Health management scheme generation method and device based on artificial intelligence

Abstract

The invention relates to the technical field of artificial intelligence, and particularly discloses a health management scheme generation method and device based on artificial intelligence. Performing entity recognition and intention recognition on the user voice to obtain disease features and user intention; obtaining target health data associated with the disease characteristics; analyzing the disease characteristics and the target health data to obtain disease information and predict health risks; and generating a health management scheme based on the user intention, the disease information and the predicted health risk. The health management scheme is generated in combination with the user intention, the individuation degree is improved, the health parameters associated with the disease features are obtained, the data richness is improved, and then the accuracy of the health management scheme is improved. When the method is applied to health management systems such as medical care, family health monitoring and the like, a health management scheme with high individuation degree and high accuracy can be generated according to the intention of the user and the specific health data, so that the user experience is improved.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to a method and apparatus for generating health management solutions based on artificial intelligence. Background Technology

[0002] With the aging population and uneven distribution of medical resources, traditional medical service models face problems such as low efficiency, slow response, and limited coverage. In recent years, artificial intelligence (AI) technology has been increasingly applied in the medical field. For example, intelligent triage using large language models can identify patient symptoms through natural language processing and convert them into standardized medical terms, thereby matching appropriate treatment plans and health management solutions. However, the application of AI technology in the medical field still has shortcomings. For instance, it only supports text input and lacks integrated analysis of multimodal data such as voice, images, and wearable devices, resulting in a single data source. Furthermore, it does not fully consider user intent, analyzing only text-input disease information to generate solutions, leading to low personalization and accuracy. Therefore, improving the personalization and accuracy of health management solutions has become an urgent problem to be solved. Summary of the Invention

[0003] This application provides a method and apparatus for generating health management plans based on artificial intelligence, so as to improve the personalization and accuracy of health management plans.

[0004] Firstly, this application provides a method for generating health management solutions based on artificial intelligence, the method comprising: Upon receiving a user's voice, entity recognition and intent recognition are performed on the user's voice to obtain symptom characteristics and user intent; Retrieve target health data associated with the disease characteristics from a pre-set health database; Based on a preset disease analysis model, the disease characteristics and target health data are analyzed to obtain the user's disease information and predict health risks. The health management plan is generated based on the user intent, the disease information, and the predicted health risks.

[0005] Secondly, this application also provides an artificial intelligence-based health management solution generation device, the device comprising: The voice processing module is used to perform entity recognition and intent recognition on the user's voice when it is received, so as to obtain the symptoms and the user's intent. The data extraction module is used to obtain target health data associated with the disease characteristics from a preset health database; The disease analysis module is used to analyze the symptom characteristics and target health data based on a preset disease analysis model to obtain the user's disease information and predict health risks. The solution generation module is used to generate the health management solution based on the user intent, the disease information, and the predicted health risks.

[0006] Thirdly, this application also provides a computer device, the computer device including a memory and a processor; the memory is used to store a computer program; the processor is used to execute the computer program and, when executing the computer program, implement the above-described method for generating a health management scheme based on artificial intelligence.

[0007] Fourthly, this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to implement the artificial intelligence-based health management scheme generation method described above.

[0008] This application discloses a method and apparatus for generating health management plans based on artificial intelligence. Upon receiving user voice, the method performs entity recognition and intent recognition on the user's voice to obtain symptom characteristics and user intent; retrieves target health data associated with the symptom characteristics from a preset health database; analyzes the symptom characteristics and target health data based on a preset disease analysis model to obtain the user's disease information and predict health risks; and generates the health management plan based on the user intent, the disease information, and the predicted health risks. This application combines user intent to generate health management plans, improving personalization, and extracts associated health parameters from the health database based on symptom characteristics, increasing data richness and thus improving the accuracy of the health management plan. Attached Figure Description

[0009] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a schematic flowchart of a health management plan generation method based on artificial intelligence provided in the first embodiment of this application; Figure 2 This is a schematic flowchart of a health management plan generation method based on artificial intelligence provided in the second embodiment of this application; Figure 3This is a schematic flowchart of a health management plan generation method based on artificial intelligence provided in the third embodiment of this application; Figure 4 A schematic block diagram of an artificial intelligence-based health management solution generation device provided for embodiments of this application; Figure 5 A schematic block diagram of the structure of a computer device provided for an embodiment of this application. Detailed Implementation

[0011] 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, 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.

[0012] The flowchart shown in the attached diagram is for illustrative purposes only and does not necessarily include all content and operations / steps, nor does it necessarily have to be performed in the order described. For example, some operations / steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.

[0013] It should be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of the application. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0014] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0015] This application provides an embodiment of an artificial intelligence-based health management plan generation method and apparatus. The method can be applied to an AI-based health management system. By combining user intent to generate health management plans, it improves personalization. Furthermore, by extracting relevant health parameters from a health database based on symptom characteristics, it enhances data richness and thus improves the accuracy of the health management plan. This AI-based health management system can be deployed on terminal devices (such as smartphones, smartwatches, tablets, etc.), local servers (such as home gateways, community health center servers), and cloud platforms (such as hospital information systems, internet hospital platforms, regional medical data centers, etc.). This AI-based health management system includes: a multimodal perception module for collecting user voice input, image data, wearable device monitoring information, and electronic health records; a knowledge enhancement and understanding module for semantic analysis and reasoning of user symptoms based on an improved large language model and medical knowledge graph; a personalized health management module for generating health management plans based on user data; an intelligent triage and resource scheduling module for recommending hospitals and making appointments based on disease information and hospital resources; a reinforcement learning decision-making module for training models on training datasets to obtain disease analysis models, health suggestion generation models, treatment plan generation models, and medical image recognition models; a doctor collaboration and feedback module for receiving corrections from doctors; and a data security and privacy protection module for encrypting model parameters and data transmission based on federated learning and national cryptographic algorithms.

[0016] In one embodiment, the multimodal perception module includes a speech recognition unit for converting speech into text; an image recognition unit for recognizing medical images based on a medical image recognition model to obtain image description data; a wearable device interface for acquiring physiological indicator data of the user monitored by a smart wearable device; and an electronic health record access interface for connecting with a medical institution system to obtain the user's electronic health record.

[0017] In one embodiment, the knowledge enhancement understanding module includes a medical semantic parser for symptom entity recognition and intent classification based on a medical semantic parsing model (such as Medical-BERT or Med-PaLM model); a knowledge graph inferencer for path reasoning and association mining based on a knowledge graph; and a multimodal fusion network for fusing speech text, medical image data, and physiological indicator data.

[0018] In one embodiment, the intelligent triage and resource scheduling module includes a triage unit for guiding patients to departments based on disease information and predicted health risks; and a resource scheduling engine for connecting to the Internet hospital platform API to realize registration, referral, and examination order flow.

[0019] In one embodiment, the doctor collaboration and feedback module includes a doctor review unit for synchronously pushing health management plans to doctors for confirmation and modification; a modification record archiving and auditing unit for storing doctors' modification records; a doctor rating unit for receiving and storing doctors' accuracy ratings of the health management plans; and a doctor editing unit for receiving doctors' edited content, such as nodes and relationships in a knowledge graph.

[0020] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0021] Please see Figure 1 , Figure 1 This is a schematic flowchart illustrating an AI-based health management plan generation method provided in an embodiment of this application. This AI-based health management plan generation method can be applied to AI-based health management systems to generate health management plans by combining user intent, thus improving personalization. Furthermore, it extracts relevant health parameters from a health database based on symptom characteristics, increasing data richness and thereby improving the accuracy of the health management plan.

[0022] like Figure 1 As shown, the method for generating a health management plan based on artificial intelligence specifically includes steps S101 to S104.

[0023] S101. Upon receiving user voice, perform entity recognition and intent recognition on the user voice to obtain symptom characteristics and user intent. In one embodiment, a user describes their health condition via voice input, such as "I've had a bad cough and a slight fever for the past two days." Upon receiving the user's voice, the system first invokes a speech recognition unit, using speech recognition technology (such as a speech recognition model) to convert the speech signal into text. Then, natural language processing techniques, such as deep learning-based named entity recognition models and medical semantic parsing models, are used to analyze the converted text and identify key entity information, such as symptoms (cough, fever) and duration (two days).

[0024] In one embodiment, structured symptom triples are generated based on the obtained key entity information, serving as symptom features.

[0025] In another embodiment, medical professional dictionaries and knowledge graphs can be combined to further improve the accuracy and professionalism of entity recognition, ensuring that feature information related to the disease can be accurately extracted.

[0026] In one embodiment, an intent classification model or rule engine is invoked to identify the user's text input intent and determine the type of user's needs, such as consulting about symptoms, seeking medical advice, or learning about disease prevention.

[0027] S102. Obtain target health data associated with the disease characteristics from a preset health database; Further, step S102 includes: performing a correlation analysis on the symptom features and health data in the user's health database to obtain the correlation between the symptom features and the health data; and extracting health data from the health database whose correlation is greater than a preset correlation threshold as the target health data.

[0028] In one embodiment, the health database includes health-related data such as users' physiological indicators, image description data, and electronic health record data.

[0029] In one embodiment, statistical correlation analysis methods, such as Pearson correlation coefficient or Spearman rank correlation coefficient, are used to measure the linear or non-linear correlation between disease characteristics and health data. For example, analyzing the correlation between the severity of cough symptoms and the white blood cell count in a complete blood count, if the absolute value of the correlation coefficient is greater than a preset threshold, the two are considered to have a strong correlation.

[0030] In another embodiment, a machine learning model is used to analyze the correlation between disease features and health data. Models such as decision trees, random forests, and support vector machines are employed. Disease features from the training dataset are used as input features, and health data is used as the target variable to train the model, learning the correlation patterns between the two. Feature importance assessment or coefficient analysis is then used to determine which health data points show a strong correlation with the disease features. For example, when predicting the risk of developing a disease, the model might find that certain combinations of disease features are highly correlated with the risk of that disease.

[0031] In one embodiment, the correlation threshold can be set according to specific application scenarios and requirements. Health data with a correlation greater than the preset threshold are extracted and used as target health data.

[0032] Furthermore, prior to step S102, the method further includes: acquiring the user's physiological indicator data, medical images, and electronic health records; recognizing the medical images based on a preset medical image recognition model to obtain image description data; preprocessing the physiological indicator data, the electronic health records, and the image description data to obtain the user's health data; and storing the user's health data in a preset database to obtain the health database.

[0033] In one embodiment, physiological data can be collected by smart wearable devices (such as smart bracelets, smartwatches, and smart blood pressure monitors). These devices monitor the user's physiological indicators in real time, such as heart rate, blood oxygen saturation, blood pressure, blood sugar, body temperature, steps, and sleep quality, through built-in sensors. The user then calls the wearable device interface to retrieve the physiological data collected by the smart wearable device. Alternatively, the user can actively upload the data to an AI-based health management system.

[0034] In one embodiment, medical images are various images captured by the user using medical equipment, such as X-ray films, CT (Computed Tomography) images, and MRI (Magnetic Resonance Imaging) images. These can be obtained directly through an interface connected to the medical institution's system, or uploaded by the user.

[0035] In one embodiment, an electronic health record access interface is invoked to obtain the user's electronic health record through a standard-compliant interface protocol. The electronic health record contains a wealth of medical information, including the user's basic personal information, medical history, surgical records, allergy history, test reports, medication records, and more.

[0036] In one embodiment, the medical image is parsed to extract key information such as patient ID, image type, capture time, and image location. An image recognition unit is then invoked to analyze the medical image using a deep learning-based medical image recognition model. Lesions and abnormal structures in the image are identified and located, generating image description data such as the size, shape, location, and density of lesions, as well as possible disease indications.

[0037] Among them, medical image recognition models are usually based on convolutional neural network architecture and are trained on a large amount of labeled medical image data.

[0038] In one embodiment, the received physiological indicator data may contain noise, outliers, or missing values. The data is filtered to remove noise interference, smooth data curves, and improve data accuracy. A reasonable threshold range is set to identify and remove outliers that exceed the normal range. For missing data points, interpolation algorithms are used to fill in the gaps, ensuring data integrity and continuity.

[0039] Physiological data collected from different devices are converted into a unified unit and format to facilitate subsequent integration and analysis. For example, all blood pressure data are uniformly expressed as millimeters of mercury, and heart rate data are uniformly expressed as beats per minute.

[0040] Clean the data in the electronic health records to remove duplicate, erroneous, or incomplete data.

[0041] In one embodiment, all data is processed in a standardized format and timestamped. The processed physiological indicator data, image description data, and electronic health record data are then integrated, and the integrated user health data is stored in a preset database to construct a health database.

[0042] In another embodiment, when a user generates new health data, such as new physiological indicator monitoring data, medical images, or new medical records, this new data is periodically acquired and integrated into the health database to maintain the timeliness and integrity of the database.

[0043] In the above embodiments, multimodal data such as voice and text data, medical images, and electronic health records are processed to generate a health database, which improves the richness of the data and can provide effective data in subsequent disease analysis and health risk prediction, thereby improving the accuracy of the output results of each model.

[0044] S103. Analyze the symptom characteristics and target health data based on a preset disease analysis model to obtain the user's disease information and predict health risks. In one embodiment, the disease analysis model includes a multimodal fusion network layer, a knowledge graph reasoning layer, and a long short-term memory network layer. The disease analysis model is trained using historical symptom characteristics, health data, disease information, and historical health risks. It is then combined with the physician's professional knowledge and clinical experience to evaluate and interpret the prediction results until the accuracy of disease information and predicted health risks reaches a preset standard, thus obtaining the final disease analysis model.

[0045] In one embodiment, based on a pre-defined disease analysis model, extracted symptom features and acquired target health data are used as input for comprehensive analysis. The disease analysis model can employ machine learning algorithms to assess and analyze the user's symptoms, thereby deriving the user's disease information, including disease type and severity.

[0046] In one embodiment, the disease analysis model also predicts the user's health risks, such as the likelihood of developing a certain disease in the future and the potential risk of disease deterioration, providing an important basis for the formulation of subsequent health management plans.

[0047] Further, step S103 includes: fusing the symptom features and the target health data based on the multimodal fusion network layer of the disease analysis model to obtain fused features; performing disease inference on the fused features based on the knowledge graph reasoning layer of the disease analysis model to obtain the disease information; and performing time series identification and risk prediction on the fused features based on the long short-term memory network layer of the disease analysis model to obtain the predicted health risk.

[0048] In one embodiment, the collected disease characteristics and target health data are preprocessed, including data cleaning, normalization, and feature extraction, to eliminate noise and inconsistencies in the data.

[0049] Different feature extraction methods are employed to process multimodal data such as text, images, and physiological signals. For example, word embedding technology is used to convert text descriptions into vector representations; convolutional neural networks are used to extract key features from medical images; and time-domain and frequency-domain analysis is performed on physiological signals to extract feature parameters.

[0050] In one embodiment, the processed multimodal data is transmitted to the multimodal fusion network layer of the disease analysis model. By mapping data from different modalities such as text, images, and physiological signals to a shared semantic space, and employing a cross-attention mechanism to achieve cross-modal information interaction, fused features are generated.

[0051] In one embodiment, each node in the knowledge graph represents a medical concept, and edges represent relationships between concepts, such as "symptoms manifest as" or "the disease may cause". Specifically, a medical knowledge graph is constructed by integrating authoritative knowledge bases in the medical field, containing concepts such as disease, symptoms, examinations, and treatments, as well as their interrelationships.

[0052] By performing path search and relationship reasoning in the knowledge graph, we can identify the diseases and related information that best match the fusion features.

[0053] In one embodiment, based on the Long Short-Term Memory network layer, the fused features are subjected to time series recognition to capture the changing trend of user health data over time. The time series data is divided into multiple segments using a sliding window method, with each segment representing the health status over a period of time. Based on the changing patterns and rules of health data over time, the health risks in the future are predicted to obtain the predicted health risks.

[0054] In one embodiment, disease reasoning and health risk prediction can be performed in parallel.

[0055] In another embodiment, when making health risk predictions, information such as weather, user lifestyle habits, and behavioral preferences can also be combined to improve the accuracy of the predictions.

[0056] S104. Based on the user intent, the disease information, and the predicted health risks, generate the health management plan.

[0057] In one embodiment, a personalized health management plan is generated by the system's health management module, taking into account user intent, disease information, and predicted health risks. The health management plan may include medical advice, self-care advice, medication advice, rehabilitation guidance, and health risk prevention and control recommendations.

[0058] In specific embodiments, different health management plans are generated depending on the user's intent. For example, when the user's intent is to seek medical treatment offline, the generated health management plan includes medical advice and a treatment plan. When the user's intent is health consultation, the health management plan includes self-care advice, medication advice, rehabilitation guidance, etc.

[0059] In another embodiment, the disease analysis model can also determine the content of the health management plan based on both the severity of the disease and the user's intent. For example, when the severity of the disease exceeds a preset threshold, even if the user's intent is health consultation, medical advice and treatment plans are generated based on disease information and predicted health risks to alert the user that their current condition requires medical attention. When the severity of the disease is below the preset threshold, the user's intent takes precedence, and a plan corresponding to that intent is generated.

[0060] The above embodiments provide a method and apparatus for generating health management solutions based on artificial intelligence. Upon receiving user voice, the method performs entity recognition and intent recognition on the user voice to obtain symptom characteristics and user intent; retrieves target health data associated with the symptom characteristics from a preset health database; analyzes the symptom characteristics and target health data based on a preset disease analysis model to obtain the user's disease information and predict health risks; and generates the health management solution based on the user intent, the disease information, and the predicted health risks. This application combines user intent to generate health management solutions, improving personalization, and extracts associated health parameters from the health database based on symptom characteristics, increasing data richness and thus improving the accuracy of the health management solution.

[0061] Please see Figure 2 , Figure 2 This is a schematic flowchart illustrating an AI-based health management plan generation method provided in an embodiment of this application. This AI-based health management plan generation method can be applied to AI-based health management systems. It combines user disease information, predicted health risks, and medical institution resources to determine recommended hospitals, fully utilize idle medical resources, ensure users can receive timely medical care, and improve efficiency. Furthermore, it rationally plans departure times and travel routes based on information such as appointment time and geographical location, improving the accuracy of the health management plan.

[0062] like Figure 2As shown, the method for generating a health management plan based on artificial intelligence specifically includes steps S201 to S204.

[0063] S201. When the user's intention is to seek medical treatment offline, an initial medical treatment plan is generated based on a preset medical treatment plan generation model, analyzing the disease information and the predicted health risks. In one embodiment, when a user intends to seek medical treatment offline, a machine learning model (such as a decision tree or neural network) is used to analyze the user's disease information (such as symptoms and preliminary diagnosis results) and predict health risks (such as the probability of future illness and the trend of disease progression) to generate an initial treatment plan.

[0064] Specifically, the initial medical plan includes the department to be consulted and the examinations to be performed. The machine learning model matches the department (such as internal medicine, surgery, or specialty) and preliminary examinations (such as blood tests and X-rays) based on the user's disease information and health risks. For example, if the user has persistent cough and fever, the model may recommend a pulmonologist and suggest a blood test and a chest X-ray.

[0065] S202. Match at least one recommended hospital based on the initial medical treatment plan, and determine the target hospital from the at least one recommended hospital based on the user's instructions; In one embodiment, based on the department and examination requirements in the initial treatment plan, hospitals with corresponding diagnostic and treatment capabilities are selected from the medical resource database to generate a recommended hospital list.

[0066] The system displays a list of recommended hospitals to the user, who can then select a target hospital by viewing detailed information such as hospital profile, department descriptions, doctor teams, and patient reviews. The system then confirms the target hospital based on the user's selection.

[0067] In another embodiment, hospitals can be filtered by combining their geographical location, the user's historical medical records and preferences (such as hospital level, medical insurance designation, etc.) to obtain a list of recommended hospitals.

[0068] S203. Obtain the medical resources of the target hospital, and make an appointment for the user based on the medical resources to obtain the user's appointment information; In one embodiment, information on the target hospital's medical resources is obtained, including doctors' schedules, available appointment slots, and the availability of examination equipment.

[0069] Based on the user's medical needs and the availability of medical resources, the system calls the interface connected to the medical platform to make appointments for the user. It automatically selects a suitable doctor and time slot for the appointment and generates appointment information, including appointment number, appointment time, department, and doctor's name.

[0070] Furthermore, after step S203, the method further includes: obtaining the user's current geographical location and the geographical location of the target hospital, and determining the appointment time based on the appointment information; determining the departure time and travel plan based on the current geographical location, the geographical location of the target hospital, and the appointment time; and generating the target medical plan based on the departure time, the travel plan, the appointment information, and the initial medical plan.

[0071] In one embodiment, the location is determined by the positioning module built into the user's mobile device (such as a smartphone, tablet, etc.), and the user's current geographical location information is obtained, including latitude and longitude coordinates, city, district, street and other detailed address information.

[0072] Obtain detailed geographical location information of the target hospital from the hospital information system or medical resource database, including latitude and longitude coordinates, as well as the hospital's address, name, department distribution, and other information.

[0073] In one embodiment, the appointment time is extracted from the user's appointment information. This can be the examination time, surgery time, etc., and includes explicit time fields such as year, month, day, hour, minute, etc.

[0074] In another embodiment, the extracted appointment time is validated for reasonableness to ensure that the time format is correct, conforms to the hospital's opening hours, and has no logical conflict with the current time (e.g., the appointment time cannot be earlier than the current time). If a problem is found with the appointment time, the user is promptly reminded to modify it or reschedule.

[0075] In one embodiment, the actual road distance between the user's current location and the target hospital's location is calculated. Considering the distance and local traffic conditions, suitable modes of transportation are recommended to the user, such as walking, cycling, driving, public transportation (subway, bus, etc.), or taxis. Using map service interfaces or traffic big data platforms, traffic information around the user's current location and the target hospital is queried, including road congestion, traffic accidents, construction, and other factors that may affect travel. Based on the traffic information and the selected mode of transportation, the estimated travel time from the user's current location to the target hospital is calculated, including a buffer period (e.g., 10%-20%) to handle potential emergencies such as increased traffic congestion or temporary detours.

[0076] Based on the appointment time and estimated travel time, the user's required departure time can be calculated. Departure time = Appointment time - Estimated travel time - Buffer time. Additionally, considering that users may need to arrive at the hospital early for check-in, registration, and other preparations, the departure time can be calculated with an additional period of time, such as 15-30 minutes in advance.

[0077] Based on the selected mode of transportation, a specific travel route is planned for the user, including detailed information such as the starting point, roads or bus routes along the way, and transfer stations. Departure time, travel route, and estimated travel time are presented to the user in a clear and intuitive way, such as generating navigation maps and travel tips, to facilitate the user's travel according to the plan.

[0078] In one embodiment, departure time, travel plan, appointment information, and initial medical plan (such as the department to be registered with, order of visits, and examination arrangements) are integrated to form a complete target medical plan. The target medical plan should include the entire process from departure to the end of the medical visit, including details such as when to depart, how to get to the hospital, and the medical process after arriving at the hospital.

[0079] In another embodiment, the target medical treatment plan can be optimized and personalized based on the user's personal preferences, lifestyle habits, or special circumstances. For example, if the user has mobility difficulties, accessible transportation options and information on accessible facilities within the hospital can be prioritized; if the user is unfamiliar with the hospital environment, hospital navigation information and department location guidance can be added to the plan.

[0080] In one embodiment, the final generated medical treatment plan can be displayed to the user via push notifications from a mobile application, SMS reminders, or other means. The plan should include key time points (such as departure time and appointment time), travel routes, hospital addresses, departments to be treated, and examination schedules, and should include a reminder function to alert the user as the departure or appointment time approaches, ensuring that the user can arrive at the hospital on time.

[0081] S204. Based on the appointment information and the initial medical plan, generate the user's target medical plan as the health management plan.

[0082] In one embodiment, the user's appointment information is integrated with the initial treatment plan to generate a complete treatment plan, including detailed information such as the user's appointment time, location, department, doctor, and examination schedule, as a health management plan.

[0083] In another embodiment, the health management plan may also include pre-visit preparations (such as fasting requirements), precautions on the day of the visit (such as bringing an ID card, medical insurance card, etc.), and follow-up recommendations after the visit.

[0084] In one embodiment, a target medical treatment plan is pushed to the user, and a reminder function is set to ensure that the user receives a timely reminder before the medical treatment.

[0085] In the above embodiments, by combining the user's disease information, predicted health risks, and the medical resources of medical institutions, recommended hospitals are determined, making full use of idle medical resources to ensure that users can receive timely medical treatment and improve the efficiency of medical treatment. Secondly, by rationally planning departure time and travel routes based on information such as appointment time and geographical location, the accuracy of the health management plan is improved.

[0086] Please see Figure 3 , Figure 3 This is a schematic flowchart illustrating an AI-based health management plan generation method provided in an embodiment of this application. This AI-based health management plan generation method can be applied to AI-based health management systems. It combines user basic data and user preference data to generate health recommendations for users, improving the personalization of health recommendations. Furthermore, the health recommendations are transmitted to doctors for modification, and the modified recommendations are used to obtain a health management plan, thus improving the accuracy of the health management plan.

[0087] like Figure 3 As shown, the method for generating a health management solution based on artificial intelligence specifically includes steps S301 to S304.

[0088] S301. When the user's intention is to seek opinions, retrieve basic user data and user preference data from the user information database; S302. Based on a preset health advice generation model, analyze the disease information, the predicted health risks, the user's basic data, and the user's preference data to generate health advice for the user. S303. Transmit the health advice to the doctor's terminal, and receive the doctor's corrections to the health advice based on the doctor's terminal; S304. Based on the revised content, revise the health recommendations to obtain the health management plan.

[0089] In one embodiment, when a user intends to seek health advice, a user information database is accessed. This database contains the user's basic health information (such as age, gender, weight, height, underlying medical conditions, etc.) and user preferences (such as dietary preferences, exercise habits, drug allergy history, preferred forms of health education, etc.). The user's basic data and preference data are then extracted from the database.

[0090] In one embodiment, the health advice generation model can be based on machine learning algorithms, such as decision trees and neural networks, and combine users' health information, disease prediction data, and preference data to generate personalized health advice.

[0091] In another embodiment, the training process of the health advice generation model includes: First, each medical institution trains its model locally to obtain a local health advice generation model. Then, the health management system obtains the parameters of each local health advice generation model, performs weighted fusion analysis on the model parameters corresponding to each medical institution to obtain the latest model parameters, and obtains the final health advice generation model based on the latest model parameters. The model is trained using federated learning to ensure the data security of each medical institution.

[0092] In one embodiment, the health advice generation model comprehensively analyzes a user's disease information (such as current symptoms and diagnosis), health risk predictions (such as future disease risk), basic data (such as age and gender), and preference data (such as dietary habits) to generate specific and feasible health advice, including diet, exercise, lifestyle adjustments, and necessary medical examinations. For example, if a user has a history of hypertension and prefers salty food, the model might suggest reducing salt intake and provide some low-sodium dietary recommendations.

[0093] In one embodiment, the generated health recommendations can be sent to the doctor's end through a communication platform or dedicated interface within the medical system, ensuring that the doctor can view and evaluate them in a timely manner.

[0094] In one embodiment, after reviewing health advice, the doctor may revise or supplement it to ensure its scientific validity and applicability. The system receives and records the doctor's revisions. For example, the doctor may adjust the advice based on the user's actual situation (such as recent test results or changes in their condition), such as changing the recommended exercise intensity or adding specific medical examinations.

[0095] In one embodiment, the original health advice is updated based on feedback from doctors, integrating their professional opinions and scientific evidence. The system then synthesizes the revised health advice into a final health management plan, including specific implementation steps and a timeline, and sends the final plan to the user.

[0096] In the above embodiments, health recommendations for users are generated by combining user basic data and user preference data, which improves the personalization of health recommendations. Furthermore, the health recommendations are transmitted to doctors for modification, and the modifications are made based on the modifications to obtain a health management plan, thereby improving the accuracy of the health management plan.

[0097] Please see Figure 4 , Figure 4 This application provides a schematic block diagram of an AI-based health management plan generation device, which is used to execute the aforementioned AI-based health management plan generation method. The AI-based health management plan generation device can be configured on a server.

[0098] like Figure 4 As shown, the AI-based health management solution generation device 400 includes: The voice processing module is used to perform entity recognition and intent recognition on the user's voice when it is received, so as to obtain the symptoms and the user's intent. The data extraction module is used to obtain target health data associated with the disease characteristics from a preset health database; The disease analysis module is used to analyze the symptom characteristics and target health data based on a preset disease analysis model to obtain the user's disease information and predict health risks. The solution generation module is used to generate the health management solution based on the user intent, the disease information, and the predicted health risks.

[0099] Furthermore, the data extraction module includes: The correlation analysis unit is used to perform correlation analysis between the disease characteristics and health data in the user's health database to obtain the correlation between the disease characteristics and the health data. A health data acquisition unit is used to extract health data from the health database whose correlation is greater than a preset correlation threshold as the target health data.

[0100] Furthermore, the scheme generation module includes: The user data acquisition unit is used to acquire basic user data and user preference data from the user information database when the user's intention is to seek opinions. The health advice generation unit is used to analyze the disease information, the predicted health risks, the user's basic data, and the user's preference data based on a preset health advice generation model, and generate health advice for the user. The correction content acquisition unit is used to transmit the health advice to the doctor's terminal and receive the doctor's correction content for the health advice based on the doctor's terminal; The health advice correction unit is used to revise the health advice based on the correction content to obtain the health management plan.

[0101] Furthermore, the scheme generation module also includes: The initial medical treatment plan generation unit is used to analyze the disease information and the predicted health risks based on a preset medical treatment plan generation model when the user's intention is to seek medical treatment offline. The target hospital determination unit is used to match at least one recommended hospital based on the initial medical treatment plan, and to determine the target hospital from the at least one recommended hospital based on user instructions; The appointment information acquisition unit is used to acquire the medical resources of the target hospital, and make an appointment for the user based on the medical resources, thereby obtaining the user's appointment information. The health management plan acquisition unit is used to generate the user's target medical treatment plan based on the medical appointment information and the initial medical treatment plan, as the health management plan.

[0102] Furthermore, the scheme generation module also includes: The appointment time determination unit is used to obtain the user's current geographical location and the geographical location of the target hospital, and determine the appointment time based on the appointment information; The travel plan determination unit is used to determine the departure time and travel plan based on the current geographical location, the geographical location of the target hospital, and the consultation time; The medical treatment plan generation unit is used to generate the target medical treatment plan based on the departure time, the travel plan, the medical appointment information, and the initial medical treatment plan.

[0103] Furthermore, the disease analysis module includes: The fusion feature acquisition unit is used to perform data fusion on the symptom features and the target health data based on the multimodal fusion network layer of the disease analysis model to obtain fusion features; The disease information acquisition unit is used to perform disease inference on the fused features based on the knowledge graph inference layer of the disease analysis model to obtain the disease information. The health risk prediction unit is used to perform time-series identification and risk prediction on the fused features based on the long short-term memory network layer of the disease analysis model, and obtain the predicted health risk.

[0104] Furthermore, the AI-based health management solution generation device 400 also includes a health database acquisition module, which includes: The data acquisition unit is used to acquire users' physiological indicator data, medical images, and electronic health records. An image recognition unit is used to recognize the medical image based on a preset medical image recognition model to obtain image description data. A data preprocessing unit is used to preprocess the physiological indicator data, the electronic health record, and the image description data to obtain the user's health data; A health database acquisition unit is used to store the user's health data in a preset database to obtain the health database.

[0105] It should be noted that those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the above-described apparatus and modules can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0106] The aforementioned device can be implemented as a computer program, which can be used in, for example... Figure 5 It runs on the computer device shown.

[0107] Please see Figure 5 , Figure 5 This is a schematic block diagram illustrating the structure of a computer device according to an embodiment of this application. The computer device may be a server.

[0108] See Figure 5 The computer device includes a processor, memory, and network interface connected via a system bus, wherein the memory may include non-volatile storage media and internal memory.

[0109] Non-volatile storage media can store operating systems and computer programs. These computer programs include program instructions that, when executed, cause the processor to perform any AI-based health management scheme generation method.

[0110] The processor provides computing and control capabilities, supporting the operation of the entire computer device.

[0111] Internal memory provides an environment for the execution of computer programs in non-volatile storage media. When the computer program is executed by the processor, it enables the processor to execute any method for generating health management solutions based on artificial intelligence.

[0112] This network interface is used for network communication, such as sending assigned tasks. Those skilled in the art will understand that... Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0113] It should be understood that the processor can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among these, a general-purpose processor can be a microprocessor or any conventional processor.

[0114] In one embodiment, the processor is configured to run a computer program stored in memory to perform the following steps: Upon receiving a user's voice, entity recognition and intent recognition are performed on the user's voice to obtain symptom characteristics and user intent; Retrieve target health data associated with the disease characteristics from a pre-set health database; Based on a preset disease analysis model, the disease characteristics and target health data are analyzed to obtain the user's disease information and predict health risks. The health management plan is generated based on the user intent, the disease information, and the predicted health risks.

[0115] In one embodiment, when the processor retrieves target health data associated with the symptom characteristics from a preset health database, it is configured to: A correlation analysis is performed between the disease characteristics and the health data in the user's health database to obtain the correlation between the disease characteristics and the health data; The target health data is the health data extracted from the health database whose correlation is greater than a preset correlation threshold.

[0116] In one embodiment, when the processor generates the health management plan based on the user intent, the disease information, and the predicted health risk, it is configured to: When the user's intent is to seek advice, basic user data and user preference data are retrieved from the user information database. Based on a preset health advice generation model, the disease information, the predicted health risks, the user's basic data, and the user's preference data are analyzed to generate health advice for the user. The health advice is transmitted to the doctor's end, and the doctor's corrections to the health advice are received based on the doctor's end. The health recommendations are revised based on the revised content to obtain the health management plan.

[0117] In one embodiment, when the processor generates the health management plan based on the user intent, the disease information, and the predicted health risk, it is further configured to: When the user's intention is to seek medical treatment offline, an initial medical treatment plan is generated based on a preset medical treatment plan generation model, analyzing the disease information and the predicted health risks. Based on the initial medical treatment plan, at least one recommended hospital is matched, and based on the user's instructions, the target hospital is determined from at least one recommended hospital; Obtain the medical resources of the target hospital, and make an appointment for the user based on the medical resources to obtain the user's appointment information; Based on the appointment information and the initial treatment plan, a target treatment plan for the user is generated, which serves as the health management plan.

[0118] In one embodiment, after acquiring the medical resources of the target hospital, making an appointment for the user based on the medical resources, and obtaining the user's appointment information, the processor is further configured to: Obtain the user's current geographical location and the geographical location of the target hospital, and determine the appointment time based on the appointment information; Based on the current geographical location, the geographical location of the target hospital, and the consultation time, determine the departure time and travel plan; Based on the departure time, the travel plan, the appointment information, and the initial medical plan, the target medical plan is generated.

[0119] In one embodiment, when the processor analyzes the symptom characteristics and the target health data based on a preset disease analysis model to obtain the user's disease information and predict health risks, it is configured to: The multimodal fusion network layer of the disease analysis model performs data fusion on the disease features and the target health data to obtain fused features; Based on the knowledge graph reasoning layer of the disease analysis model, disease inference is performed on the fused features to obtain the disease information; The long short-term memory network layer of the disease analysis model is used to perform time series identification and risk prediction on the fused features to obtain the predicted health risk.

[0120] In one embodiment, before retrieving target health data associated with the symptom characteristics from a preset health database, the processor is further configured to: Acquire users' physiological data, medical images, and electronic health records; Based on a preset medical image recognition model, the medical images are recognized to obtain image description data; The physiological indicator data, the electronic health record, and the image description data are preprocessed to obtain the user's health data. The user's health data is stored in a preset database to obtain the health database.

[0121] The embodiments of this application also provide a computer-readable storage medium storing a computer program, the computer program including program instructions, and the processor executing the program instructions to implement any of the artificial intelligence-based health management scheme generation methods provided in the embodiments of this application.

[0122] The computer-readable storage medium may be an internal storage unit of the computer device described in the foregoing embodiments, such as the hard disk or memory of the computer device. The computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, SmartMedia Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the computer device.

[0123] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for generating health management plans based on artificial intelligence, characterized in that, include: Upon receiving a user's voice, entity recognition and intent recognition are performed on the user's voice to obtain symptom characteristics and user intent; Retrieve target health data associated with the disease characteristics from a pre-set health database; Based on a preset disease analysis model, the disease characteristics and target health data are analyzed to obtain the user's disease information and predict health risks. The health management plan is generated based on the user intent, the disease information, and the predicted health risks.

2. The method for generating a health management plan based on artificial intelligence according to claim 1, characterized in that, The step of obtaining target health data associated with the disease characteristics from a preset health database includes: A correlation analysis is performed between the disease characteristics and the health data in the user's health database to obtain the correlation between the disease characteristics and the health data; The target health data is the health data extracted from the health database whose correlation is greater than a preset correlation threshold.

3. The method for generating a health management plan based on artificial intelligence according to claim 1, characterized in that, The process of generating the health management plan based on the user intent, the disease information, and the predicted health risks includes: When the user's intent is to seek advice, basic user data and user preference data are retrieved from the user information database. Based on a preset health advice generation model, the disease information, the predicted health risks, the user's basic data, and the user's preference data are analyzed to generate health advice for the user. The health advice is transmitted to the doctor's end, and the doctor's corrections to the health advice are received based on the doctor's end. The health recommendations are revised based on the revised content to obtain the health management plan.

4. The method for generating a health management plan based on artificial intelligence according to claim 1, characterized in that, The process of generating the health management plan based on the user intent, the disease information, and the predicted health risks further includes: When the user's intention is to seek medical treatment offline, an initial medical treatment plan is generated based on a preset medical treatment plan generation model, analyzing the disease information and the predicted health risks. Based on the initial medical treatment plan, at least one recommended hospital is matched, and based on the user's instructions, the target hospital is determined from at least one recommended hospital; Obtain the medical resources of the target hospital, and make an appointment for the user based on the medical resources to obtain the user's appointment information; Based on the appointment information and the initial treatment plan, a target treatment plan for the user is generated, which serves as the health management plan.

5. The method for generating a health management plan based on artificial intelligence according to claim 4, characterized in that, After obtaining the medical resources of the target hospital and making an appointment for the user based on the medical resources, and obtaining the user's appointment information, the method further includes: Obtain the user's current geographical location and the geographical location of the target hospital, and determine the appointment time based on the appointment information; Based on the current geographical location, the geographical location of the target hospital, and the consultation time, determine the departure time and travel plan; Based on the departure time, the travel plan, the appointment information, and the initial medical plan, the target medical plan is generated.

6. The method for generating a health management plan based on artificial intelligence according to claim 1, characterized in that, The analysis of the symptom characteristics and target health data based on a preset disease analysis model to obtain the user's disease information and predict health risks includes: The multimodal fusion network layer of the disease analysis model performs data fusion on the disease features and the target health data to obtain fused features; Based on the knowledge graph reasoning layer of the disease analysis model, disease inference is performed on the fused features to obtain the disease information; The long short-term memory network layer of the disease analysis model is used to perform time series identification and risk prediction on the fused features to obtain the predicted health risk.

7. The method for generating a health management plan based on artificial intelligence according to any one of claims 1 to 6, characterized in that, Before retrieving the target health data associated with the disease characteristics from the preset health database, the method further includes: Acquire users' physiological data, medical images, and electronic health records; Based on a preset medical image recognition model, the medical images are recognized to obtain image description data; The physiological indicator data, the electronic health record, and the image description data are preprocessed to obtain the user's health data. The user's health data is stored in a preset database to obtain the health database.

8. A health management solution generation device based on artificial intelligence, characterized in that, include: The voice processing module is used to perform entity recognition and intent recognition on the user's voice when it is received, so as to obtain the symptoms and the user's intent. The data extraction module is used to obtain target health data associated with the disease characteristics from a preset health database; The disease analysis module is used to analyze the symptom characteristics and target health data based on a preset disease analysis model to obtain the user's disease information and predict health risks. The solution generation module is used to generate the health management solution based on the user intent, the disease information, and the predicted health risks.

9. A computer device, characterized in that, The computer device includes a memory and a processor; The memory is used to store computer programs; The processor is configured to execute the computer program and, in executing the computer program, implement the AI-based health management scheme generation method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, causes the processor to implement the artificial intelligence-based health management scheme generation method as described in any one of claims 1 to 7.

Images