Time-series data characteristics: Statistical characteristics: Central tendency: mean, median; Dispersion: standard deviation, variance, coefficient of variation, range (maximum value - minimum value); Examples: average resting heart rate over the past 7 days, standard deviation of daytime systolic blood pressure (reflecting blood pressure variability), average daily steps. Time-period characteristics: Divide the day into different time periods (e.g., sleep, early morning, daytime, nighttime) and calculate characteristics separately; Examples: average nighttime heart rate, morning blood pressure, afternoon blood glucose peak. Circadian rhythm and pattern characteristics: Sleep: Extract total sleep duration, deep sleep / light sleep / REM ratio, sleep onset time, number of awakenings, and sleep efficiency (total sleep time / bedtime) using accelerometer and heart rate data; Activity: duration of moderate to high intensity activity, longest continuous sedentary time, and calorie expenditure during activity; Heart rate: resting heart rate, heart rate recovery rate after exercise (an important cardiovascular health indicator). Physiological rhythm and pattern characteristics: Sleep: Total sleep duration, deep sleep / light sleep / REM ratio, sleep onset time, number of awakenings, and sleep efficiency (total sleep time / time in bed) were extracted using accelerometer and heart rate data; Activity: Duration of moderate-to-vigorous intensity activity, longest continuous sedentary time, and calorie expenditure during activity; Heart rate: Resting heart rate and post-exercise heart rate recovery rate (important cardiovascular health indicators). Event and trend characteristics: Abnormal event count: Number of days with excessively high heart rate (>100 bpm), number of weekly hypoglycemic events; Trend: The rate of change in average steps taken over the past 7 days compared to the previous 7 days was calculated using a sliding window, and monthly weight trends (increasing / stable / decreasing).

[0023] Clinical data features: Named entity recognition: using NLP models to identify entities such as diseases, symptoms, drugs, and surgeries; Sentiment analysis: analyzing users' self-reported health notes to determine their emotional state (such as positive, anxious, negative).

[0024] S2.2 Generate rule-based labels and machine learning model-based labels; Rule-based labeling is suitable for scenarios where medical knowledge is clear and logic is well-defined. Its advantages include strong interpretability and ease of physician review. Single-dimensional rule: Label: Blood Pressure Level: Mean systolic blood pressure < 120 and mean diastolic blood pressure < 80 are defined as "normal"; mean systolic blood pressure between 130-139 or mean diastolic blood pressure between 80-89 is defined as "high normal"; mean systolic blood pressure >= 140 or mean diastolic blood pressure >= 90 is defined as "hypertension". Multi-dimensional composite rule: Label: Metabolic Syndrome Risk: Meeting at least three of the following criteria—excessive waist circumference, elevated triglycerides, low HDL-C, high blood pressure, and high fasting blood glucose—is defined as "high risk".

[0025] Machine learning-based labeling is suitable for scenarios with complex patterns that require learning patterns from data. Classification models include: Fall risk level (high / medium / low): trained using features such as gait speed, gait variability, muscle strength data, and past fall history; Sleep quality classification (good / average / poor): using features such as sleep duration, deep sleep percentage, number of awakenings, and HRV; Depressive tendency (yes / no): combining activity level, sleep patterns, app usage behavior, and questionnaire scores. Clustering models include: Health behavior clustering: clustering users into categories such as "health-conscious individuals," "weekend warriors," "sedentary office workers," and "sleep-disordered individuals" based on activity levels, sleep patterns, and dietary records; Anomaly detection models include: Abnormal physiological signals: training a model using historical heart rate data to automatically label "abnormal heart rate pattern" when persistent abnormal fluctuations occur; Behavioral mutations: users who have consistently exercised for a long time and then suddenly experience a sharp drop in activity level for a week can be labeled as having "significantly decreased activity level."

[0026] S2.3. Bind the user's multi-source health data and the tags to the corresponding users to generate a structured health profile.

[0027] Define the tagging system: First-level categories: basic information, health status, health risks, behavioral habits, psychosocial issues, etc.; Second-level categories (under "health status"): chronic diseases, physiological indicators, physical fitness, etc.; Specific tags (under "chronic diseases"): hypertension, type 2 diabetes, coronary heart disease, etc.

[0028] The core of a user’s health profile is to generate a dynamic, structured set of tags for each user.

[0029] Furthermore, the health profile is not a static result, but is constantly updated with the input of new data; the system constructs a "time series-profile node" link to support trend modeling of individual health status; and utilizes the reasoning capabilities of large models to identify potential changes in health status (such as "continuous increase in blood pressure + insufficient sleep → risk of prehypertension").

[0030] Existing technologies generally lack dynamic updating capabilities, and health profiles are usually static results that fail to reflect health trends over time. This invention proposes a time-series-based health profile evolution mechanism. When new data is input, the system automatically updates the health profile and generates historical trajectories and trend prediction curves. This mechanism ensures that the health profile reflects changes in an individual's health status in real time, avoiding the influence of "outdated data" on user judgment.

[0031] In some feasible embodiments, step S4 specifically includes: Intervention recommendations are personalized, for example: For individuals at high risk of diabetes: a low-sugar diet and 30 minutes of brisk walking daily are recommended. For individuals with sleep disorders: We recommend adjusting behavioral habits and intervening in their sleep schedules.

[0032] The intervention effect can be adjusted again based on user feedback and new data, forming a closed loop.

[0033] In some feasible embodiments, it also includes: Streaming technology is used to update the health profile in real time.

[0034] Real-time data is sent to Kafka, a high-throughput, distributed, persistent message queue that decouples data producers and consumers and buffers peak traffic.

[0035] Apache Flink, as a stream processing engine, continuously consumes data from Kafka topics. Internally, a Flink Job implements a series of complex business logics, performing window aggregation, feature extraction, and model inference on the data stream, ultimately generating or updating the user's health tag.

[0036] The processing results (such as newly calculated tags) are written back to another Kafka Topic, or directly to downstream profile storage (such as Redis for real-time queries, Elasticsearch for searching, and graph databases for relational queries).

[0037] like Figure 2 As shown, an artificial intelligence-based health profile generation and dynamic monitoring system includes: The data acquisition unit is used to execute step S1; A profile building unit is used to execute step S2; The early warning unit is used to execute step S3; A personalized suggestion unit is used to execute step S4.

[0038] The content of the above method embodiments is applicable to this system embodiment. The specific functions implemented in this system embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0039] A health profile generation and dynamic monitoring device based on artificial intelligence: At least one processor; At least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor implements the AI-based health profile generation and dynamic monitoring method described above.

[0040] The content of the above method embodiments is applicable to the device embodiments. The specific functions implemented by the device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0041] A storage medium storing processor-executable instructions, which, when executed by a processor, are used to implement the AI-based health profile generation and dynamic monitoring method described above.

[0042] The content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0043] The above is a detailed description of the preferred embodiments of the present invention. However, the present invention is not limited to the embodiments described. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention. All such equivalent modifications or substitutions are included within the scope defined by the claims of this application.

Claims

1. A method for generating and dynamically monitoring health profiles based on artificial intelligence, characterized in that, Includes the following steps: Acquire user health data from multiple sources; Map the user's multi-source health data into a structured health profile; Based on the structured health profile and combined with preset risk rules, early warnings are issued for abnormal trends.

2. The method for generating and dynamically monitoring health profiles based on artificial intelligence according to claim 1, characterized in that, The sources of the user's multi-source health data include electronic medical records, physical examination data, imaging data, physiological indicators from wearable devices, behavioral data, and customized consultation data.

3. The method for generating and dynamically monitoring health profiles based on artificial intelligence according to claim 2, characterized in that, Also includes: The user's multi-source health data is cleaned, standardized, and semantically parsed.

4. The method for generating and dynamically monitoring health profiles based on artificial intelligence according to claim 2, characterized in that, Also includes: Intervention recommendations are generated based on the early warning.

5. The method for generating and dynamically monitoring health profiles based on artificial intelligence according to claim 1, characterized in that, The step of mapping the user's multi-source health data into a structured health profile specifically includes: Feature extraction is performed on the user's multi-source health data to obtain basic static features, time-series data features, and clinical data features; Based on preset rules and pre-trained models, labels are generated according to the basic static features, the time-series data features, and the clinical data. The user's multi-source health data and the tags are bound to the corresponding users to generate a structured health profile.

6. The method for generating and dynamically monitoring health profiles based on artificial intelligence according to claim 5, characterized in that, Also includes: If it is determined that there are missing values ​​of consecutive points less than the first threshold in the user's multi-source health data, they are filled using a linear interpolation method; If it is determined that there are missing values ​​of consecutive points in the user's multi-source health data that are greater than a first threshold and less than a second threshold, the time series model is used to predict the value of the missing points, and the missing points are filled in according to the prediction results; If it is determined that there are missing values ​​of consecutive points greater than the second threshold in the user's multi-source health data, no padding is performed and a missing label is added.

7. The method for generating and dynamically monitoring health profiles based on artificial intelligence according to claim 6, characterized in that, Also includes: Streaming technology is used to update the health profile in real time.

8. A health profile generation and dynamic monitoring system based on artificial intelligence, characterized in that, include: The data acquisition unit is used to acquire multi-source health data from users. A profile building unit is used to map the user's multi-source health data into a structured health profile; The early warning unit generates early warnings for abnormal trends based on the structured health profile and in conjunction with preset risk rules.

9. A health profile generation and dynamic monitoring device based on artificial intelligence, characterized in that, include: At least one processor; At least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor implements the AI-based health profile generation and dynamic monitoring method as described in any one of claims 1-7.

Images