Health data capsule personal health information pushing method and device

Abstract

The invention discloses a health data capsule personal health information pushing method and device, and the method comprises the steps: obtaining target data related to a user, and carrying out the preprocessing of the target data through a preprocessing process matched with the characteristics of the target data; extracting data features of the preprocessed data, and generating a user portrait in combination with a user health data capsule personal data basis; according to each recommendation reference factor, executing an information pushing operation; the execution of the information pushing operation comprises the following steps: for any type of health service, determining a recommendation scheme matched with the type of health service; according to the user portrait and the user-related timely health data, in combination with the health service resource data, determining a health service recommendation scheme matched with the health demand of the user, determining health information needing to be pushed under the health service, and pushing the health information according to a preset pushing strategy; and executing information push optimization operation, and pushing the optimized health information under various health services according to the optimized push strategy.

Description

Technical Field

[0001] The present application relates to the field of information recommendation technology, and in particular to a method and device for pushing personal health information in a health data capsule. Background Art

[0002] Driven by digital advancements, the healthcare system is undergoing profound transformation. Traditional medical services are often limited by time and location, resulting in limited efficiency. Existing healthcare push services suffer from several design flaws. First, existing healthcare push systems face challenges integrating and mining high-quality, multimodal data. These systems struggle to integrate health data from diverse sources, such as personal health records and questionnaires. These data often come in diverse formats and vary in quality, limiting the depth and precision of data mining and, consequently, the accuracy and relevance of recommendations. Second, the recommendation algorithms in existing healthcare push systems are often not specifically designed for different service types. For example, services such as health management, disease prevention, and doctor recommendations each have their own unique characteristics, but many systems use a one-size-fits-all recommendation algorithm that fails to effectively adapt to these differences. This results in inaccurate recommendations that fail to meet users' personalized needs. Furthermore, data security and privacy protection are significant issues. User health data often contains sensitive information, making ensuring secure storage and transmission of this data and preventing unauthorized access a key challenge in system design. Current security measures are inadequate, posing a significant risk of data leakage and undermining user trust in the system. Finally, the collection and analysis of user reviews and feedback are also inadequate. Existing health service push systems often lack effective mechanisms to leverage user feedback to optimize recommendation algorithms and improve service quality, resulting in suboptimal service optimization. Summary of the Invention

[0003] The present application provides a method and device for pushing personal health information via a health data capsule, which is used to solve the technical problem of how to improve the effect and efficiency of health information push.

[0004] To solve the above technical problems, this application provides the following technical solutions:

[0005] This application provides a method for pushing personal health information in a health data capsule, the method comprising:

[0006] Acquire user-related target data, and preprocess the target data using a preprocessing process that matches the characteristics of the target data;

[0007] Utilize machine learning algorithms to extract data features from pre-processed data, and generate user profiles based on the user's health data capsule personal data and the data features;

[0008] Executing an information push operation based on each recommendation reference factor; wherein the recommendation reference factor includes the user profile and timely health data related to the user; executing the information push operation includes: for any type of health service, determining a recommended solution that matches the health service; determining the health information that needs to be pushed under the health service based on the user profile and timely health data related to the user, combined with the health service resource data and the recommended solution that matches the health service, and pushing the health information according to a preset push strategy;

[0009] Perform information push optimization operations, including: optimizing health information and push strategies under various health services based on optimization reference factors, and pushing optimized health information under various health services according to the optimized push strategies.

[0010] Optionally, preprocessing the target data using a preprocessing process that matches the characteristics of the target data includes:

[0011] According to the structural or non-structural characteristics of the target data, a matching preprocessing process is determined, and the target data is preprocessed through the determined preprocessing process.

[0012] Optionally, the health service includes a health information service, and the recommendation scheme corresponding to the health information service is an association rule matching scheme.

[0013] Optionally, based on the user profile and timely health data related to the user, combined with the health service resource data and the recommended solution matching the health service of this type, the health information that needs to be pushed under the health service of this type is determined, including:

[0014] For health information services, the health information that needs to be pushed under the health information service is determined based on the matching of the characteristics of the user portrait and the user's timely health data with the rule information matching library, combined with the health service resource data.

[0015] Optionally, the health service includes a task management service, and the recommendation plan corresponding to the task management service is a phased task recommendation plan.

[0016] Optionally, based on the user profile and timely health data related to the user, combined with the health service resource data and the recommended solution matching the health service of this type, the health information that needs to be pushed under the health service of this type is determined, including:

[0017] For task management services, corresponding health tasks are formulated based on the characteristics of the user portrait and the user's timely health data, combined with health service resource data, and the health tasks are used as health information that needs to be pushed under the task management service.

[0018] Optionally, based on the user profile and the characteristics of the user's timely health data, combined with health service resource data, corresponding health tasks are formulated, including:

[0019] Obtain target tasks regularly and determine the predicted progress of the corresponding target tasks;

[0020] Comparing the predicted progress with the preset progress of the target task to determine the user's progress evaluation result;

[0021] Based on the characteristics of user portraits and user-related timely health data, combined with health service resource data and user progress evaluation results, corresponding health tasks are formulated.

[0022] Optionally, the health service includes a dynamic information service, and the recommendation scheme corresponding to the dynamic information service is a classic recommendation algorithm scheme.

[0023] Optionally, based on the user profile and timely health data related to the user, combined with the health service resource data and the recommended solution matching the health service of this type, the health information that needs to be pushed under the health service of this type is determined, including:

[0024] For dynamic information services, a variety of classic recommendation algorithms are used to determine the similarity between the characteristics of the user portrait and user-related timely health data and the object portrait characteristics, and combined with health service resource data to determine the health information that needs to be pushed under dynamic information services.

[0025] This application provides a device for pushing personal health information using a health data capsule, the device comprising:

[0026] A data processing module is used to obtain target data related to the user and pre-process the target data through a pre-processing process that matches the characteristics of the target data;

[0027] A user portrait module is used to extract data features of pre-processed data using a machine learning algorithm, and generate a user portrait based on the user's personal data in the health data capsule and the data features;

[0028] An information push module is configured to execute an information push operation based on various recommendation reference factors, wherein the recommendation reference factors include the user profile and timely health data related to the user; executing the information push operation includes: determining, for any type of health service, a recommended solution matching that type of health service; determining, based on the user profile and timely health data related to the user, the recommended solution matching that type of health service in combination with the health service resource data, health information that needs to be pushed under that type of health service, and pushing the health information according to a preset push strategy;

[0029] The push optimization module is used to perform information push optimization operations, including: optimizing the health information and push strategies under various health services based on optimization reference factors, and pushing the optimized health information under various health services according to the optimized push strategies.

[0030] At least one of the above technical solutions adopted in this application can achieve the following beneficial effects:

[0031] This application adopts corresponding pre-processing processes for target data with different characteristics, which can improve the data processing effect and efficiency, make the efficiency after processing more conducive to feature extraction, effectively improve the utilization efficiency of data, and also help improve the effect and efficiency of health information push. According to different health service types and characteristics, the corresponding recommendation scheme is adopted for health information push, which can effectively improve the accuracy and effectiveness of health information push, thereby improving the effect and efficiency of health information push. Health information push refers to factors such as user portraits and user-related timely health data, which can further improve the accuracy and effectiveness of health information push, thereby improving the effect and efficiency of health information push. Through information push optimization operations, the information push content and strategy can be adjusted in a timely manner, making the health information push more adaptive, and can always and better match the user's needs and preferences, thereby improving the effect and efficiency of health information push. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions in the embodiments of this specification or the prior art, the following briefly describes the drawings required for use in the embodiments of this specification or the prior art description. Obviously, the following only describes the drawings required for use in some embodiments of this application. For those of ordinary skill in the art, other drawings can be derived from these drawings without inventive effort.

[0033] Figure 1 This is a flowchart of the method for pushing personal health information in the health data capsule provided in the first embodiment of this specification.

[0034] Figure 2 This is a schematic diagram of the module architecture of the main body for executing the method for pushing personal health information of the health data capsule in the first embodiment of this specification.

[0035] Figure 3 This is a schematic diagram of the data feature extraction process in the first embodiment of this specification.

[0036] Figure 4 This is a schematic diagram of the information push process under the health information service in the first embodiment of this specification.

[0037] Figure 5 This is a schematic diagram of the information push process under the task management service in the first embodiment of this specification.

[0038] Figure 6 This is a schematic diagram of the information push process under the dynamic information service in the first embodiment of this specification.

[0039] Figure 7 It is a structural diagram of the health information push device in the second embodiment of this specification. DETAILED DESCRIPTION

[0040] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments involved in the specific implementation methods are only part of the embodiments of this application, not all of the embodiments. All other embodiments obtained based on the embodiments in the specific implementation methods by those skilled in the art without making creative work should fall within the scope of protection of this application.

[0041] It should be noted that the term "comprise" and any variations thereof are intended to be inclusive rather than exclusive. For example, a process or method comprising a series of steps or modules is not necessarily limited to those steps or modules explicitly listed, but may include other steps or modules not explicitly listed, or a combination of other steps or modules inherent to such process or method.

[0042] The first embodiment of this specification (hereinafter referred to as "Example 1") provides a method for pushing personal health information of a health data capsule. The execution subject of Example 1 includes but is not limited to a terminal or a server or an operating system or an application, that is, the execution subject can be diverse and can be set, used or changed as needed. In addition, a third-party application can also assist the execution subject in executing Example 1. For example, the health data capsule personal health information push method in Example 1 can be executed by a server, and a corresponding application can be installed on a terminal (the terminal can be held by a user). Data can be transmitted between the terminal or the application and the server, thereby assisting the server in executing the health data capsule personal health information push method in Example 1.

[0043] like Figure 1 As shown, the method for pushing personal health information in a health data capsule provided in Example 1 includes:

[0044] S101: Acquire user-related target data, and preprocess the target data using a preprocessing process that matches the characteristics of the target data;

[0045] In the first embodiment, user-related data (which may be referred to as "target data") may be obtained from different data sources through various suitable methods. Figure 2 The target data may also be multi-dimensional and can be used to describe the multi-dimensional features of the user portrait. The following describes the available methods for obtaining target data (embodiment 1 is not limited to the following methods):

[0046] Method 1: Target data can be obtained from stored data, for example, user-related target data can be obtained from a cloud server or other data storage system (can be obtained from Figure 2 The archive acquisition module 111 executes).

[0047] Method 2: Users can enter relevant data through appropriate methods (including but not limited to online questionnaires, health assessment tools, etc.) Figure 2 The information acquisition module 112 in the embodiment is executed) to obtain the target data.

[0048] Method 3: You can collect the user's historical operation data (including but not limited to the user's historical browsing records and operation records in the relevant system, platform or application, which can be Figure 2 The record acquisition module 113 in the process is executed) to obtain the target data.

[0049] As described above, the target data can be multi-dimensional, including but not limited to personal health records (including but not limited to basic information such as age, gender, height and weight, and / or health history information such as previous diseases, allergies, family medical history, and surgical history, and / or information on recent physical symptoms such as headaches and fatigue, and / or data on living habits such as eating habits, sleep conditions, and exercise frequency), health concepts, living habits, work environment, emotional state, interest preferences, and other data, not limited to embodiment one. In addition, the target data can include both objective data and the user's subjective emotional information and objective environmental information. For example, the data obtained through method two can reflect the user's subjective emotional information and objective environmental information, which is conducive to a more comprehensive understanding of the user's situation.

[0050] Each of the above methods for acquiring target data has its own unique characteristics. For example, data acquired through method 1 can comprehensively and accurately reflect user information, facilitating personalized health recommendations. Data acquired through method 2 provides a comprehensive understanding of a user's lifestyle and health concerns. Data acquired through method 3 is particularly critical for real-time recommendations for dynamic information services. For example, if a user frequently browses content related to chronic disease management, analyzing this behavior can optimize health service delivery and provide recommendations that better suit the user's interests.

[0051] Various methods for acquiring target data can be combined to obtain more comprehensive and detailed target data, which is more conducive to providing personalized health recommendations for users. For example, user input can be used to obtain user interest preference data, while historical user operation data can be used to further refine user interests and preferences. By combining various methods for acquiring target data, user data can be acquired and integrated from multiple dimensions, thereby achieving more accurate and personalized health service recommendations.

[0052] Optionally, for different categories of keyword information, the three methods can be combined for each keyword to calculate the recommendation degree E. For example, for keyword Key1, the weight of data A obtained by method 1 is set to α, the weight of data B obtained by method 2 is set to β, and the weight of data C obtained by method 3 is set to γ. The final recommendation degree E is calculated using the following formula:

[0053] E=αA+βB+((Tt) / T+γ)*C,

[0054] Where T is the number of days in a month, typically 30, and t is the actual date of the data obtained in method 3, the number of days from the current date. This gives more recent data a higher weight, and (Tt) / T+ is used together with the γ value as the weight of data C. If t is greater than 30, t is set to 30, where α+β+γ=1. This yields the recommendation score E for keyword Key1. Similarly, the same calculation can be performed for keyword Key2, obtaining the recommendation scores of different keywords, ranking them, and obtaining the optimal recommended keyword, which can then be used to recommend relevant information.

[0055] In the first embodiment, the acquired target data can be preprocessed (by Figure 2 The image data processing module 12 is executed), and the target data is preprocessed through a preprocessing process that matches the characteristics of the target data.

[0056] Among them, preprocessing the target data through a preprocessing process that matches the characteristics of the target data may include the following contents (these contents can be Figure 2(Executed by the data preprocessing module 121 in the data preprocessing module): According to the structural or unstructured characteristics of the target data, a matching preprocessing process is determined, and the target data is preprocessed through the determined preprocessing process. Specifically, for structured target data, the matching preprocessing process may include data cleaning, formatting, standardization and other processing processes to deal with the problems of missing values, abnormal values ​​and inconsistent data formats that may exist in the target data. For unstructured target data, if it is text data, the matching preprocessing process may include text cleaning, vocabulary standardization, text segmentation, text annotation and vocabulary encoding and other processing processes, so as to convert the text into a feature vector combination representation; if it is image data, the matching preprocessing process may include resizing, cropping, denoising and normalization and other processing to ensure that the image is suitable for the input requirements of the neural network model.

[0057] The above preprocessing ensures that the preprocessed data is more suitable for further analysis and processing, including data feature mining. Furthermore, methods such as topic identification and statistical learning can be combined to mine data features and complete the classification and induction of user profiles to form a complete user profile, as explained below.

[0058] S103: Using a machine learning algorithm to extract data features from the preprocessed data, and generating a user profile based on the user's health data capsule personal data and the data features;

[0059] In the first embodiment, the data features of the pre-processed data can be extracted using a machine learning algorithm (which can be obtained by Figure 2 The deep health feature mining module 122 in the data is executed), and a user profile is generated based on the data features (which can be performed by Figure 2 Specifically, a data feature extraction model can be constructed using a supervised / unsupervised machine learning algorithm, and the preprocessed data can be input into the data feature extraction model to deeply mine the preprocessed data and extract the data features of the preprocessed data (including but not limited to health features and potential patterns), thereby providing support for personalized health recommendations and intervention strategies. The architecture of the established data feature extraction model can be set as needed, and is not limited to the first embodiment.

[0060] like Figure 3 As shown, data features can be extracted in the following ways:

[0061] Step S110: Use statistical learning methods (including but not limited to chi-square test, correlation analysis, and other statistical learning methods) to identify and select features that have a significant impact on health status. At the same time, health indicators and change trends can be calculated based on existing features to create new features and enhance the expressiveness of the data. Clustering algorithms can also be used to discover natural groupings in user health data and identify groups with different health statuses.

[0062] Step S120: Apply supervised / unsupervised machine learning models, such as basic models (e.g., decision trees, linear regression, SVM, etc.), ensemble models (e.g., random forests, boosted trees, etc.), various neural networks (e.g., convolutional neural networks and recurrent neural networks, etc.), and clustering and dimensionality reduction methods (e.g., k-means clustering, principal component analysis, etc.), to identify key features and potential patterns in the preprocessed data, deeply explore complex features and long-term dependencies in the data, and generate data features, such as user in-depth health features;

[0063] In step S130, cross-validation techniques are used to calculate the model's accuracy, recall, F1 score, and other indicators to evaluate the effectiveness of feature extraction and pattern recognition. Pre-defined conditions can be set, and data features that meet these conditions are considered usable data features.

[0064] Taking a decision tree model as an example, a user's basic health data (such as age, blood pressure, body mass index, family medical history, etc., typically pre-processed data) can be used to predict their risk of developing chronic diseases (such as diabetes and hypertension). Based on these input features, the decision tree identifies the key features most strongly associated with chronic diseases and provides personalized prevention and management plans for high-risk users. Furthermore, when detecting lung nodules in X-rays, convolutional neural networks (CNNs) can automatically identify subtle features in the images, significantly improving diagnostic accuracy and efficiency. By training on large numbers of labeled images, CNNs can learn to extract key information from complex imaging data, helping doctors more quickly detect abnormal nodules, thereby improving diagnostic processes and enabling earlier intervention. Furthermore, when processing large datasets containing multiple health indicators, PCA reduces the data dimensionality to its principal components, retaining the most important information while removing redundant features and leaving behind abstract health features with less interpretability. This not only speeds up model training and improves computational efficiency, but also makes data visualization and interpretation more intuitive, thus providing a clearer foundation for subsequent data analysis and model building, and then using the constructed model to extract data features.

[0065] In Example 1, a user portrait can be generated based on the above-mentioned data features and in combination with the personal data of the user's health data capsule through a suitable method (including but not limited to cluster analysis and / or pattern recognition method). The above-mentioned target data may be target data related to a single user, and the generated user portrait may be a user portrait of a single user. In particular, the above-mentioned target data may be target data related to multiple users (i.e., group users), and the generated user portrait may be a group user portrait, which may be used to describe the commonalities and differences of group characteristics and support group-level health management and decision-making. Among them, the group user portrait can be defined and generated through multiple dimensions, for example, it can be classified, defined, and generated from the health risk level dimension (corresponding to the health risk level portrait) and the user participation level dimension (corresponding to the user participation level portrait).

[0066] Health data capsules are defined as a blockchain-based toolkit for personal lifelong health-related data, including personal identity verification and personal health data archiving. Health data capsules are owned and managed independently by individuals, and can be authorized for access and use by others. Health data capsules contain personal health records (PHR), including:

[0067] PHR identification information refers to the digital identity created by users when they register on the relevant platform. This identity usually includes a public key and a private key. Users can generate their own public and private keys to create their own unique digital identity. The public key is synchronized with the platform, and the private key is kept by the user.

[0068] Digital identity is based on public key encryption and asymmetric cryptography, which enable it to effectively support the secure storage and transmission of personal health records.

[0069] A digital identity consists of a public key and a private key. The public key is used to receive PHR, while the private key is used to sign and authorize access to the PHR. These key pairs are randomly generated and uniquely associated with each user.

[0070] When a user wishes to store their PHR, they sign the hash value of the information using their private key and send the signed information to nodes on the network for verification. If the data is verified, the hash value and the corresponding PHR information are stored in the authorized partition. During this entire process, the private key is not transmitted; only the digital signature is broadcast to the network. Similarly, accessing a user's PHR information requires the user's private key signature authorization.

[0071] Health information is used to record data related to individual health;

[0072] Service information is a set of related tools for user health record data services, including self-maintenance of health record data, such as self-entry, authorization of wearable device collection result data merging, self-query, authorized access, and other user-oriented functions.

[0073] Management information is a set of related tools for user health record data management, including identity authentication, data archiving, data credibility verification, institutional access, public key management and other system management-oriented function sets.

[0074] In the first embodiment, the personal data basis of the user health data capsule refers to the user's personal health data contained in the health data capsule.

[0075] Taking health risk level profiling as an example, it can be generated through cluster analysis based on multi-dimensional indicators (i.e., extracted data features) such as the user's living environment, health status, potential chronic diseases, and psychological state, as well as the user's personal data in the health data capsule, to divide users into low-risk, medium-risk, and high-risk groups. Low-risk groups generally exhibit relatively good health and can be provided with routine health maintenance and preventive advice; medium-risk groups may have potential health risks and need to be closely monitored and receive more personalized health management plans to help them effectively prevent disease; high-risk groups face serious health threats and should be provided with emergency intervention and intensive medical service recommendations to ensure their health is continuously monitored and managed.

[0076] Taking the user engagement profile as an example, the user engagement profile can be divided into entry-level, stable, high-value, and outlier levels by analyzing the user's interaction behavior with various health service systems or platforms (i.e., extracted data features) and the personal data of the user's health data capsule. Entry-level users are usually novices or users with low usage frequency. Their engagement can be improved by pushing basic health information and providing guiding services; stable users have formed relatively stable usage habits and can continue to provide corresponding health services to maintain their stickiness and activity in the corresponding health services; high-value users generally use various health services frequently and have a high dependence on various health service systems or platforms. They can be provided with advanced customized health management services to further enhance user loyalty and satisfaction; for outlier users, push strategies and service content can be adjusted through behavioral analysis in an attempt to re-attract and reactivate their usage.

[0077] Through the above classification and profiling based on health risks and user participation levels, we can carry out more refined user management, optimize the push effect of health services, and improve overall operational efficiency and service quality through continuous iteration.

[0078] In the first embodiment, various data (including but not limited to target data and / or pre-processed data and / or data features) can be stored in a secure database (which can be Figure 2 The portrait data storage module 13 in the image data storage module 13 is executed) and provides efficient data access capabilities, data integrity protection and security encryption measures. Among them, the data to be stored can be encrypted (can be encrypted by Figure 2 The encryption module 131 in the storage module is executed) and combined with one or more privacy protection mechanisms to ensure the security of data during storage and transmission to prevent unauthorized access and tampering. In addition, by interacting with the secure storage channel, the encrypted data (which can be stored by Figure 2 At the same time, it provides a fast retrieval and access interface to ensure the integrity and availability of data.

[0079] In addition, Example 1 does not limit the specific content of the encryption algorithm used in the above-mentioned encryption process and the privacy protection mechanism used. For example, for static data stored in the database, Advanced Encryption Standard (AES) can be used for symmetric encryption to ensure efficient encryption and decryption of data. For dynamic data that needs to be protected during transmission, an asymmetric encryption algorithm, such as RSA, can be used to ensure secure key transmission and data integrity. In addition, when selecting an encryption algorithm, hybrid encryption technology can be introduced to combine symmetric encryption with asymmetric encryption to balance security and efficiency. To further enhance privacy protection, single or more privacy protection mechanisms, such as differential privacy or homomorphic encryption, can be combined. Differential privacy can effectively protect user privacy information and prevent the user's personal data from being exposed even during data analysis. Homomorphic encryption allows calculations to be performed on encrypted data, avoiding the exposure of its content when decrypting the data, thereby further improving the security of data processing. Through these multi-layered security mechanisms, the security of the data that needs to be stored is ensured during storage and transmission, preventing unauthorized access and tampering.

[0080] The first embodiment does not limit the storage channels and retrieval methods adopted. For example, secure storage channels may include various distributed storage systems (such as IPFS, Hadoop HDFS, etc.) and cloud servers (such as AWS, Azure, Google Cloud, etc.), which provide data storage solutions with high reliability and scalability. In these channels, data is stored after being encrypted to prevent unauthorized access and tampering. In order to ensure the integrity and availability of the data, a fast retrieval and access interface can be provided. Among them, the retrieval method may include precise retrieval based on hash values, full-text index search, and multi-dimensional query based on attributes to meet the query requirements in different scenarios. For example, on a large-scale data set, full-text retrieval technology (such as Elasticsearch) can be used to quickly locate relevant records, and when a specific user data needs to be efficiently retrieved, a hash-based index mechanism can be used for fast access. In addition, dynamic retrieval optimization, such as caching strategies and parallel processing technologies, can also be supported to further improve the response speed of data retrieval.

[0081] The above modules 11, 12, 13, 111, 112, 113, 121, 122, 123, 131, and 132 can be integrated into a user portrait generation module 1.

[0082] S105: Executing an information push operation based on each recommendation reference factor; wherein the recommendation reference factor includes the user profile and timely health data related to the user; executing the information push operation includes: for any type of health service, determining a recommendation plan that matches the health service; determining health information that needs to be pushed under the health service type based on the user profile and timely health data related to the user, combined with the health service resource data and the recommended plan that matches the health service type, and pushing the health information according to a preset push strategy;

[0083] In the first embodiment, the user profile and user-related timely health data can be used as recommendation reference factors. Among them, the user-related timely health data can refer to the user-related health data within a period of time before the information push operation is performed, that is, the user's recent relevant health data, that is, timely health data focuses on the real-time nature of the data.

[0084] In the first embodiment, different types of health services can be provided, and each type of health service can have a matching recommendation solution. Specifically, health services may include health information services, for which the corresponding recommendation solution is an association rule matching solution; and / or health services may include task management services, for which the corresponding recommendation solution is a phased task recommendation solution; and / or health services may include dynamic information services, for which the corresponding recommendation solution is a classic recommendation algorithm solution.

[0085] In the first embodiment, an information push operation can be performed based on various recommendation reference factors. The information push operation may include: determining a recommendation scheme that matches any type of health service; determining the health information that needs to be pushed under this type of health service based on the user portrait and the timely health data related to the user, combined with the health service resource data and the recommendation scheme that matches this type of health service, and pushing the health information according to the preset push strategy. The following is a detailed description:

[0086] As mentioned above, each type of health service can have a matching recommendation plan. For any type of health service, a recommendation plan that matches this type of health service can be determined. Based on the user portrait and the timely health data related to the user, combined with the health service resource data and the recommendation plan that matches this type of health service, the health information that needs to be pushed under this type of health service is determined. The following is an explanation of the following situations (Example 1 is not limited to the following situations):

[0087] Case 1: Health information services

[0088] For health information services, based on user profiles and timely health data related to users, combined with health service resource data and recommended solutions matching this type of health service, the health information that needs to be pushed under this type of health service is determined, which may include:

[0089] For health information services, according to the matching of the characteristics of "user portraits and timely health data related to users" with the rule information matching library (that is, matching the characteristics of user portraits and timely health data related to users with the rule information matching library), combined with health service resource data, determine the health information that needs to be pushed under the health information service (which can be Figure 2 The health service resource data herein may refer to resource data included in health information services, such as health knowledge, health information and other resource data.

[0090] like Figure 4 As shown, information push under health information services can be carried out in the following ways (not limited to the following ways):

[0091] Step S210: extract key features (which may be referred to as user profile features, including but not limited to age, gender, health status, and historical behavior) from the user profile and user-related timely health data. These features will serve as the basic input for rule matching;

[0092] Step S220: Build a rule base based on the established user profile features. These rules can be derived from domain experts’ knowledge, from existing user data, or through some classic association rule mining algorithms.

[0093] Step S230: During the rule matching process, the extracted user profile features are compared with the conditions in the rule library one by one. For each rule that meets the conditions, the corresponding recommendation result is triggered;

[0094] In step S240, based on the matched rules and combined with the health service resource data, personalized recommendations are generated for the user, specifically the health information that needs to be pushed under the health information service. This health information can include specific health service suggestions, medical resource suggestions, health management suggestions, or health consultations, such as healthy diet advice, health checkup services, and disease prevention information.

[0095] For example, suppose that the key features of user A (i.e., user profile features) are extracted from the user portrait, such as age 45, female gender, high blood pressure, and frequent browsing of information related to cardiovascular disease in the past year. A rule base can be pre-built. Assume that the rule base includes the following rule: "If the user is over 40 years old and has high blood pressure, and has browsed information related to cardiovascular disease in the past year, then cardiovascular health management services are recommended." During the rule matching process, the characteristics of user A are compared one by one with the conditions in the rule base, and it is found that user A meets all the conditions of the rule. Therefore, the relevant recommendations will be triggered, and combined with the existing resource data under the health information service, a cardiovascular health management plan will be recommended to user A, including health services for controlling high blood pressure, recommendations for appropriate medical resources, and relevant health management advice. This method can accurately push personalized health services to users, effectively improving the relevance and practicality of the services.

[0096] Case 1 is particularly applicable to the recommendation of health knowledge and health information services.

[0097] Case 2: Task Management Services

[0098] For task management services, based on user profiles and timely health data related to users, combined with health service resource data and recommended solutions matching this type of health service, the health information that needs to be pushed under this type of health service is determined, which may include:

[0099] For task management services, according to the characteristics of "user portrait and timely health data related to users", combined with health service resource data, corresponding health tasks are formulated, and health tasks are used as health information that needs to be pushed under task management services (which can be Figure 2 The health service resource data here may refer to resource data included in the task management service, such as resource data such as the task content of various health tasks.

[0100] Among them, according to the characteristics of the user portrait and the timely health data related to the user, combined with the health service resource data, the corresponding health tasks are formulated, which may include:

[0101] Obtain target tasks regularly and determine the predicted progress of the corresponding target tasks;

[0102] Comparing the predicted progress with the preset progress of the target task to determine the user's progress evaluation result;

[0103] Based on the characteristics of user portraits and user-related timely health data, combined with health service resource data and user progress evaluation results, corresponding health tasks are formulated.

[0104] like Figure 5 As shown, information push under the task management service can be performed in the following ways (not limited to the following ways):

[0105] Step S310 , periodically obtaining task data (belonging to the target task) from sources such as the user's health monitoring device, application logs, and electronic health records;

[0106] Step S320: Use time series analysis techniques such as ARIMA and LSTM to model the user's historical behavior data and predict future task completion progress (i.e., predict progress);

[0107] Step S330 , by comparing the predicted result with the expected progress (i.e., the preset progress), as well as various evaluation indicators such as task completion rate and behavior consistency, the user's current task stage is determined, and the user's task completion status is evaluated to obtain a progress evaluation result;

[0108] Step S340, combining the user's individual characteristics (characteristics of "user portrait and user-related timely health data"), health service resource data and progress assessment results, dynamically adjusts the recommended content to ensure that the recommended tasks (i.e., corresponding phased health tasks are formulated, which belong to the health information that needs to be pushed under the task management service) are both challenging and can maintain user engagement.

[0109] For example, suppose user B's step count data, heart rate records, and health app logs are regularly collected from their health monitoring device. By applying time series analysis techniques such as ARIMA or LSTM, modeling of user B's step count and heart rate changes over time is performed to predict future trends. The predicted results are compared with user B's health goals to assess their progress towards completing the task. For example, if the prediction indicates that user B will not reach their daily step goal within the set timeframe, the system analyzes their current task completion rate and behavioral consistency, determining that user B may be mid-way through their task. Based on these assessments and combined with existing resource data from task management services, recommendations are dynamically adjusted to push challenging tasks suitable for user B, such as increasing their step goal or adjusting their exercise plan. This ensures that the recommendations maintain user engagement and motivation. This allows for personalized adjustments tailored to user B's actual progress, improving the effectiveness of health management.

[0110] In the second scenario, user portrait characteristics, health service resource data, task progress, and phased needs are combined to formulate and push corresponding tasks and plans, which is particularly suitable for recommending task management services such as health behavior promotion and chronic disease management.

[0111] Case 3: Dynamic information services

[0112] For dynamic information services, based on user profiles and timely health data related to users, combined with health service resource data and recommended solutions matching this type of health service, the health information that needs to be pushed under this type of health service is determined, which may include:

[0113] For dynamic information services, a variety of classic recommendation algorithms are used to determine the similarity between the characteristics of "user portraits and timely health data related to users" and the characteristics of object portraits (object portrait characteristics can be determined in advance), and combined with health service resource data, the health information that needs to be pushed under dynamic information services is determined (which can be determined by Figure 2 The health service resource data here may refer to resource data included in dynamic information services, such as various types of dynamic consultation resource data stored or obtained.

[0114] like Figure 6 As shown, information push under dynamic information services can be carried out in the following ways (not limited to the following ways):

[0115] Step S410: Collect relevant descriptions of the recommended objects, including descriptions that clearly reflect the basic characteristics of the objects, as well as descriptions that implicitly reflect the deep characteristics of the objects, such as user comments, ratings, etc.

[0116] Step S420: Combine natural language processing algorithms and relevant evaluation models to extract all the features of the recommended object from its description to construct a profile of the recommended object (which belongs to the user profile). In this process, technologies such as knowledge graphs can be optionally used to enhance the expression of semantic features.

[0117] Step S430: Convert the extracted features (i.e., features of the user profile and user-related timely health data) into high-dimensional feature vectors through word embeddings, graph embeddings, or other feature embedding techniques. Feature engineering methods such as dimensionality reduction can be optionally applied to optimize the vector representation.

[0118] Step S440: Using similarity measurement methods such as cosine similarity and Euclidean distance, calculate the similarity between the user feature vector and the recommended object feature vector (belonging to the object portrait feature), thereby determining the degree of match between the recommended object and the user's needs;

[0119] Step S450: Based on the similarity matching results and combined with the health service resource data, the most relevant recommendation objects (i.e., health information that needs to be pushed under the dynamic information service) are recommended to the user;

[0120] Step S460, optionally, applies a collaborative filtering recommendation algorithm to recommend to the user the preferences of similar users or content similar to the preference objects (which also belongs to the health information that needs to be pushed under the dynamic information service) to improve the relevance and personalization of the recommendation.

[0121] For example, suppose we need to recommend appropriate medical resources for user C. First, descriptions of relevant medical resources, such as user reviews, are collected. Then, keywords are extracted from the user reviews using a keyword extraction algorithm (such as the TF-IDF algorithm). These keywords include, but are not limited to, specialized fields (e.g., cardiology), geographic locations (e.g., hospitals in a certain city), years of experience (e.g., doctor's work experience), doctor's attitude (e.g., "patient," "kind," etc.), condition descriptions (e.g., "dizziness," "colic," etc.), treatment options (e.g., "surgery," "medication," etc.), treatment outcomes (e.g., "recovery," "improvement," etc.), and patient gratitude (e.g., "medical ethics," "thank you," etc.). Some keywords can be directly used as basic features of the medical resource, while others can be combined with sentiment analysis methods and, based on the SERVQUAL evaluation model, derived from deep features such as the medical resource's degree of empathy, reliability, responsiveness, and assurance. All features of the medical resource (which constitute the object profile) are then converted into high-dimensional feature vectors. Feature dimensionality reduction methods can be used to optimize the vector representation during this process. Then, using metrics such as cosine similarity or Euclidean distance, the similarity between user C's health feature vector and the medical resource feature vector is calculated to determine the degree of match between the resources and the user's needs. Based on these calculation results and combined with existing resource data from dynamic information services, the most relevant medical resources (i.e., health information that needs to be pushed under dynamic information services) are recommended to user C. As a supplement, collaborative filtering recommendation algorithms can be applied to further improve the relevance and personalization of recommendations based on the preferences of other similar users or the content of similar medical resources.

[0122] In case three, a variety of classic recommendation algorithms are integrated. By comparing the similarity between user portrait features and object portrait features, combined with health service resource data, relevant content is recommended to users, such as doctor recommendations, forum information push, etc., which is especially suitable for recommendations of dynamic information services.

[0123] The above-mentioned association rule matching and recommendation module 21 , phased task push and optimization module 22 , and classic recommendation algorithm integration module 23 can be integrated into a health service push module 2 .

[0124] As described above, the health information that needs to be pushed under each type of health service can be determined, and the health information that needs to be pushed under each type of health service can be pushed according to the preset push strategy. Specifically, the health information that needs to be pushed under each type of health service can be used as a push task. The push tasks can be sorted by priority based on the urgency of the user's health status, the priority of the service demand, and the timeliness of the recommended service, and the push tasks can be pushed according to the priority (belonging to the push strategy) to ensure that the most important health services can be pushed to the user in a timely manner (can be determined by Figure 2 The priority queue management module 31 in the process is executed).

[0125] S107: Executing information push optimization operations, including: optimizing health information and push strategies under various health services according to optimization reference factors, and pushing the optimized health information under various health services according to the optimized push strategies.

[0126] In the first embodiment, an information push optimization operation can be performed, thereby optimizing both the push information and the push strategy. The information push optimization operation can include optimizing the health information and push strategy for various health services based on optimization reference factors, and pushing the optimized health information for various health services according to the optimized push strategy.

[0127] In the first embodiment, the following methods (not limited to the following methods) can be used to optimize the push information and push strategy:

[0128] By analyzing the user's historical behavior data and health feedback (which are optimization reference factors), combined with reinforcement learning or related optimization algorithms, the push plan can be optimized and the frequency, time and form of push content can be adjusted (that is, the recommendation strategy can be optimized by Figure 2 The push plan optimization module 32 in the system is executed) to improve the accuracy of health service push and user satisfaction.

[0129] It is possible to collect user feedback after receiving health information push, and also to evaluate the effect of health information push, such as collecting feedback and evaluating the push effect through questionnaire surveys (which can be done by Figure 2 Based on the feedback and / or evaluation results, push strategies and health information under various health services can be continuously improved or optimized. For example, the push priority or push frequency of push information or push services with better feedback and / or evaluation results can be increased.

[0130] The above-mentioned priority queue management module 31 , push plan optimization module 32 and user questionnaire feedback module 33 can be integrated into a push plan management module 3 .

[0131] The above modules can be integrated into a health service push system to execute the health data capsule personal health information push method provided in the first embodiment.

[0132] The first embodiment can achieve the following beneficial effects:

[0133] By acquiring multi-dimensional data from various data sources, pre-processing and extracting features (including deep health feature extraction), we generate detailed user profiles and store them in encrypted form, providing the foundational data for subsequent personalized service recommendations and push notifications. By acquiring target data through multiple channels and dimensions, including but not limited to data storage systems, application logs, and user input, we increase the richness of target data.

[0134] Corresponding preprocessing processes are used for target data with different characteristics. For example, corresponding preprocessing methods, such as data cleaning and format conversion, are used to target the characteristics of structured and unstructured data, combined with machine learning technology to explore deep-seated health characteristics and potential patterns. This method of preprocessing target data with different characteristics using corresponding preprocessing processes can improve data processing effectiveness and efficiency, making the processed data more efficient and more conducive to feature extraction. It can effectively improve data utilization efficiency and also help improve the accuracy and efficiency of health information push.

[0135] We combine topic recognition and statistical learning techniques to extract data features, ensuring the accuracy and comprehensiveness of the generated user profiles. The generated user profile data is securely stored with efficient access and strict security encryption, providing strong support for the precise delivery of personalized health services.

[0136] According to the different types and characteristics of health services, corresponding recommendation schemes are used to push health information. For example, health information services, including but not limited to diet recommendations, health check-ups and preventive knowledge recommendations, mainly provide stable health information and suggestions to support users' daily health management and decision-making. Its focus is on providing reliable health knowledge to help users prevent diseases and maintain health. It adopts an association rule matching scheme, combined with health service resource data, to provide highly relevant suggestions based on the user's health information. It has high efficiency in processing structured health knowledge, including quickly matching the user's health characteristics with corresponding health suggestions. It is especially suitable for personalized diet suggestions, physical examination services and disease prevention knowledge recommendations.

[0137] Task management services, including but not limited to health behavior promotion and chronic disease management, usually involve specific tasks or goals of users. These services focus on helping users develop and implement long-term health plans, and adopt phased task push and optimization solutions. By analyzing the user's health profile characteristics, task progress and phased needs, combined with health service resource data, tasks and plans are dynamically pushed, and recommended content can be dynamically adjusted according to the user's task progress and needs. It is especially suitable for task management services such as health behavior promotion and chronic disease management. It can be adjusted according to the user's actual progress and needs, providing high real-time and personalized services, thereby improving user participation and health management effects.

[0138] Dynamic information services, including but not limited to medical resource matching and forum information push, focus on real-time updated information. They can provide the latest information and resources based on users' immediate needs to adapt to the rapidly changing health environment. Classic recommendation algorithms are applied to them, integrating multiple recommendation algorithms. By calculating the similarity between user profile features and service recipients, combined with health service resource data, dynamic information such as doctor recommendations and forum information push is recommended to users. They are particularly capable of adapting to different user needs, handling complex recommendation scenarios, and improving the relevance and accuracy of recommendations. Their high flexibility allows adjustments based on user behavior data and changes in interests, thereby optimizing the user experience and ensuring that real-time updated information is accurately pushed to users.

[0139] This combination and personalized recommendation method for health information push, which uses corresponding recommendation schemes for different health service types and characteristics, can more clearly identify and meet the needs of different health services, improve the relevance and effectiveness of services, meet personalized service requirements, and effectively improve the accuracy and effectiveness of health information push, thereby improving the effectiveness and efficiency of health information push. Health information push refers to factors such as user portraits and timely health data related to users, which can further improve the accuracy and effectiveness of health information push, thereby improving the effectiveness and efficiency of health information push.

[0140] Moreover, the combination of the above-mentioned recommendation schemes overcomes the possible defects of a single scheme, effectively covers various services such as health knowledge, task management and dynamic information, and provides comprehensive personalized recommendation services by combining multiple technologies such as rule matching, dynamic task push and classic recommendation algorithms.

[0141] By presetting push strategies, we can ensure that the most urgent and important services are pushed first.

[0142] Through information push optimization operations, it is possible to utilize historical data and feedback to timely manage and optimize push information content and strategies, making health information push more adaptive, always and better matching user needs and preferences, and ensuring push timeliness and personalization, thereby improving the effectiveness and efficiency of health information push, and improving the accuracy, effectiveness, and user satisfaction of services. In particular, based on user feedback and changes in health status, combined with optimization algorithms and reinforcement learning, push strategies (including the frequency, time, and form of push content) can be optimized, and the content of push information can be optimized, further improving the effectiveness and adaptiveness of recommendations, so that push content can better match user needs and preferences, ensuring the timeliness and effectiveness of push content.

[0143] By adopting secure storage methods, encryption algorithms and health data capsules, the security and privacy protection of various stored data can be improved, effectively preventing data leakage and unauthorized access.

[0144] As can be seen from the above, Example 1 can accurately capture and analyze the health needs of users, thereby realizing the intelligent push of personalized health services.

[0145] like Figure 7 As shown, the second embodiment of this specification provides a health data capsule personal health information push device corresponding to the method described in the first embodiment, the device comprising:

[0146] The data processing module 202 is used to obtain target data related to the user and pre-process the target data through a pre-processing process that matches the characteristics of the target data;

[0147] The user portrait module 204 is used to extract data features of the pre-processed data using a machine learning algorithm, and generate a user portrait based on the user's health data capsule personal data and the data features;

[0148] The information push module 206 is configured to execute an information push operation based on various recommendation reference factors, wherein the recommendation reference factors include the user profile and timely health data related to the user. Executing the information push operation includes: for any type of health service, determining a recommended solution matching the health service; determining health information to be pushed for the health service based on the user profile and timely health data related to the user, combined with the recommended solution matching the health service resource data and the health service, and pushing the health information according to a preset push strategy;

[0149] The push optimization module 208 is used to perform information push optimization operations, including: optimizing the health information and push strategies under various health services according to optimization reference factors, and pushing the optimized health information under various health services according to the optimized push strategies.

[0150] Optionally, preprocessing the target data using a preprocessing process that matches the characteristics of the target data includes:

[0151] According to the structural or non-structural characteristics of the target data, a matching preprocessing process is determined, and the target data is preprocessed through the determined preprocessing process.

[0152] Optionally, the health service includes a health information service, and the recommendation scheme corresponding to the health information service is an association rule matching scheme.

[0153] Optionally, based on the user profile and timely health data related to the user, combined with the health service resource data and the recommended solution matching the health service of this type, the health information that needs to be pushed under the health service of this type is determined, including:

[0154] For health information services, the health information that needs to be pushed under the health information service is determined based on the matching of the characteristics of the user portrait and the user's timely health data with the rule information matching library, combined with the health service resource data.

[0155] Optionally, the health service includes a task management service, and the recommendation plan corresponding to the task management service is a phased task recommendation plan.

[0156] Optionally, based on the user profile and timely health data related to the user, combined with the health service resource data and the recommended solution matching the health service of this type, the health information that needs to be pushed under the health service of this type is determined, including:

[0157] For task management services, corresponding health tasks are formulated based on the characteristics of the user portrait and the user's timely health data, combined with health service resource data, and the health tasks are used as health information that needs to be pushed under the task management service.

[0158] Optionally, based on the user profile and the characteristics of the user's timely health data, combined with health service resource data, corresponding health tasks are formulated, including:

[0159] Obtain target tasks regularly and determine the predicted progress of the corresponding target tasks;

[0160] Comparing the predicted progress with the preset progress of the target task to determine the user's progress evaluation result;

[0161] Based on the characteristics of user portraits and user-related timely health data, combined with health service resource data and user progress evaluation results, corresponding health tasks are formulated.

[0162] Optionally, the health service includes a dynamic information service, and the recommendation scheme corresponding to the dynamic information service is a classic recommendation algorithm scheme.

[0163] Optionally, based on the user profile and timely health data related to the user, combined with the health service resource data and the recommended solution matching the health service of this type, the health information that needs to be pushed under the health service of this type is determined, including:

[0164] For dynamic information services, a variety of classic recommendation algorithms are used to determine the similarity between the characteristics of the user portrait and user-related timely health data and the object portrait characteristics, and combined with health service resource data to determine the health information that needs to be pushed under dynamic information services.

[0165] The second embodiment can achieve the same technical effect as the first embodiment, and the embodiments can be used in combination.

[0166] The foregoing is merely an embodiment of the present invention and is not intended to limit the present application. For those skilled in the art, various modifications and variations may be made to the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included within the scope of the claims of the present application.

Claims

1. A method for pushing personal health information of a health data capsule, characterized in that: The method comprises: Acquire user-related target data, and preprocess the target data using a preprocessing process that matches the characteristics of the target data; Utilize machine learning algorithms to extract data features from pre-processed data, and generate user profiles based on the user's health data capsule personal data and the data features; Executing an information push operation based on each recommendation reference factor; wherein the recommendation reference factor includes the user profile and timely health data related to the user; executing the information push operation includes: for any type of health service, determining a recommended solution that matches the health service; determining the health information that needs to be pushed under the health service based on the user profile and timely health data related to the user, combined with the health service resource data and the recommended solution that matches the health service, and pushing the health information according to a preset push strategy; Perform information push optimization operations, including: optimizing health information and push strategies under various health services based on optimization reference factors, and pushing optimized health information under various health services according to the optimized push strategies.

2. The method according to claim 1, wherein Preprocessing the target data through a preprocessing process that matches the characteristics of the target data includes: According to the structural or non-structural characteristics of the target data, a matching preprocessing process is determined, and the target data is preprocessed through the determined preprocessing process.

3. The method according to claim 1, wherein The health service includes a health information service, and the recommendation scheme corresponding to the health information service is an association rule matching scheme.

4. The method according to claim 3, wherein Based on the user profile and the user's timely health data, combined with the health service resource data and the recommended solution matching this type of health service, the health information that needs to be pushed under this type of health service is determined, including: For health information services, the health information that needs to be pushed under the health information service is determined based on the matching of the characteristics of the user portrait and the user's timely health data with the rule information matching library, combined with the health service resource data.

5. The method according to claim 1, wherein The health service includes a task management service, and the recommendation scheme corresponding to the task management service is a phased task recommendation scheme.

6. The method according to claim 5, wherein Based on the user profile and the user's timely health data, combined with the health service resource data and the recommended solution matching this type of health service, the health information that needs to be pushed under this type of health service is determined, including: For task management services, corresponding health tasks are formulated based on the characteristics of the user portrait and the user's timely health data, combined with health service resource data, and the health tasks are used as health information that needs to be pushed under the task management service.

7. The method according to claim 6, wherein Based on the characteristics of the user profile and the user's timely health data, combined with health service resource data, corresponding health tasks are formulated, including: Obtain target tasks regularly and determine the predicted progress of the corresponding target tasks; Comparing the predicted progress with the preset progress of the target task to determine the user's progress evaluation result; Based on the characteristics of user portraits and user-related timely health data, combined with health service resource data and user progress evaluation results, corresponding health tasks are formulated.

8. The method according to claim 1, wherein The health service includes a dynamic information service, and the recommendation scheme corresponding to the dynamic information service is a classic recommendation algorithm scheme.

9. The method according to claim 7, wherein Based on the user profile and the user's timely health data, combined with the health service resource data and the recommended solution matching this type of health service, the health information that needs to be pushed under this type of health service is determined, including: For dynamic information services, a variety of classic recommendation algorithms are used to determine the similarity between the characteristics of the user portrait and user-related timely health data and the object portrait characteristics, and combined with health service resource data to determine the health information that needs to be pushed under dynamic information services.

10. A health data capsule personal health information push device, characterized in that: The device comprises: A data processing module is used to obtain target data related to the user and pre-process the target data through a pre-processing process that matches the characteristics of the target data; A user portrait module is used to extract data features of pre-processed data using a machine learning algorithm, and generate a user portrait based on the user's personal data in the health data capsule and the data features; An information push module is configured to execute an information push operation based on various recommendation reference factors, wherein the recommendation reference factors include the user profile and timely health data related to the user; executing the information push operation includes: determining, for any type of health service, a recommended solution matching that type of health service; determining, based on the user profile and timely health data related to the user, the recommended solution matching that type of health service in combination with the health service resource data, health information that needs to be pushed under that type of health service, and pushing the health information according to a preset push strategy; The push optimization module is used to perform information push optimization operations, including: optimizing the health information and push strategies under various health services based on optimization reference factors, and pushing the optimized health information under various health services according to the optimized push strategies.

Images