Health service recommendation method and device

Abstract

The invention discloses a health service recommendation method and device, and relates to the technical field of internet medical treatment. A specific embodiment of the method comprises the steps of obtaining user data, wherein the user data comprises health question and answer data, health state data and behavior data of a user; inputting the health state data and the behavior data into a pre-constructed first recommendation model, and obtaining a service recommendation first weight output by the first recommendation model; inputting the health question and answer data into a pre-constructed second recommendation model, and obtaining a service recommendation second weight output by the second recommendation model; and according to the service recommendation first weight and the service recommendation second weight, determining a recommended health service corresponding to the user. According to the embodiment, personalized and accurate health service recommendation is realized, the accuracy of the recommendation service is ensured while the personalization of the recommendation service is improved, and then the adoption rate of the recommended health service by the user is improved, so that specialized and personalized recommendation service is provided for the user, and the user is better served.

Description

Technical Field

[0001] The present invention relates to the field of Internet medical technology, and in particular to a method and device for recommending health services. Background Art

[0002] With the rapid development of Internet technology, Internet medical scenarios have gradually matured and currently provide users with convenient online medical consultation and treatment services. Taking into account the professionalism of the medical field, in order to further provide users with professional and personalized health services, existing service recommendation methods usually configure corresponding recommendation services based on business rules, or based on a single-point recommendation approach, determine the user's recommended service under a specified recommendation dimension or strategy, such as recommending health advice, preventive measures, rehabilitation plans, diet plans, etc. to users.

[0003] In the process of implementing the present invention, the inventors found that the prior art has the following problems:

[0004] The existing service recommendation methods are not very accurate, and the user adoption rate of recommended services is not high, which leads to poor recommendation effects and easily affects user satisfaction. Summary of the invention

[0005] In view of this, an embodiment of the present invention provides a method and device for service recommendation, which realizes personalized and accurate service recommendation, fully mines user data that is valuable for service recommendation, and collaboratively determines the final recommended service through the weights output by two recommendation models, thereby improving the personalization of the recommended service while also ensuring the accuracy of the recommended service, thereby improving the user's adoption rate of the recommended service, so as to provide users with professional and personalized recommendation services and better serve users.

[0006] To achieve the above object, according to one aspect of an embodiment of the present invention, a method for recommending health services is provided, comprising:

[0007] Acquire user data, wherein the user data includes the user's health question and answer data, health status data, and behavior data;

[0008] Inputting the health status data and the behavior data into a pre-built first recommendation model, and obtaining a first weight of a service recommendation corresponding to the user output by the first recommendation model;

[0009] Inputting the health question and answer data into a pre-built second recommendation model, and obtaining a second weight of service recommendation corresponding to the user output by the second recommendation model;

[0010] The recommended health service corresponding to the user is determined according to the service recommendation first weight and the service recommendation second weight.

[0011] Optionally, the behavior data includes the user's search behavior on the website, and the search keywords corresponding to the search behavior; the health status data and the behavior data are input into a pre-constructed first recommendation model, and the first weight of the service recommendation corresponding to the user output by the first recommendation model is obtained, including: inputting the health status data and the user's search behavior, and the search keywords corresponding to the search behavior into a pre-constructed first recommendation model, and determining the user category to which the user belongs based on the user health behavior characteristics extracted from the health status data, the search behavior and the corresponding search keywords by the first recommendation model, so that the first recommendation model matches the corresponding service according to the user category and outputs the matching weight as the first weight of the service recommendation.

[0012] Optionally, the health question and answer data is input into a pre-constructed second recommendation model, and a second weight of service recommendation corresponding to the user output by the second recommendation model is obtained, including: inputting the health question and answer data into a pre-constructed second recommendation model, extracting user intention features of the health question and answer data according to the second recommendation model, and determining the user intent of the user, so that the second recommendation model matches the corresponding service according to the user intent and outputs the matching weight as the second weight of service recommendation.

[0013] Optionally, the health status data includes user personal information, user medical history and user current symptoms; the behavior data is a search behavior sequence consisting of user search behavior and corresponding search keywords within a specified time period before obtaining the user data, and each behavior element in the search behavior sequence has a corresponding influence coefficient.

[0014] Optionally, the second recommendation model is constructed based on a language model application construction framework, and the prompt words of the second recommendation model come from a prompt word template corresponding to personal characteristics and symptom characteristics contained in the health status data, and the prompt word template is obtained by optimizing the initial prompt words.

[0015] Optionally, determining the recommended health service corresponding to the user according to the first weight of the service recommendation and the second weight of the service recommendation includes: determining the corresponding first service to be recommended and the weight corresponding to the first service to be recommended according to the first weight of the service recommendation and a preset first weight coefficient; determining the corresponding second service to be recommended and the weight corresponding to the second service to be recommended according to the second weight of the service recommendation and a preset second weight coefficient; merging the first service to be recommended and the second service to be recommended to obtain the health service to be recommended, and calculating the weight of the health service to be recommended according to the weight corresponding to the first service to be recommended and the weight corresponding to the second service to be recommended; and screening out recommended health services from the health services to be recommended according to the weights of the health services to be recommended.

[0016] Optionally, after determining the recommended health service corresponding to the user, the method further includes: monitoring feedback information from the user on the recommended health service, and iteratively training the first recommendation model and the second recommendation model based on the feedback, the first weight of the service recommendation, and the second weight of the service recommendation.

[0017] Optionally, after obtaining the user data, the method further includes: performing data cleaning, text vectorization processing and normalization processing on the user data to obtain processed user data, so as to input the health question and answer data, health status data and behavior data in the processed user data into the first recommendation model and the second recommendation model.

[0018] Optionally, the user data comes from at least one business scenario among a user consultation scenario, a user health management scenario, and a user medical knowledge question and answer scenario.

[0019] Optionally, the first recommendation model is trained based on a pre-constructed training set of health status data and behavior data of sample users, and the health status data and behavior data training set are labeled with corresponding recommended services; the second recommendation model is trained based on a pre-constructed training set of health question and answer data of sample users, and the health question and answer data training set are labeled with corresponding recommended services.

[0020] According to a second aspect of an embodiment of the present invention, there is provided a device for recommending health services, including:

[0021] A user data acquisition module, used to acquire user data, wherein the user data includes the user's health question and answer data, health status data and behavior data;

[0022] A first weight determination module, configured to input the health status data and the behavior data into a pre-built first recommendation model, and obtain a first weight of a service recommendation corresponding to the user output by the first recommendation model;

[0023] A second weight determination module is used to input the health question and answer data into a pre-built second recommendation model to obtain a second weight of the service recommendation corresponding to the user output by the second recommendation model;

[0024] The recommended service determination module is used to determine the recommended health service corresponding to the user according to the service recommendation first weight and the service recommendation second weight.

[0025] According to a third aspect of an embodiment of the present invention, there is provided an electronic device for service recommendation, including:

[0026] one or more processors;

[0027] a storage device for storing one or more programs,

[0028] When the one or more programs are executed by the one or more processors, the one or more processors implement the method provided in the first aspect of the embodiment of the present invention.

[0029] According to a fourth aspect of an embodiment of the present invention, a computer-readable medium is provided, on which a computer program is stored, and when the program is executed by a processor, the method provided by the first aspect of the embodiment of the present invention is implemented.

[0030] According to a fifth aspect of an embodiment of the present invention, there is provided a computer program product, including a computer program, wherein when the computer program is executed by a processor, the method provided in the first aspect of the embodiment of the present invention is implemented.

[0031] An embodiment of the invention has the following advantages or beneficial effects: by obtaining user data, the user data includes the user's health question and answer data, health status data and behavior data; inputting the health status data and behavior data into a pre-constructed first recommendation model, and obtaining the first weight of the service recommendation corresponding to the user output by the first recommendation model; inputting the health question and answer data into a pre-constructed second recommendation model, and obtaining the second weight of the service recommendation corresponding to the user output by the second recommendation model; according to the first weight of the service recommendation and the second weight of the service recommendation, a technical solution for determining the recommended health service corresponding to the user is realized, a personalized and accurate health service recommendation is realized, user data valuable for health service recommendation is fully mined, and the final recommended health service is collaboratively determined through the weights output by the two recommendation models, while improving the personalization of the recommended health service, it also ensures the accuracy of the recommended health service, thereby improving the user's adoption rate of the recommended health service, so as to provide users with professional and personalized health services and better serve users. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The accompanying drawings are used to better understand the present invention and do not constitute an improper limitation of the present invention.

[0033] Figure 1 is a schematic diagram of the main process of the method for health service recommendation according to an embodiment of the present invention;

[0034] Figure 2 is a flow chart of a method for health service recommendation according to a reference embodiment of the present invention;

[0035] Figure 3 is a schematic diagram of the data structure composition of health status data and behavior data in an embodiment of the present invention;

[0036] Figure 4 is an overall schematic diagram of a method for determining recommended health services including a user data processing flow according to an embodiment of the present invention;

[0037] Figure 5 It is a flow chart of generating a second weight for service recommendation by a second recommendation model constructed based on the LLM programming framework model according to an embodiment of the present invention;

[0038] Figure 6 is a schematic diagram of a determination process of a recommended health service according to an embodiment of the present invention;

[0039] Figure 7 This is a schematic diagram of the main principles of the method for health service recommendation according to an embodiment of the present invention;

[0040] Figure 8 is a schematic diagram of main modules of a device for health service recommendation according to an embodiment of the present invention;

[0041] Fig. 9 is an exemplary system architecture diagram to which embodiments of the present invention may be applied;

[0042] Fig.10 It is a schematic diagram of the structure of a computer system of a terminal device or a server suitable for implementing an embodiment of the present invention. DETAILED DESCRIPTION

[0043] It should be noted that in the technical solution of the present invention, the collection / collection, update, analysis, use, transmission, storage and other aspects of user personal information involved are in compliance with the provisions of relevant laws and regulations, are used for legal and reasonable purposes, are not shared, leaked or sold outside of these legal uses, and are subject to supervision and management by national regulatory authorities. Necessary measures should be taken for user personal information to selectively block the use or access to personal information data to prevent illegal access to such personal information data, ensure that persons who have access to personal information data comply with the provisions of relevant laws and regulations, and ensure the security of user personal information. In addition, once these user personal information data are no longer needed, the risks should be minimized by limiting or even prohibiting data collection and / or deleting data.

[0044] The following is a description of exemplary embodiments of the present invention in conjunction with the accompanying drawings, including various details of the embodiments of the present invention to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0045] The existing health service recommendation methods are not very accurate, and the user adoption rate of the recommended health services is not high, which leads to poor recommendation effects. It is also easy to affect user satisfaction and cannot meet practical applications well.

[0046] In order to solve the above-mentioned problems existing in the prior art, the present invention proposes a method for health service recommendation. Based on the recommendation dimensions corresponding to the first recommendation model and the second recommendation model, the acquired user data including health question and answer data, health status data and behavior data are divided into two categories, and are sent to the first recommendation model and the second recommendation model respectively to obtain the first service recommendation weight and the second service recommendation weight output by the two models, and then determine the recommended health service, thereby realizing a personalized and accurate health service recommendation, fully mining the user data that is valuable for health service recommendation, and collaboratively determining the final recommended health service through the weights output by the two recommendation models, which improves the personalization of the recommended health service while also ensuring the accuracy of the recommended health service, thereby improving the user's adoption rate of the recommendation results, so as to provide users with professional and personalized health services and better serve users.

[0047] Figure 1 is a schematic diagram of the main process of the method for health service recommendation according to an embodiment of the present invention, such as Figure 1 As shown, the method for health service recommendation in the embodiment of the present invention includes the following steps S101 to S103.

[0048] Step S101: Acquire user data, where the user data includes the user's health question and answer data, health status data, and behavior data.

[0049] Specifically, when users conduct online medical consultations, health consultations, or purchase health service products through the Internet medical platform, in order to protect the rights and interests of users and merchants, as well as the security of the platform, the platform will record the user's access records on the platform, the interaction information with merchants or language models, and related order information and other data information, and by default, store these data information in the database with the user as the dimension. It can be understood that when each user registers, the platform will collect the user's basic information, such as gender, age, email address, personal health status, etc., with the user's consent. When the platform receives a new registered user, it will establish a user ID based on the collected user's basic information, so as to store the user's data information on the platform in the storage unit corresponding to the user ID.

[0050] Furthermore, in view of the fact that the embodiments of the present invention mainly solve the problem of personalized service recommendation on the Internet medical platform, in order to ensure the accuracy of health service recommendations, when the recommendation system receives a health service recommendation task for a user, it obtains the interaction information between the user and the merchant or language model, the user's access records and health order information within a specified time period from the background database according to the user ID, and organizes the interaction information into health question and answer data, and the access records into user behavior data, and summarizes the user's health status data based on the health question and answer data and health order information, and finally organizes and summarizes the health question and answer data, health status data and behavior data to form the user's user data.

[0051] According to an embodiment of the present invention, the user data comes from at least one business scenario among a user consultation scenario, a user health management scenario, and a user medical knowledge question and answer scenario.

[0052] Specifically, the embodiments of the present invention support the collection of user data for any business scenario provided by the Internet medical platform, which can be health question and answer data, health status data and behavior data in the user consultation scenario; health question and answer data, health status data and behavior data in the user health management scenario; and health question and answer data, health status data and behavior data in the user medical knowledge question and answer scenario of the platform. Of course, when a user is involved in multiple business scenarios, user data for the corresponding scenarios can be obtained at the same time, and the intercommunication of data resources can be realized on the Internet medical platform, and the service recommendation of the recommendation system can be collaboratively supported.

[0053] When user data comes from multiple business scenarios, you can directly summarize and integrate the user data of each business scenario, or you can set the proportion parameters corresponding to the user data of various business scenarios, and integrate and summarize them according to the set proportion parameters to obtain the final user data, so as to achieve the purpose of flexibly adjusting the proportion of user data in key business scenarios.

[0054] Step S102: input the health status data and the behavior data into a pre-built first recommendation model, and obtain a first weight of service recommendation corresponding to the user output by the first recommendation model.

[0055] Specifically, in order to ensure the accuracy of the recommended service, the embodiment of the present invention comprehensively determines the corresponding recommended service from the user's multiple dimensional features. To this end, the embodiment of the present invention uses two recommendation models with different focuses (a first recommendation model and a second recommendation model) to respectively determine the corresponding service recommendation weights. Based on the fact that the first recommendation model mainly recommends services through the user's dynamic data, the embodiment of the present invention uses the behavior data in the user data as the input of the first recommendation model. In addition, considering that the user's access record reflects the user's needs to a certain extent, such as accessing cardiovascular and cerebrovascular physical examination service products, accessing heart-nourishing and heart-protecting health products, etc., but it may also be a situation of helping relatives and friends to place orders or consult, so in order to further improve the accuracy of the recommendation, the health status data in the user data is also added to the first recommendation model, so that the first recommendation model uses the health status data to verify or filter out the target behavior data that is closely related to the user in the behavior data, and then the first recommendation model uses the filtered target behavior data to determine the first weight of the service recommendation corresponding to the health service suitable for recommendation to the user among all the health services of the current medical platform.

[0056] According to an embodiment of the present invention, the first recommendation model is trained based on a pre-constructed training set of health status data and behavior data of sample users, and the health status data and behavior data training set are annotated with corresponding recommended services.

[0057] Specifically, the first recommendation model of the embodiment of the present invention selects the KNN (K-NearestNeighbors Model) model based on classification regression to construct the initial first recommendation model, and then selects sample users from the background database, and obtains the health status data and behavior data of the sample users accordingly. According to the health service products ordered by the sample users after these behavior data occur, the corresponding recommended services are marked for the sample users, and a training set of health status data and behavior data is constructed. The training set is used to train the initial first recommendation model, and the parameters of the initial first recommendation model are updated. When the training converges, a first recommendation model that can be used for online health service recommendations to users is obtained.

[0058] Step S103: input the health question and answer data into a pre-built second recommendation model, and obtain a second weight of the service recommendation corresponding to the user output by the second recommendation model.

[0059] Specifically, based on the second recommendation model, the user's needs are deeply mined mainly through the user's original question and answer text, and then the corresponding recommended services are determined. Therefore, the embodiment of the present invention uses the health question and answer data in the user data as the input of the second recommendation model. It can be understood that the health question and answer data contains any question and answer data of the user about health, and the second recommendation model has the ability to search and retrieve historical data related to it from the background according to the health question and answer data. For example, the acquired health question and answer data is the health question and answer data of the user's second follow-up visit. Then, based on the health question and answer data, the second recommendation model can find the health question and answer data of the user's first consultation from the background to obtain comprehensive health question and answer data information. The second recommendation model determines the second weight of the service recommendation corresponding to the health service suitable for recommendation to the user among all services of the current medical platform based on the acquired health question and answer data and the health question and answer data associated with it.

[0060] According to one embodiment of the present invention, the second recommendation model is trained based on a pre-constructed health question and answer data training set of sample users, and the health question and answer data training set is annotated with corresponding recommended services.

[0061] Specifically, the second recommendation model of the embodiment of the present invention chooses to use the large language LLM (Large Language Model) model to construct an initial second recommendation model, then selects sample users from the background database, and obtains the health question and answer data of the sample users accordingly, and marks the corresponding recommended services for the sample users based on the health service products ordered by the sample users after generating the health question and answer data, constructs a health question and answer data training set, and uses the training set to train the initial second recommendation model, and updates the parameters of the initial second recommendation model. When the training converges, a second recommendation model that can be used for online service recommendations to users is obtained.

[0062] Step S104: Determine the recommended health service corresponding to the user according to the service recommendation first weight and the service recommendation second weight.

[0063] Specifically, based on the first weight of service recommendation and the second weight of service recommendation output by the first recommendation model and the second recommendation model, the first weight and / or second weight corresponding to various health services screened out from all medical services on the platform are obtained, and the first weight and second weight corresponding to each health service are linearly summed to obtain the service weight of each health service. Finally, a specified number of recommended health services are screened out from each health service according to the weight.

[0064] According to one embodiment of the present invention, determining the recommended health service corresponding to the user according to the first weight of the service recommendation and the second weight of the service recommendation includes: determining the corresponding first service to be recommended and the weight corresponding to the first service to be recommended according to the first weight of the service recommendation and a preset first weight coefficient; determining the corresponding second service to be recommended and the weight corresponding to the second service to be recommended according to the second weight of the service recommendation and a preset second weight coefficient; merging the first service to be recommended and the second service to be recommended to obtain the health service to be recommended, and calculating the weight of the health service to be recommended according to the weight corresponding to the first service to be recommended and the weight corresponding to the second service to be recommended; and screening out recommended health services from the health services to be recommended according to the weights of the health services to be recommended.

[0065] Specifically, in order to more flexibly and better meet usage needs, the first weight coefficient and the second weight coefficient corresponding to the first weight of service recommendation and the second weight of service recommendation can also be pre-set to flexibly adjust the proportion of the first weight of service recommendation and the second weight of service recommendation in the final health service recommendation.

[0066] The first weight of the service recommendation is analyzed, and the first recommended service to be recommended to the user and the first weights corresponding to these services are obtained, and the first weight coefficient is applied to the first weight to obtain the weight corresponding to the first recommended service. Similarly, the second weight of the service recommendation is analyzed, and the second recommended service to be recommended to the user and the second weights corresponding to these services are obtained, and the second weight coefficient is applied to the second weight to obtain the weight corresponding to the second recommended service.

[0067] Furthermore, the first service to be recommended and the second service to be recommended are combined to obtain the health service to be recommended and the weight corresponding to the health service to be recommended. It can be understood that the health service in the first service to be recommended and the health service in the second service to be recommended may have the same health service, then correspondingly, it has two weights (from the first weight and the second weight respectively), and only one is retained for the same health service, but the sum of the two weights needs to be used as the weight corresponding to the same health service.

[0068] Finally, the services to be recommended are sorted from large to small according to their corresponding weights, and a specified number of health services corresponding to large weight values ​​are selected from the sorting results as recommended health services.

[0069] Figure 2 is a flow chart of a method for recommending health services according to a reference embodiment of the present invention. As another embodiment of the present invention, Figure 2 As shown, the method recommended by the health service may include:

[0070] Step S201: Acquire user data, where the user data includes the user's health question and answer data, health status data, and behavior data.

[0071] According to one embodiment of the present invention, the health status data includes user personal information, user medical history and user current symptoms; the behavior data is a search behavior sequence consisting of user search behavior and corresponding search keywords within a specified time period before obtaining the user data, and each behavior element in the search behavior sequence has a corresponding influence coefficient.

[0072] Specifically, the embodiment of the present invention defines the specific content of health status data. In order to simplify the data structure, all static data are classified as health status data. The static data in the embodiment of the present invention includes user personal information, user medical history and user current symptoms. These static data are obtained by filtering the user registration and login information and the information recently submitted by the user on the platform website.

[0073] In addition, the behavior data of users in the embodiment of the present invention preferably uses the search behavior of users on the platform website. The search behavior itself clearly represents the needs of users, and the needs of users can be more accurately located through the search keywords corresponding to the search behavior. Therefore, the embodiment of the present invention obtains the search behavior of users on the platform website in the recent period of time, and finds the search keywords corresponding to the search behavior, combines the search behavior and the corresponding search keywords into a behavior element, and sorts them in chronological order to obtain a search behavior sequence. It can be understood that, in view of the time attribute of the search behavior sequence, generally speaking, the search behavior closest to the current service recommendation can better represent the current needs of users, so the corresponding influence coefficient is assigned to each behavior element in the behavior sequence in chronological order to obtain a search behavior sequence with an influence coefficient, so that the acquirer of the search behavior sequence can use the key and important behavior elements therein according to the influence coefficient.

[0074] For example, Figure 3It is a schematic diagram of the data structure composition of the health status data and behavior data of an embodiment of the present invention. The health status data mainly includes three types of static data (personal information I, user disease history M and current symptoms F), among which personal information I is divided into two subcategories according to the degree of association with service recommendations. The information marked with 1 represents information related to service recommendations, such as gender, age, etc. in the figure, and the information marked with 0 represents information unrelated to service recommendations, such as telephone number, name, etc. in the figure. The user's disease history M mainly includes the historical time T of the occurrence of the disease, the description of the condition S, the diagnosis result K at the time, and the corresponding treatment plan P. There is also the user's self-perception of the current symptoms F. In addition, the behavior data is mainly based on the search behavior sequence C. Finally, the data structure of {user ID: personal information I-user disease history M-current symptoms F, search behavior sequence C} is formed.

[0075] Furthermore, based on the data structure of the health status data and behavior data defined in the user dimension, by specifying the data type and field representation of each data, it helps to ensure the consistency of the data and the accuracy of the query.

[0076] Preferably, in view of the high performance and powerful full-text search capability of the distributed search engine database, in order to improve the efficiency of service recommendations, the user data is stored in the distributed search engine database, and the user ID is used as an index to establish a user data storage document. Each user is stored as a document in the corresponding index, which facilitates the use of various query conditions and filters of the distributed search engine database to efficiently and accurately search for user data.

[0077] According to another embodiment of the present invention, after acquiring the user data, the method further includes: performing data cleaning, text vectorization processing and normalization processing on the user data to obtain processed user data, so as to input the health question and answer data, health status data and behavior data in the processed user data into the first recommendation model and the second recommendation model.

[0078] Specifically, considering that user data may come from different business modules, there will inevitably be noise data in the collected user data. Therefore, after obtaining the user data, the user data is first cleaned, including filtering out duplicate data, expired data, and the above Figure 3 The data labeled 0 in the personal information of the user data, such as phone number and name, which are not related to the service recommendation, and in order to ensure the integrity of the data and facilitate subsequent calculations, the filtered data are filled with specified data, such as using 0 to fill the filtered missing data, to obtain the user data after data cleaning and filling processing.

[0079] Furthermore, considering that the subsequent first recommendation model and the second recommendation model are models for numerical data, it is necessary to perform text vectorization processing on the user data. The user data after data cleaning and filling processing is divided into two categories: numerical data and text data. For numerical data, such as age, gender (0-1), height, weight, historical time of disease occurrence T, etc., they can be directly input as input. For text data, such as health problem data, disease description S, and diagnosis result K at the time, text vectorization processing is performed. Specifically, a common weighted technology TF-IDF algorithm used for information retrieval and data mining can be used to extract text features to achieve text vectorization conversion.

[0080] In addition, in order to standardize user data, the data after the above text vectorization processing and the numerical data that have not been processed by text vectorization are normalized and integrated, so that the subsequent first recommendation model and the second recommendation model can directly use the user data that has been cleaned, text vectorized and normalized.

[0081] Figure 4 It is an overall schematic diagram of a method for determining recommended health services including a user data processing flow in an embodiment of the present invention. User data is queried and obtained from a distributed search engine database, and the user data is cleaned to obtain the cleaned user data. Text data in the cleaned user data is vectorized, and the data after text vectorization and the numerical data in the cleaned user data are integrated, and the integrated user data is numerically normalized to obtain user data that has undergone data cleaning, text vectorization, and normalization. Then, using the user data that has undergone data cleaning, text vectorization, and normalization, a first recommendation model and a second recommendation model are used to obtain a first weight for service recommendation and a second weight for service recommendation, thereby obtaining a recommended health service corresponding to the user.

[0082] Step S202, the behavior data includes the search behavior of the user on the website, and the search keywords corresponding to the search behavior; the health status data and the search behavior of the user, as well as the search keywords corresponding to the search behavior are input into a pre-constructed first recommendation model, and the user health behavior characteristics extracted from the health status data, the search behavior and the corresponding search keywords by the first recommendation model are used to determine the user category to which the user belongs, so that the first recommendation model matches the corresponding service according to the user category and outputs the matching weight as the first weight for service recommendation.

[0083] Specifically, the behavior data of the embodiment of the present invention mainly includes a search behavior sequence, which includes the user's search behavior on the website in the recent period of time, and behavior elements composed of search keywords corresponding to the search behavior, and each behavior element has a corresponding influence coefficient. After receiving the health status data and the search behavior sequence, the first recommendation model filters out the search behaviors with less relevance to the service recommendation in the search behavior sequence based on the influence coefficient of each behavior element in the search behavior sequence and the user's medical history and current symptoms in the health status data, and extracts the corresponding features as the user's health behavior features based on the remaining search behaviors and corresponding search keywords in the search behavior sequence, as well as the corresponding influence coefficients, using the convolutional network in the first recommendation model.

[0084] Furthermore, the user category is determined based on the correspondence between the preset user health behavior characteristics and the user category, or based on the preset category characteristics of various user categories, the user category is determined by obtaining the similarity between the user health behavior characteristics and various category characteristics. Specific user categories may include users recommended for cardiovascular and cerebrovascular drug management, users for chronic disease rehabilitation tracking, users recommended for women's maintenance diet therapy, users for consultation with a specialist, etc. The first recommendation model then selects health services that match the user category from all the medical services provided by the current platform website based on the determined user category, and determines the corresponding matching weight based on the degree of matching, and appends the number of the matching health service to the matching weight to obtain the first weight of the service recommendation, that is, the first weight of the service recommendation includes both the matching health service and the weight corresponding to the matching health service.

[0085] The first weight of service recommendation in the embodiment of the present invention is mainly determined from the user's behavior data dimension, that is, the first recommendation model mainly recommends matching health services to users from the user's behavior data dimension, and sorts them according to the first weight, so as to give priority to recommending the health services with the highest matching degree to users, thereby improving the user experience.

[0086] Step S203: input the health question and answer data into a pre-constructed second recommendation model, extract user intention features from the health question and answer data based on the second recommendation model, and determine the user intention of the user, so that the second recommendation model matches the corresponding service according to the user intention and outputs the matching weight as the second weight for service recommendation.

[0087] According to one embodiment of the present invention, the second recommendation model is constructed based on a language model application construction framework, and the prompt words of the second recommendation model come from a prompt word template corresponding to the personal characteristics and symptom characteristics contained in the health status data, and the prompt word template is obtained by optimizing the initial prompt words.

[0088] Specifically, in order to improve the construction efficiency of the second recommendation model, the embodiment of the present invention uses a language model application construction framework (LLM programming framework) model to construct the second recommendation model. The LLM programming framework is a powerful framework designed to help developers build end-to-end applications using language models. The LLM programming framework can support a variety of open source language models and has the ability to integrate external services. It has good flexibility and scalability, can easily interact with various language models, link multiple components together, and integrate additional resources.

[0089] The second recommendation model constructed by the LLM programming framework model supports flexible custom prompts. Based on preset personal characteristics and symptom characteristics, the initial prompt is optimized to obtain prompt templates suitable for different personal characteristics and symptom characteristics, so as to maximize the reuse of prompt templates.

[0090] For example, in the case where the personal characteristics are infants and young children, the interactive user objects are often the parents of infants and young children. Common words and expressions corresponding to various symptom characteristics of infants and young children are collected. Taking into account the anxious mentality of parents, the prompt needs to iteratively optimize the initial prompt words based on the principles of simplicity, clarity, easy to understand, and emotional elements, and finally obtain prompt word templates corresponding to different symptom characteristics under the personal characteristics of infants and young children.

[0091] Take the symptom characteristics of fever for infants and young children as an example of a prompt word template: "Hello, welcome to the online consultation service. Do you want to consult about your baby's health?" "Can you tell me how high my baby's fever is? Are there any other signs of discomfort?" (Description of symptoms) “How long has your baby had these symptoms? When did they start?” (Symptom duration) "What measures have you taken for your baby? For example, have you given him medicine or used physical cooling methods?" (Measures already taken) "Does your baby have a similar medical history? Or is he allergic to certain drugs?" (Medical History Inquiry) "Let me confirm. The baby's temperature is now 38.5° and has lasted for about 6 hours, right?" (Confirmation information) “I understand you may be worried, but don’t worry, we will help you resolve the issue as soon as possible.” (Emotional care) "If your baby has difficulty breathing, keeps crying or is restless, or has a temperature over 39°, please contact a doctor or go to the hospital immediately" (Emergency Guide).

[0092] Similarly, for the case where the individual characteristic is an elderly person, common words for various common symptom characteristics of the elderly are collected. The prompt needs to be based on a friendly and approachable language style and reduce professional terms; complex processes should be broken down into simple steps and the elderly should be guided to complete them one by one; and based on important information, repeated confirmation is required to ensure correct understanding; for those who are not sure how to answer, some preset options can be provided; charts and pictures can be used to enhance the elderly's understanding; and patient communication is given emotionally. Based on these principles, the initial prompt words are iteratively optimized, and finally prompt word templates corresponding to different symptom characteristics under the personal characteristics of the elderly are obtained to help the elderly interact more sequentially and ensure that complete and accurate health information is collected.

[0093] Similarly, for the case where the personal characteristics are young people, common words for various common symptoms and characteristics of young people are collected. The prompt needs to be based on a relaxed and friendly language style; provide concise and clear communication expressions; provide quick options to save interaction time; and appropriately introduce multimedia elements, such as emoticons, pictures and other multimedia elements, to make the interaction vivid. Based on these principles, the initial prompt words are iteratively optimized, and finally prompt word templates corresponding to different symptom characteristics under the personal characteristics of young people are obtained to improve communication efficiency.

[0094] The embodiment of the present invention inputs the acquired health question and answer data into the second recommendation model. The second recommendation model searches for the corresponding prompt word template according to the user's personal characteristics and symptom characteristics, performs template conversion on the health question and answer data according to the found prompt word template, and passes the template converted health question and answer data to the parsing unit in the second recommendation model, so as to facilitate the second recommendation model to have a deeper and more accurate understanding and analysis of the health question and answer data.

[0095] Furthermore, the second recommendation model is based on the acquired templated converted health question and answer data, using the word segmentation module and the semantic parsing module, and combined with its short-term and long-term memory capabilities to find the context, extract the user intention features of the health question and answer data, and obtain the user intention contained in the health question data. Then, according to the user intention, the health service that matches the user intention is screened out from all the medical services provided by the current platform website, and the corresponding matching weight is determined according to the matching degree, and the matching health service number is attached to the matching weight to obtain the second weight of service recommendation, that is, the second weight of service recommendation includes both the matching health service and the weight corresponding to the matching health service.

[0096] Figure 5It is a flow chart of the second recommendation model constructed based on the LLM programming framework model in the embodiment of the present invention to generate the second weight of service recommendation. The user's health question and answer data is obtained, and the health question and answer data is input into the second recommendation model. The second recommendation model determines the prompt word template according to the user's personal characteristics and symptom characteristics, calls the corresponding prompt word template, and performs template conversion on the health question and answer data. The LLM programming framework in the second recommendation model is used to parse the converted health question and answer data and extract the user intention feature to obtain the user intention, and then obtain the second weight of service recommendation through service matching. In addition, the LLM programming framework will also output the main calculation basis for determining the second weight of service recommendation as a reason in the corresponding second weight of service recommendation in the remarks, so as to analyze the results of the recommended service later.

[0097] Step S204: Determine the recommended health service corresponding to the user according to the service recommendation first weight and the service recommendation second weight.

[0098] Figure 6 1 is a flow chart of determining the recommended health service according to an embodiment of the present invention. After receiving a user health service recommendation task, obtaining user data from a distributed search engine database, inputting health status data and behavior data in the user data into a pre-built first recommendation model, and obtaining a first weight A of the service recommendation corresponding to the user output by the first recommendation model. i . Input the health question and answer data in the user data into the pre-built second recommendation model, and obtain the second service recommendation weight Bi corresponding to the user output by the second recommendation model. Merge the first service to be recommended determined by the first weight of the service recommendation and the second service to be recommended determined by the second weight of the service recommendation to obtain the health service to be recommended, and according to the preset first weight coefficient α and the preset second weight coefficient β, the service recommendation first weight A i and service recommendation second weight B i Perform linear weighting to obtain the weight W of the health service to be recommended i =αA i +βB i Finally, according to the calculated weights of the health services to be recommended, the recommended health services for the user are determined from the health services to be recommended, and the corresponding weights are also noted as the reasons for recommending the health services.

[0099] Step S205: monitor the user's feedback information on the recommended health service, and iteratively train the first recommendation model and the second recommendation model according to the feedback, the first service recommendation weight, and the second service recommendation weight.

[0100] Specifically, considering that the first recommendation model and the second recommendation model have iterative training capabilities, in order to further improve the recommendation capabilities of the first recommendation model and the second recommendation model, after determining the recommended health service corresponding to the user, monitor the user's feedback information on the recommended health services pushed by the platform, and update the training sample sets of the first recommendation model and the second recommendation model based on the feedback information, as well as the current service recommendation first weight and the service recommendation second weight. Use the updated training sample sets to iteratively train the corresponding first recommendation model and the second recommendation model, respectively, to obtain further optimized first recommendation models and second recommendation models.

[0101] Figure 7 This is a schematic diagram of the main principles of the method for recommending health services in an embodiment of the present invention. This schematic diagram only illustrates the determination principle of the recommended service from the user feature level. Based on the health status data and behavior data in the user data, the first recommendation model is used to determine the user category from the user behavior dimension, and a personal health portrait of the user is constructed. Based on the health question and answer data in the user data, the second recommendation model is used to extract user intention features from the health question and answer data. The personal health portrait determined by the user behavior and the user intention features are combined to determine the recommended health service for the user and form a knowledge base. Finally, the feedback mechanism is used to iteratively train the first recommendation model and the second recommendation model according to the user's feedback information on the recommended health service, optimize the recommendation capabilities of the first recommendation model and the second recommendation model, and obtain recommended health services with higher user satisfaction.

[0102] Figure 8 Schematic diagram of the main modules of the device for health service recommendation according to an embodiment of the present invention. Figure 8 As shown, the service recommendation device 800 mainly includes a user data acquisition module 801 , a first weight determination module 802 , a second weight determination module 803 and a recommended service determination module 804 .

[0103] A user data acquisition module 801 is used to acquire user data, wherein the user data includes the user's health question and answer data, health status data and behavior data;

[0104] A first weight determination module 802 is used to input the health status data and the behavior data into a pre-built first recommendation model, and obtain a first weight of the service recommendation corresponding to the user output by the first recommendation model;

[0105] A second weight determination module 803 is used to input the health question and answer data into a pre-built second recommendation model to obtain a second weight of the service recommendation corresponding to the user output by the second recommendation model;

[0106] The recommended service determination module 804 is used to determine the recommended health service corresponding to the user according to the service recommendation first weight and the service recommendation second weight.

[0107] According to one embodiment of the present invention, the behavioral data includes the user's search behavior on the website, and the search keywords corresponding to the search behavior; the first weight determination module 802 is also used to: input the health status data and the user's search behavior, and the search keywords corresponding to the search behavior into a pre-constructed first recommendation model, and determine the user category to which the user belongs based on the user health behavior characteristics extracted from the health status data, the search behavior and the corresponding search keywords by the first recommendation model, so that the first recommendation model matches the corresponding service according to the user category and outputs the matching weight as the first weight for service recommendation.

[0108] According to another embodiment of the present invention, the second weight determination module 803 is also used to: input the health question and answer data into a pre-built second recommendation model, extract user intention features of the health question and answer data based on the second recommendation model, determine the user intention of the user, so that the second recommendation model matches the corresponding service according to the user intention and outputs the matching weight as the second weight for service recommendation.

[0109] According to another embodiment of the present invention, the health status data includes user personal information, user medical history and user current symptoms; the behavior data is a search behavior sequence consisting of user search behavior and corresponding search keywords within a specified time period before obtaining the user data, and each behavior element in the search behavior sequence has a corresponding influence coefficient.

[0110] According to another embodiment of the present invention, the second recommendation model is constructed based on a language model application construction framework, and the prompt words of the second recommendation model come from a prompt word template corresponding to the personal characteristics and symptom characteristics contained in the health status data, and the prompt word template is obtained by optimizing the initial prompt words.

[0111] According to another embodiment of the present invention, the recommended service determination module 804 is further used to: determine the corresponding first service to be recommended and the weight corresponding to the first service to be recommended according to the first weight of the service recommendation and the preset first weight coefficient; determine the corresponding second service to be recommended and the weight corresponding to the second service to be recommended according to the second weight of the service recommendation and the preset second weight coefficient; merge the first service to be recommended and the second service to be recommended to obtain a health service to be recommended, and calculate the weight of the health service to be recommended according to the weight corresponding to the first service to be recommended and the weight corresponding to the second service to be recommended; and screen out recommended health services from the health services to be recommended according to the weights of the health services to be recommended.

[0112] According to another embodiment of the present invention, the service recommendation device 800 also includes an iterative training module (not shown in the figure), which is used to: after determining the recommended health service corresponding to the user, monitor the user's feedback information on the recommended health service, and iteratively train the first recommendation model and the second recommendation model based on the feedback, the service recommendation first weight and the service recommendation second weight.

[0113] According to another embodiment of the present invention, the service recommendation device 800 also includes a data cleaning module (not shown in the figure), which is used to: after acquiring the user data, perform data cleaning, text vectorization processing and normalization processing on the user data to obtain processed user data, so as to input the health question and answer data, health status data and behavior data in the processed user data into the first recommendation model and the second recommendation model.

[0114] According to yet another embodiment of the present invention, the user data comes from at least one business scenario of a user consultation scenario, a user health management scenario, and a user medical knowledge question and answer scenario.

[0115] According to another embodiment of the present invention, the first recommendation model is trained based on a pre-constructed training set of health status data and behavior data of sample users, and the health status data and behavior data training set are labeled with corresponding recommended services; the second recommendation model is trained based on a pre-constructed training set of health question and answer data of sample users, and the health question and answer data training set are labeled with corresponding recommended services.

[0116] Fig. 9 is an exemplary system architecture diagram to which embodiments of the present invention can be applied.

[0117] like Fig. 9As shown, system architecture 900 may include terminal devices 901, 902, 903, network 904 and server 905. Network 904 is used to provide a medium for communication links between terminal devices 901, 902, 903 and server 905. Network 904 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0118] Users can use terminal devices 901, 902, 903 to interact with server 905 through network 904 to receive or send messages, etc. Terminal devices 901, 902, 903 can be installed with various communication client applications, such as service recommended applications, etc. (only as an example).

[0119] The terminal devices 901 , 902 , and 903 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, and desktop computers, etc.

[0120] Server 905 may be a server that provides various services, such as a background management server that provides support for service recommendations made by users using terminal devices 901, 902, and 903 (for example only). The background management server may obtain user data, which includes the user's health question and answer data, health status data, and behavior data; input the health status data and the behavior data into a pre-built first recommendation model to obtain a first weight for service recommendations corresponding to the user output by the first recommendation model; input the health question and answer data into a pre-built second recommendation model to obtain a second weight for service recommendations corresponding to the user output by the second recommendation model; determine the recommended health service corresponding to the user based on the first weight for service recommendation and the second weight for service recommendation, and feed back the processing results (such as recommended health services, etc. - for example only) to the terminal device.

[0121] It should be noted that the service recommendation method provided in the embodiment of the present invention is generally executed by the server 905 , and accordingly, the service recommendation device is generally set in the server 905 .

[0122] It should be understood that Fig. 9 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0123] Reference below Fig.10 , Fig.10 It is a schematic diagram of the structure of a computer system of a terminal device or a server suitable for implementing an embodiment of the present invention. Fig.10 The terminal device or server shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0124] like Fig.10 As shown, the computer system 1000 includes a central processing unit (CPU) 1001, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage part 1008 into a random access memory (RAM) 1003. In the RAM 1003, various programs and data required for the operation of the system 1000 are also stored. The CPU 1001, the ROM 1002, and the RAM 1003 are connected to each other through a bus 1004. An input / output (I / O) interface 1005 is also connected to the bus 1004.

[0125] The following components are connected to the I / O interface 1005: an input section 1006 including a keyboard, a mouse, etc.; an output section 1007 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 1008 including a hard disk, etc.; and a communication section 1009 including a network interface card such as a LAN card, a modem, etc. The communication section 1009 performs communication processing via a network such as the Internet. A drive 1010 is also connected to the I / O interface 1005 as needed. A removable medium 1011, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 1010 as needed, so that a computer program read therefrom is installed into the storage section 1008 as needed.

[0126] In particular, according to the embodiments disclosed in the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present invention include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 1009, and / or installed from the removable medium 1011. When the computer program is executed by the central processing unit (CPU) 1001, the above-mentioned functions defined in the system of the present invention are executed.

[0127] It should be noted that the computer-readable medium shown in the present invention may be a computer-readable signal medium or a computer-readable storage medium or any combination of the two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present invention, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination thereof.

[0128] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present invention. In this regard, each box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a sequence different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0129] The units involved in the embodiments of the present invention may be implemented in software or hardware. The units described may also be set in a processor, for example, may be described as: a processor including: a user data acquisition module, a first weight determination module, a second weight determination module and a recommendation service determination module.

[0130] Among them, the names of these modules do not constitute a limitation of the modules themselves in certain cases. For example, the recommended service determination module can also be described as "a module for determining the recommended health service corresponding to the user based on the first weight of the service recommendation and the second weight of the service recommendation."

[0131] On the other hand, the present invention also provides a computer-readable medium, which may be included in the device described in the embodiment; or it may exist independently without being assembled into the device. The computer-readable medium carries one or more programs, and when the one or more programs are executed by a device, the device includes: obtaining user data, the user data including the user's health question and answer data, health status data and behavior data; inputting the health status data and the behavior data into a pre-built first recommendation model, and obtaining the first weight of the service recommendation corresponding to the user output by the first recommendation model; inputting the health question and answer data into a pre-built second recommendation model, and obtaining the second weight of the service recommendation corresponding to the user output by the second recommendation model; and determining the recommended health service corresponding to the user according to the first weight of the service recommendation and the second weight of the service recommendation.

[0132] The technical solution according to the embodiment of the present invention has the following advantages or beneficial effects: by obtaining user data, the user data includes the user's health question and answer data, health status data and behavior data; inputting the health status data and behavior data into a pre-constructed first recommendation model, and obtaining the first weight of the service recommendation corresponding to the user output by the first recommendation model; inputting the health question and answer data into a pre-constructed second recommendation model, and obtaining the second weight of the service recommendation corresponding to the user output by the second recommendation model; according to the first weight of the service recommendation and the second weight of the service recommendation, a technical solution for determining the recommended health service corresponding to the user is realized, a personalized and accurate health service recommendation is realized, user data valuable for health service recommendation is fully mined, and the final recommended health service is collaboratively determined through the weights output by the two recommendation models, while improving the personalization of the recommended health service, it also ensures the accuracy of the recommended health service, thereby improving the user's adoption rate of the recommendation results, so as to provide users with professional and personalized health services and better serve users.

[0133] The specific implementation methods described herein do not constitute limitations on the scope of protection of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may occur depending on design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention shall be included in the scope of protection of the present invention.

Claims

1. A method for recommending health services, characterized in that: include: Acquire user data, wherein the user data includes the user's health question and answer data, health status data, and behavior data; Inputting the health status data and the behavior data into a pre-built first recommendation model, and obtaining a first weight of a service recommendation corresponding to the user output by the first recommendation model; Inputting the health question and answer data into a pre-built second recommendation model, and obtaining a second weight of service recommendation corresponding to the user output by the second recommendation model; The recommended health service corresponding to the user is determined according to the service recommendation first weight and the service recommendation second weight.

2. The method according to claim 1, characterized in that The behavior data includes the user's search behavior on the website and the search keywords corresponding to the search behavior; Inputting the health status data and the behavior data into a pre-built first recommendation model, and obtaining a first weight of a service recommendation corresponding to the user output by the first recommendation model, comprises: The health status data, the user's search behavior, and the search keywords corresponding to the search behavior are input into a pre-constructed first recommendation model, and the user category to which the user belongs is determined based on the user health behavior characteristics extracted from the health status data, the search behavior and the corresponding search keywords by the first recommendation model, so that the first recommendation model matches the corresponding service according to the user category and outputs the matching weight as the first weight for service recommendation.

3. The method according to claim 1, characterized in that Inputting the health question and answer data into a pre-built second recommendation model, and obtaining a second weight of service recommendation corresponding to the user output by the second recommendation model, including: The health question and answer data is input into a pre-built second recommendation model, and the user intent features of the health question and answer data are extracted based on the second recommendation model to determine the user intent of the user, so that the second recommendation model matches the corresponding service according to the user intent and outputs the matching weight as the second weight for service recommendation.

4. The method according to claim 1, characterized in that: The health status data includes user personal information, user medical history and user current symptoms; the behavior data is a search behavior sequence consisting of user search behavior and corresponding search keywords within a specified time period before obtaining the user data, and each behavior element in the search behavior sequence has a corresponding influence coefficient.

5. The method according to claim 1, characterized in that The second recommendation model is constructed based on a language model application construction framework, and the prompt words of the second recommendation model come from the prompt word template corresponding to the personal characteristics and symptom characteristics contained in the health status data, and the prompt word template is obtained by optimizing the initial prompt words.

6. The method according to claim 1, characterized in that Determining the recommended health service corresponding to the user according to the service recommendation first weight and the service recommendation second weight includes: Determine a corresponding first service to be recommended and a weight corresponding to the first service to be recommended according to the service recommendation first weight and a preset first weight coefficient; Determine a corresponding second service to be recommended and a weight corresponding to the second service to be recommended according to the service recommendation second weight and a preset second weight coefficient; The first service to be recommended and the second service to be recommended are combined to obtain a health service to be recommended, and a weight of the health service to be recommended is calculated according to a weight corresponding to the first service to be recommended and a weight corresponding to the second service to be recommended; According to the weights of the health services to be recommended, recommended health services are screened out from the health services to be recommended.

7. The method according to claim 1, characterized in that After determining the recommended health service corresponding to the user, the method further includes: The user's feedback information on the recommended health service is monitored, and the first recommendation model and the second recommendation model are iteratively trained according to the feedback, the first service recommendation weight, and the second service recommendation weight.

8. The method according to claim 1, characterized in that After obtaining the user data, the method further includes: The user data is cleaned, text vectorized and normalized to obtain processed user data, so as to input the health question and answer data, health status data and behavior data in the processed user data into the first recommendation model and the second recommendation model.

9. The method according to claim 1, characterized in that: The user data comes from at least one business scenario among a user consultation scenario, a user health management scenario, and a user medical knowledge question and answer scenario.

10. The method according to claim 1, characterized in that The first recommendation model is trained based on a pre-constructed training set of health status data and behavior data of sample users, wherein the training set of health status data and behavior data is labeled with corresponding recommended services; The second recommendation model is trained based on a pre-constructed health question and answer data training set of sample users, and the health question and answer data training set is annotated with corresponding recommended services.

11. A device for recommending health services, characterized in that: include: A user data acquisition module, used to acquire user data, wherein the user data includes the user's health question and answer data, health status data and behavior data; A first weight determination module, configured to input the health status data and the behavior data into a pre-built first recommendation model, and obtain a first weight of a service recommendation corresponding to the user output by the first recommendation model; A second weight determination module is used to input the health question and answer data into a pre-built second recommendation model to obtain a second weight of the service recommendation corresponding to the user output by the second recommendation model; The recommended service determination module is used to determine the recommended health service corresponding to the user according to the service recommendation first weight and the service recommendation second weight.

12. A mobile electronic device terminal, characterized in that: include: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 10.

13. A computer readable medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 10 is implemented.

14. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 10 is implemented.

Images