Personalized health follow-up visit method, system and equipment based on health portrait and large language model and medium

Abstract

The invention relates to the technical field of personalized health follow-up visit, in particular to a personalized health follow-up visit method, system and device based on a health portrait and a large language model and a medium, and the method comprises the steps: collecting multi-dimensional health data of a user, carrying out the standardization processing and feature extraction of the health data, and generating a plurality of user feature tags; based on the user feature tag, constructing a dynamically updated user health portrait; inputting the user health portrait into a service classification model, and determining a target follow-up visit service line and a follow-up visit mode corresponding to the current user; on the basis of the target follow-up visit service line, the follow-up visit mode and the user health portrait, a prompt word is constructed and input to a large language model, so that personalized follow-up visit content is generated; and based on the follow-up visit content, executing a follow-up visit task on the user through the intelligent call-out system. The method has the beneficial effects that the follow-up visit efficiency and coverage are remarkably improved, and refined, dynamic and humanized health management of each user is really realized.

Description

Technical Field

[0001] This invention relates to the field of personalized health follow-up technology, and in particular to a personalized health follow-up method, system, device and medium based on health profile and large language model. Background Technology

[0002] With the continuous advancement of information technology and artificial intelligence, the field of health management is gradually evolving towards personalization and intelligence. Traditional health follow-up methods mainly rely on manual telephone or face-to-face communication, which is inefficient and difficult to implement on a large scale. In recent years, some systems have attempted to introduce data analysis and automation tools to improve follow-up efficiency.

[0003] However, existing technologies still have many shortcomings in practical applications: First, they lack sufficient data integration and a unified modeling and dynamic updating mechanism for multi-source and multi-dimensional health information, resulting in one-sided and static patient profiles; second, current follow-up strategies have limited personalization, with most systems still relying on predefined rules to generate questionnaires or reminders, making it difficult to dynamically adjust follow-up content and methods based on the patient's real-time status; third, traditional automated outbound calling systems have rigid dialogue, low levels of human-computer interaction intelligence, and are unable to naturally understand and respond to patient intentions, and lack the ability to analyze and learn from new data acquired during follow-up, preventing health management from achieving a true closed loop. Summary of the Invention

[0004] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In the first aspect, the present invention provides a personalized health follow-up method based on health profiles and large language models, including collecting multi-dimensional health data of users, and performing standardized processing and feature extraction on the health data to generate multiple user feature tags; Based on user feature tags, construct dynamically updated user health profiles; Input the user's health profile into the business classification model to determine the target follow-up business line and follow-up method for the current user; Based on the target follow-up business line, follow-up method and user health profile, prompt words are constructed and input into the large language model to generate personalized follow-up content; Based on the follow-up content, follow-up tasks are performed on users through an intelligent outbound calling system; Collect user feedback data during the follow-up process, analyze the feedback data, and extract key information; Based on key information, update users' health profiles and user characteristic tags to optimize subsequent follow-up strategies.

[0005] As a preferred embodiment of the personalized health follow-up method based on health profiling and large language models of the present invention, wherein: multiple user feature tags are generated, including, The system uses a rule engine and machine learning algorithms to extract features and cross-validate standardized health data, generating labels to identify health risk levels, disease types, or lifestyles.

[0006] As a preferred embodiment of the personalized health follow-up method based on health profiling and large language models of the present invention, wherein: generating personalized follow-up content includes, The context of building a large language model based on user health profiles, combined with content templates and prompts that match the target follow-up business lines and follow-up methods; Generate complete follow-up dialogue content that includes an opening, a brief description of the patient's condition, a questionnaire assessment, rehabilitation guidance, and a concluding reminder.

[0007] As a preferred embodiment of the personalized health follow-up method based on health profiling and large language models of the present invention, the method includes: performing follow-up tasks on users through an intelligent outbound calling system, including: The text-based follow-up content is converted into speech information using speech synthesis technology; Establish a voice communication connection with the user through an automatic dialing system to perform follow-up.

[0008] As a preferred embodiment of the personalized health follow-up method based on health profiling and large language models of the present invention, the method includes: collecting user feedback data during the follow-up process, parsing the feedback data, and extracting key information, including... During the intelligent outbound calling process, the voice recognition system receives user voice feedback in real time and converts it into text data. Natural language processing is performed on the converted text data to extract key information related to health status, symptom changes, or emotional tendencies.

[0009] As a preferred embodiment of the personalized health follow-up method based on health profiles and large language models of the present invention, the business classification model is a model trained based on machine learning algorithms, used to map users to predefined follow-up business lines, and the follow-up business lines are divided according to the user's health risk level.

[0010] As a preferred embodiment of the personalized health follow-up method based on health profiling and large language model of the present invention, the multi-dimensional health data includes at least one of basic information, medical history information, lifestyle information and psychological state information.

[0011] Secondly, the present invention provides a personalized health follow-up system based on health profiles and large language models, including: a collection and extraction module, used to collect multi-dimensional health data of users, and to perform standardized processing and feature extraction on the health data to generate multiple user feature tags; The first building module is used to construct dynamically updated user health profiles based on user feature tags; The determination module is used to input the user's health profile into the business classification model to determine the target follow-up business line and follow-up method for the current user. The second construction module is used to build prompt words based on the target follow-up business line, follow-up method and user health profile and input them into the large language model to generate personalized follow-up content. The execution module is used to perform follow-up tasks on users through an intelligent outbound calling system based on the follow-up content. The data collection and analysis module is used to collect user feedback data during the follow-up process, analyze the feedback data, and extract key information. The update module is used to update the user's health profile and user characteristic tags based on key information, in order to optimize subsequent follow-up strategies.

[0012] Thirdly, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method described above.

[0013] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the method described above.

[0014] Compared with existing technologies, the beneficial effects of this invention are as follows: By automatically collecting and deeply analyzing users' multidimensional health data, a precise dynamic health profile is constructed, and the most suitable follow-up strategy is automatically matched for patients with different risk levels based on an intelligent classification model; highly personalized follow-up dialogue content is generated using a large language model, covering the entire process from disease review to rehabilitation guidance; efficient and natural proactive communication is achieved through intelligent outbound calling and voice recognition technology, and patient feedback is analyzed in real time, forming a closed-loop learning mechanism of "execution-feedback-optimization". While significantly improving follow-up efficiency and coverage, it truly realizes refined, dynamic, and humanized health management for each user. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a flowchart illustrating a personalized health follow-up method based on health profiling and a large language model. Detailed Implementation

[0017] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0018] Example 1, referring to Figure 1 This is the first embodiment of the present invention, which provides a personalized health follow-up method based on health profiles and large language models, including: S100: Collects multi-dimensional health data from users, performs standardized processing and feature extraction on the health data, and generates multiple user feature tags; S200: Based on user feature tags, construct dynamically updated user health profiles; S300: Input the user health profile into the business classification model to determine the target follow-up business line and follow-up method for the current user; S400: Based on the target follow-up business line, follow-up method and user health profile, construct prompt words and input them into the large language model to generate personalized follow-up content; S500: Based on follow-up content, perform follow-up tasks to users through an intelligent outbound calling system; S600: Collects user feedback data during the follow-up process, analyzes the feedback data, and extracts key information; S700: Update the user's health profile and user characteristic tags based on key information to optimize subsequent follow-up strategies.

[0019] It should be noted that in health follow-up scenarios such as chronic disease management and postoperative rehabilitation, traditional follow-up methods heavily rely on manual telephone or face-to-face communication, resulting in inefficiency, limited coverage, and difficulty in personalization. Furthermore, due to the dynamic changes in patients' health conditions, the lack of a unified integration and intelligent analysis mechanism often leads to follow-up content that lacks specificity and fails to respond to the patient's latest condition in real time. In addition, existing systems are insufficient in natural language interaction, intelligent feedback parsing, and continuous learning optimization, making it difficult to achieve large-scale, precise, and automated health management services.

[0020] Therefore, addressing the issues of low efficiency, lack of personalization, and lack of intelligence in the aforementioned health follow-up processes, this method, through steps S100-S700, firstly constructs a dynamically updated user health profile through multi-dimensional data collection and feature extraction, achieving accurate characterization of the patient's condition; secondly, it determines personalized follow-up strategies through a classification model and generates natural, fluent, and patient-appropriate follow-up content using a large language model; furthermore, it achieves automatic interaction and feedback collection by leveraging intelligent outbound calling and speech recognition technologies; finally, it continuously optimizes the profile and strategies through feedback data, realizing self-iteration and upgrading of the follow-up process, significantly improving the efficiency, accuracy, and patient satisfaction of follow-up.

[0021] Example 2, refer to Figure 1 This is one embodiment of the present invention. Based on the above embodiment, a personalized health follow-up method based on health profiles and large language models is provided.

[0022] In this embodiment of the application, step S100 involves collecting multi-dimensional health data from users, standardizing the health data, extracting features, and generating multiple user feature tags, including the following steps A1-A2: A1: Multidimensional health data includes at least one of the following: basic information, medical history information, lifestyle information, and psychological state information; It is understandable that medical history information includes, but is not limited to, biochemical indicators, examination results and medical imaging conclusions, follow-up records, medical history records (family medical history) and health questionnaires. Regularly updating health data can provide a foundation for subsequent data analysis and user profile building.

[0023] It should be noted that the standardization of health data includes: ① Data cleaning and invalidation: used to identify and correct data that does not conform to medical common sense or numerical logic (such as negative age, blood pressure values ​​exceeding the reasonable range, etc.) and to merge or delete duplicate data caused by system synchronization or multiple entries; ② Data format standardization: used to convert date and time in different formats (such as 2024-05-27, 27 / 05 / 2024, May27,2024) into a standard format (such as YYYY-MM-DDHH:mm:ss), convert clinical indicators into a unified unit of measurement, such as unifying blood pressure to mmHg, blood glucose values ​​to mmol / L or mg / dL, and performing basic natural language processing (NLP) such as word segmentation and stop word removal on unstructured text information (such as chief complaint, doctor's notes) to prepare for subsequent feature extraction; ③ Medical terminology and code standardization: used to standardize diagnostic results. ④ Textual descriptions of drug names, surgical procedures, etc., are mapped to standard medical dictionaries or coding systems (such as ICD-10 (disease codes), ATC (drug codes), LOINC (laboratory test codes), etc.) to eliminate ambiguity caused by differences in terminology; ⑤ Data discretization and segmentation: Continuous clinical indicators (such as age, BMI index, blood glucose level) are divided into clinically meaningful intervals (for example, age is divided into "youth, middle-aged, elderly", and BMI is divided into "underweight, normal, overweight, obese"), and categorical variables (such as gender, smoking status) are converted into numerical or One-Hot coding forms to facilitate processing by machine learning models; ⑥ Data normalization / standardization: Numerical features are scaled to make them the same dimension. Specifically, Min-Max normalization can be used to convert values ​​to the [0,1] interval, or Z-Score standardization can be used to process the data into a distribution with a mean of 0 and a standard deviation of 1.

[0024] A2: Use rule engines and machine learning algorithms to extract features and cross-validate standardized health data to generate labels that identify health risk levels, disease types, or lifestyles.

[0025] Understandably, feature extraction via a rule engine, for example: if systolic blood pressure consistently >140 mmHg and diastolic blood pressure consistently >90 mmHg, a user will be labeled "high blood pressure risk." Feature extraction via machine learning algorithms means that by analyzing data from thousands of patients, the algorithm may discover a feature combination of "age + nocturnal heart rate variability + medication adherence in the last three months," which can predict the risk of heart failure recurrence earlier and more accurately than blood pressure alone. Then, it will automatically label high-risk users as "high-risk for heart failure." Cross-validation is performed to prevent the model from being effective only for a certain part of the data (overfitting), thereby ensuring that the generated labels are accurate and stable enough under all circumstances.

[0026] It should be noted that health risk level labels include, but are not limited to, low risk, medium risk, high cardiovascular risk, and high risk of falls; disease type labels include, but are not limited to, type 2 diabetes, stage II hypertension, and post-coronary artery disease surgery; and lifestyle labels include, but are not limited to, sedentary office workers, those with high-salt diets, and those who exercise regularly.

[0027] For example, basic user information can be obtained, including name, gender, age, and contact information. For instance, the basic information for user Zhang Sanfeng could be obtained as follows: Name: Zhang Sanfeng, Gender: Male, Age: 38, Contact Information: 186xxxxxxxx.

[0028] Extract users' medical history information, including the department visited, diagnosis results, and treatment plans. For example, extract the medical history information of user Zhang Sanfeng: Department visited: Cardiology; Diagnosis result: Coronary heart disease; Treatment plan: Aspirin, statins, beta-blockers.

[0029] Extract user lifestyle information, including diet, exercise, smoking and drinking habits. For example, extract user Zhang Sanfeng's lifestyle information: Diet: low carbohydrates, high protein; Exercise: five aerobic exercise sessions per week; Smoking and drinking: non-smoker, but drinks alcohol.

[0030] The extracted information is standardized and features are extracted to obtain the user's feature vector. For example, the user's basic information, medical history information, and lifestyle information are converted into a standardized text format, and the TF-IDF (Term Frequency-Inverse Document Frequency) algorithm is used to extract key features to obtain the feature vector of user Zhang Sanfeng.

[0031] In one optional implementation, the acquisition of the user's multi-dimensional health data in step S100 can be achieved through the automatic collection of physiological data using IoT devices. Specifically, by authorizing access to the user's daily-used smart health IoT devices, real-time physiological data can be automatically and continuously collected. The system establishes a secure connection with the cloud platform through the device's provided API (Application Programming Interface) or SDK (Software Development Kit), and retrieves data periodically or triggered on a regular basis.

[0032] In another optional implementation, the acquisition of the user's multi-dimensional health data in step S100 can also be based on multi-source data aggregation authorized by the API of a third-party health platform. That is, with the user's authorization and consent, the system can collect the user's health data scattered across different applications and devices in one stop by calling the standardized API of a third-party health platform (such as Apple Health Kit, Google Fit, WeChat Sports) or the open platform of a medical institution.

[0033] In this embodiment of the application, step S200 involves constructing a dynamically updated user health profile based on user feature tags.

[0034] Understandably, before building a health profile, a large number of user characteristic tags have already been generated, such as hypertension risk, type 2 diabetes, exercise 3-5 times per week, high-salt diet, sleep disorders, and 3 months post-surgery. Each tag is a quantitative or qualitative description of a user's health in a specific dimension. Building a dynamically updated user health profile involves organically combining these scattered, multi-dimensional user characteristic tags to form a complete and three-dimensional panoramic view of the user's health. Specifically, the tags are first categorized and organized. For example, hypertension risk and high cholesterol are categorized into disease risk; high-salt diet and sedentary lifestyle are categorized into lifestyle; and high scores on anxiety self-rating scales are categorized into psychological state. The internal relationships between the tags are then analyzed to construct a logical and interconnected whole. For example, the system will identify a high correlation between a user's diabetes tag and their high-sugar diet and obesity tags. The final health profile is a structured data model in the computer (such as a JSON object or feature vector containing various tags and their weights and relationships), which can be directly read and calculated by subsequent algorithms.

[0035] It should be noted that dynamic updates mainly occur in two situations: Firstly, when the system obtains new data through new follow-ups (such as intelligent outbound calls), users uploading the latest physical examination reports, or connecting to smart devices (such as wristbands), it will trigger a profile update; secondly, even without new data, the system will periodically (or event-triggered) rerun the machine learning model with the latest data, which may discover potential changes in user risk. For example, if the system periodically analyzes all user data and finds that the changing trends of multiple indicators of a certain user are consistent with the characteristics of pre-heart failure, even if there is no current diagnosis, it may add a new label of pre-heart failure risk to the user, thereby achieving early warning.

[0036] In this embodiment of the application, in step S300, the user health profile is input into the business classification model to determine the target follow-up business line and follow-up method (method) corresponding to the current user. Furthermore, the business classification model is a model trained based on machine learning algorithms, used to map users to predefined follow-up business lines, which are divided according to the user's health risk level.

[0037] Understandably, machine learning algorithms can be random forests, gradient boosting trees, or neural network algorithms.

[0038] It should be noted that a large amount of historical patient data and manually labeled data indicating which follow-up service line each patient should belong to are used to train the service classification model. This allows the service classification model to learn the relationship between these data and the labels, and ultimately, by inputting new user "health profile" data into the service classification model, it can output a prediction result, namely, which predefined service line the current user should be assigned to.

[0039] For example, the "Cardiovascular Rehabilitation Management" business line is selected based on the label "cardiovascular disease rehabilitation patients".

[0040] It should be further clarified that the follow-up service is divided into three levels: high, medium, and low. Specifically, the high-risk service targets recently discharged patients with myocardial infarction or late-stage cancer; the medium-risk service targets patients with stable chronic diseases such as hypertension and diabetes; and the low-risk service targets healthy individuals or users who only require rehabilitation guidance. Furthermore, follow-up methods include, but are not limited to, intelligent outbound telephone calls (suitable for important follow-ups requiring interaction, such as for high-risk patients), SMS reminders (suitable for simple medication reminders or follow-up appointment notifications), APP push questionnaires (suitable for young users familiar with mobile phone operation to conduct self-assessments), and human customer service intervention (automatically transferring to a human operator when the system detects that the user is emotionally distressed or reports an emergency).

[0041] In this embodiment of the application, step S400 involves constructing prompt words based on the target follow-up business line, follow-up method, and user health profile, and inputting them into a large language model to generate personalized follow-up content, including the following steps B1-B2: B1: Construct a large language model based on user health profiles, and combine it with content templates and prompts that match the target follow-up business lines and follow-up methods; B2: Generate complete follow-up dialogue content including opening remarks, a brief description of the condition, questionnaire assessment, rehabilitation guidance, and concluding remarks.

[0042] Understandably, the disease summary includes the patient's current condition and a brief description of the disease, and the questionnaire assessment includes rehabilitation indicators and symptom assessment.

[0043] For example, the opening line could be: "Hello! I'm XXX, and I'm happy to provide you with rehabilitation guidance." The brief description of your condition is as follows: "Your recovery progress is as follows: You have been diagnosed with coronary heart disease and are currently receiving treatment with aspirin, statins, and beta-blockers." The rehabilitation guidelines are: "Please remember to contact your doctor at any time and pay attention to maintaining healthy lifestyle habits, including a balanced diet, moderate exercise, and quitting smoking and drinking." The closing message reads: "Thank you for your cooperation and patience. Our next follow-up will be on [Date] at [Time]. Please note any changes to the follow-up time. If you have any questions, please feel free to contact us. Best regards!" In this embodiment of the application, step S500 involves performing a follow-up task on the user based on the follow-up content through an intelligent outbound calling system, including the following steps C1-C2: C1: Converting text-based follow-up content into speech information using speech synthesis technology; C2: Establishes a voice communication connection with the user through an automatic dialing system to perform follow-up.

[0044] It should be noted that the intelligent outbound calling system can generate natural and fluent voice dialogues based on the patient's health profile and the content of the follow-up questionnaire to conduct health consultations and information collection.

[0045] Ideally, intelligent outbound calling can improve the coverage and efficiency of the system's follow-up.

[0046] In this embodiment of the application, step S600 involves collecting user feedback data during the follow-up process, parsing the feedback data, and extracting key information, including the following steps D1-D2: D1: During the intelligent outbound calling process, the voice recognition system receives user voice feedback in real time and converts it into text data. D2: Perform natural language processing on the converted text data to extract key information related to health status, symptom changes, or emotional tendencies.

[0047] It should be noted that speech recognition systems can identify keywords, emotions, and intentions in natural language, helping to analyze patient feedback.

[0048] Ideally, efficient speech recognition enables the system to acquire patients' health information and emotional state in real time, providing support for subsequent data analysis.

[0049] In this embodiment of the application, step S700 updates the user's health profile and user feature tags based on key information to optimize subsequent follow-up strategies.

[0050] Understandably, the intelligent data extraction and feedback system updates the user's health profile and user characteristic tags based on key information to optimize subsequent follow-up strategies. Furthermore, the system can identify new needs or potential problems of patients and adjust follow-up strategies based on feedback.

[0051] It should be noted that through continuous data feedback and learning, the system can continuously optimize personalized health management plans to improve patient satisfaction.

[0052] In summary, this invention constructs a precise dynamic health profile by automatically collecting and deeply analyzing users' multidimensional health data, and automatically matches the most suitable follow-up strategy for patients with different risk levels based on an intelligent classification model; it generates highly personalized follow-up dialogue content using a large language model, covering the entire process from disease review to rehabilitation guidance; and it achieves efficient and natural proactive communication through intelligent outbound calling and speech recognition technology, and analyzes patient feedback in real time to form a closed-loop learning mechanism of "execution-feedback-optimization". While significantly improving follow-up efficiency and coverage, it truly realizes refined, dynamic, and humanized health management for each user.

[0053] Example 3 illustrates a method for personalized health follow-up based on health profiling and a large language model. It should be noted that the technical solution of this personalized health follow-up system based on health profiling and a large language model belongs to the same concept as the technical solution of the personalized health follow-up method based on health profiling and a large language model described above. Details not described in detail in the technical solution of the personalized health follow-up system based on health profiling and a large language model in this example can be found in the description of the technical solution of the personalized health follow-up method based on health profiling and a large language model described above.

[0054] This embodiment also provides a personalized health follow-up system based on health profiles and large language models, including: The data collection and extraction module is used to collect multi-dimensional health data from users, and to perform standardized processing and feature extraction on the health data to generate multiple user feature tags. The first building module is used to construct dynamically updated user health profiles based on user feature tags; The determination module is used to input the user's health profile into the business classification model to determine the target follow-up business line and follow-up method for the current user. The second construction module is used to build prompt words based on the target follow-up business line, follow-up method and user health profile and input them into the large language model to generate personalized follow-up content. The execution module is used to perform follow-up tasks on users through an intelligent outbound calling system based on the follow-up content. The data collection and analysis module is used to collect user feedback data during the follow-up process, analyze the feedback data, and extract key information. The update module is used to update the user's health profile and user characteristic tags based on key information, in order to optimize subsequent follow-up strategies.

[0055] This embodiment also provides an electronic device suitable for personalized health follow-up based on health profiles and large language models, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the personalized health follow-up method based on health profiles and large language models proposed in the above embodiment.

[0056] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, it implements the personalized health follow-up method based on health profiles and large language models as proposed in the above embodiments.

[0057] The storage medium proposed in this embodiment belongs to the same inventive concept as the personalized health follow-up method based on health profile and large language model proposed in the above embodiments. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0058] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0059] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A personalized health follow-up method based on health profiling and large language models, characterized in that: include, Collect multi-dimensional health data from users, and perform standardized processing and feature extraction on the health data to generate multiple user feature tags; Based on the user feature tags, a dynamically updated user health profile is constructed; The user health profile is input into the business classification model to determine the target follow-up business line and follow-up method for the current user. Based on the target follow-up business line, the follow-up method, and the user health profile, prompt words are constructed and input into the large language model to generate personalized follow-up content; Based on the follow-up content, follow-up tasks are performed on users through an intelligent outbound calling system; Collect user feedback data during the follow-up process, parse the feedback data and extract key information; Based on the key information, update the user's health profile and user feature tags to optimize subsequent follow-up strategies.

2. The personalized health follow-up method based on health profiling and large language model as described in claim 1, characterized in that: The generation of multiple user feature tags include, The standardized health data is used to extract features and cross-validate using a rule engine and machine learning algorithms to generate labels that identify health risk levels, disease types, or lifestyles.

3. The personalized health follow-up method based on health profiling and large language model as described in claim 2, characterized in that: The generation of personalized follow-up content, include, The context of constructing a large language model based on the user health profile is combined with content templates and prompts that match the target follow-up business line and the follow-up method; Generate complete follow-up dialogue content that includes an opening, a brief description of the patient's condition, a questionnaire assessment, rehabilitation guidance, and a concluding reminder.

4. The personalized health follow-up method based on health profiling and large language model as described in claim 3, characterized in that: The process of performing follow-up tasks on users through an intelligent outbound calling system includes, The text-based follow-up content is converted into speech information using speech synthesis technology; Establish a voice communication connection with the user through an automatic dialing system to perform follow-up.

5. The personalized health follow-up method based on health profiles and large language models as described in claim 4, characterized in that: Collect user feedback data during the follow-up process, parse the feedback data, and extract key information, including: During the intelligent outbound calling process, the voice recognition system receives user voice feedback in real time and converts it into text data. Natural language processing is performed on the converted text data to extract key information related to health status, symptom changes, or emotional tendencies.

6. The personalized health follow-up method based on health profiles and large language models as described in claim 5, characterized in that: The business classification model is a model trained based on machine learning algorithms, used to map users to predefined follow-up business lines, which are divided according to the user's health risk level.

7. A personalized health follow-up method based on health profiling and large language models as described in any one of claims 1-6, characterized in that: The multidimensional health data includes at least one of the following: basic information, medical history information, lifestyle information, and psychological state information.

8. A personalized health follow-up system based on health profiling and large language models, employing the method described in any one of claims 1-7, characterized in that, include: The data collection and extraction module is used to collect multi-dimensional health data from users, and to perform standardized processing and feature extraction on the health data to generate multiple user feature tags. The first construction module is used to construct a dynamically updated user health profile based on the user feature tags; The determination module is used to input the user health profile into the business classification model to determine the target follow-up business line and follow-up method corresponding to the current user. The second construction module is used to construct prompt words based on the target follow-up business line, the follow-up method, and the user health profile, and input them into the large language model to generate personalized follow-up content. The execution module is used to perform follow-up tasks on users through an intelligent outbound calling system based on the follow-up content; The data collection and analysis module is used to collect user feedback data during the follow-up process, analyze the feedback data, and extract key information. The update module is used to update the user's health profile and user feature tags based on the key information, in order to optimize subsequent follow-up strategies.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the 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 a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images