User portrait-based scheme generation method and device, equipment, storage medium

Abstract

The embodiment of the application provides a scheme generation method and device based on user portrait, equipment and storage medium, and belongs to the technical field of artificial intelligence. The method comprises the following steps: acquiring multi-source heterogeneous data, wherein the multi-source heterogeneous data comprises insurance data and at least one of medical data and health data, the medical data and the health data are derived from a slave node; performing desensitization processing on the multi-source heterogeneous data to obtain multi-source desensitization data; performing portrait extraction on the multi-source desensitization data to obtain at least two kinds of data: a health risk portrait, a health behavior portrait and a health improvement portrait; performing risk assessment based on multi-dimensional portrait data to obtain risk assessment information; performing scheme prediction based on the risk assessment information to obtain a current adjustment scheme, and sending the current adjustment scheme to the slave node. The application can be applied to the financial technology and health care scene, can break the information island, can mine multi-dimensional user portrait, and can improve the generation efficiency and accuracy of the guarantee scheme generation.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology and is applied to fintech and healthcare scenarios, particularly to a method, apparatus, device, and storage medium for generating solutions based on user profiles. Background Technology

[0002] In the current insurance industry, rigid coverage rules fail to adapt to customers' current health conditions. For example, once the coverage details (such as sum insured, premium, and scope of coverage) are determined during the underwriting stage, the coverage rules remain unchanged throughout the insurance period. Even if a customer achieves significant improvements in their health through medical treatment and health management (such as weight loss, smoking cessation, and meeting chronic disease control targets), they cannot obtain coverage upgrades (such as increased sum insured) or premium discounts. Furthermore, when a customer's health deteriorates, such as with abnormal chronic disease indicators, the insurance company can only passively wait for claims to occur. The inability of insurance companies to promptly obtain information about customers' health and adjust coverage accordingly leads to a vicious cycle of retaining high-risk customers and losing healthy customers, increasing operating costs and reducing customer satisfaction. Therefore, how to generate coverage details in a timely manner and improve the accuracy of coverage content generation has become an urgent problem to be solved. Summary of the Invention

[0003] The main objective of this application is to propose a method, apparatus, device, and storage medium for generating solutions based on user profiles, aiming to solve the technical problems of low efficiency and low accuracy in the generation of security documents in the current technology, and to improve the efficiency and accuracy of security document generation.

[0004] To achieve the above objectives, a first aspect of this application proposes a scheme generation method based on user profiles, wherein the master node is communicatively connected to at least one slave node, and the method includes: Acquire multi-source heterogeneous data; wherein the multi-source heterogeneous data originates from the master node and the slave node; The multi-source heterogeneous data is de-identified to obtain multi-source de-identified data; The multi-source de-identified data is used to extract profiles to obtain multi-dimensional profile data; wherein, the multi-dimensional profile data includes at least two of the following: health risk profile, health behavior profile, and health improvement profile; Risk assessment is performed based on the multi-dimensional profile data to obtain risk assessment information; Based on the risk assessment information, a solution prediction is performed to obtain the current adjustment solution, and the current adjustment solution is sent to the slave node.

[0005] To achieve the above objectives, a second aspect of this application proposes a scheme generation apparatus based on user profiles. The apparatus is applied to a master node, which is communicatively connected to at least one slave node. The apparatus includes: A multi-source heterogeneous data acquisition module is used to acquire multi-source heterogeneous data; wherein, the multi-source heterogeneous data originates from the master node and the slave node; The data desensitization module is used to desensitize the multi-source heterogeneous data to obtain multi-source desensitized data; The multi-dimensional profile extraction module is used to extract profiles from the multi-source de-identified data to obtain multi-dimensional profile data; wherein, the multi-dimensional profile data includes at least two of the following: health risk profile, health behavior profile, and health improvement profile; The risk assessment module is used to perform risk assessment based on the multidimensional profile data to obtain risk assessment information; The scheme generation module is used to predict the scheme based on the risk assessment information, obtain the current adjustment scheme, and send the current adjustment scheme to the slave node.

[0006] To achieve the above objectives, a third aspect of the present application provides a computer device, the computer device including a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method described in the first aspect.

[0007] To achieve the above objectives, a fourth aspect of the present application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described in the first aspect.

[0008] The user profile-based scheme generation method, apparatus, device, and storage medium proposed in this application obtain multi-source heterogeneous data by acquiring local insurance data from the master node and medical and health data from the slave nodes. This multi-source heterogeneous data is then anonymized, and a profile is extracted based on the anonymized data to obtain multi-dimensional profile data including health risk profiles, health behavior profiles, and health improvement profiles. Risk assessment is performed based on this multi-dimensional profile data to obtain risk assessment information. Scheme prediction is then based on this risk assessment information to obtain a current adjustment scheme. This current adjustment scheme includes at least one of a rights allocation scheme, an intervention service scheme, an underwriting conclusion, and a claims conclusion. The current adjustment scheme is promptly sent to the slave nodes. This application can process massive amounts of multi-source heterogeneous data, break down information silos, and mine multi-dimensional user profiles, thereby improving the efficiency and accuracy of insurance scheme generation. Attached Figure Description

[0009] Figure 1 This is a schematic diagram of the application environment for the solution generation method based on user profiles provided in the embodiments of this application; Figure 2 This is a flowchart of the user profile-based scheme generation method provided in the embodiments of this application; Figure 3 yes Figure 2 The flowchart for step 202 in the document; Figure 4 This is another flowchart of the solution generation method based on user profiles provided in the embodiments of this application; Figure 5 yes Figure 2 The flowchart for step 203 in the document; Figure 6 yes Figure 2 The flowchart for step 205 in the document; Figure 7 This is another flowchart of the user profile-based scheme generation method provided in the embodiments of this application; Figure 8 This is a schematic diagram of the structure of the solution generation device based on user profile provided in the embodiments of this application; Figure 9 This is a schematic diagram of the hardware structure of the computer device provided in the embodiments of this application. Detailed Implementation

[0010] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0011] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0012] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0013] First, let's analyze some of the terms used in this application: Artificial intelligence (AI) is a new branch of computer science that studies, develops, and applies theories, methods, technologies, and systems to simulate, extend, and expand human intelligence. It aims to understand the essence of intelligence and produce intelligent machines that can react in a way similar to human intelligence. Research in this field includes robotics, speech recognition, image recognition, natural language processing, and expert systems. AI can simulate the information processes of human consciousness and thought. Furthermore, AI utilizes digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceiving the environment, acquiring knowledge, and using that knowledge to achieve optimal results.

[0014] Natural Language Processing (NLP): NLP uses computers to process, understand, and utilize human language (such as Chinese and English). NLP is a branch of artificial intelligence and an interdisciplinary field of computer science and linguistics, often referred to as computational linguistics. NLP includes syntactic analysis, semantic analysis, and discourse understanding. It is commonly used in machine translation, handwritten and printed character recognition, speech recognition and text-to-speech conversion, intent recognition, information extraction and filtering, text classification and clustering, sentiment analysis, and opinion mining. It involves data mining, machine learning, knowledge acquisition, knowledge engineering, artificial intelligence research, and linguistic research related to language computation.

[0015] The insurance industry currently suffers from a common pain point: a disconnect between static coverage and dynamic health needs. This manifests in several ways: Fixed coverage rules fail to adapt to changes in a customer's health. For example, once the coverage details (such as sum insured, premium, and scope of coverage) are determined during the underwriting stage, the rules remain unchanged throughout the insurance period. Even if a customer achieves significant improvements in their health through health management (such as weight loss, smoking cessation, and meeting chronic disease control targets), they cannot obtain coverage upgrades (such as increased sum insured) or premium discounts. When a customer's health deteriorates, such as with abnormal chronic disease indicators, the insurance company can only passively wait for a claim to be paid. Furthermore, in practical applications, the health data obtained by insurance companies is often limited to medical examination reports and past medical history questionnaires at the time of application. It is difficult to integrate dynamic data after underwriting, leading to fragmented health data. For example, data such as heart rate and blood sugar from wearable devices, follow-up records from partner hospitals, and exercise and diet tracking data from health management platforms result in inaccurate and outdated customer profiles, leading to static and one-sided health profiles. In addition, insurance companies currently face problems such as passive risk control and insufficient customer loyalty. Specifically, due to the lack of a positive cycle of health improvement and protection incentives, healthy customers are easily lost because they cannot obtain additional value; high-risk customers, due to the lack of targeted health interventions, experience worsening conditions leading to increased claims rates. Insurance companies are caught in a vicious cycle of retaining high-risk customers and losing healthy customers, which increases operating costs and reduces customer satisfaction.

[0016] Based on this, embodiments of this application provide a method, apparatus, device, and storage medium for generating solutions based on user profiles, aiming to improve the efficiency and accuracy of generating protection documents and provide users with at least one of the following in a timely manner: rights allocation plan, intervention service plan, underwriting conclusion, and claims conclusion.

[0017] The user profile-based scheme generation method, apparatus, device, and storage medium provided in this application are specifically described through the following embodiments. First, the user profile-based scheme generation method in this application embodiment is described.

[0018] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0019] Foundational technologies for artificial intelligence generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies mainly encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0020] The user profile-based solution generation method provided in this application relates to the field of artificial intelligence technology. This user profile-based solution generation method can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, etc.; the server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application implementing the user profile-based solution generation method, but is not limited to the above forms.

[0021] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0022] It should be noted that in all specific embodiments of this application, when processing data related to user identity or characteristics, such as user audio, user voice, user behavior, user historical data, and user attribute information, user permission or consent is obtained first. Furthermore, the collection, use, and processing of this data comply with relevant laws, regulations, and standards. In addition, when embodiments of this application require access to sensitive personal information of users, separate permission or consent from the user is obtained through pop-ups or redirects to confirmation pages. Only after obtaining the user's separate permission or consent is the necessary user-related data required for the proper functioning of these embodiments obtained.

[0023] The user profile-based scheme generation method provided in this application can be applied to, for example, Figure 1 In this application environment, communication connections are established between the master node and slave nodes. In one application scenario, the master node can be an insurance company, and the slave nodes can include medical devices, equipment, etc. Slave nodes can provide medical and health data, while the master node can acquire multi-source heterogeneous data, perform anonymization processing on this data, extract profiles based on the anonymized data, conduct risk assessments based on the multi-dimensional profiles, obtain risk assessment information, predict solutions based on the risk assessment information, derive the current adjustment plan, and send the current adjustment plan to the slave nodes. Slave nodes can be, but are not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices. The server side can be implemented using a standalone server or a server cluster consisting of multiple servers. The following detailed description of specific embodiments further illustrates this application.

[0024] Figure 2 This is an optional flowchart of the scheme generation method based on user profiles provided in the embodiments of this application. Figure 2 The method may include, but is not limited to, steps 201 to 205.

[0025] Step 201: Obtain multi-source heterogeneous data; wherein, the multi-source heterogeneous data comes from the master node and the slave node; Step 202: De-identify the multi-source heterogeneous data to obtain multi-source de-identified data; Step 203: Extract profiles from the multi-source de-identified data to obtain multi-dimensional profile data; wherein, the multi-dimensional profile data includes at least two of the following: health risk profile, health behavior profile, and health improvement profile; Step 204: Conduct a risk assessment based on the multi-dimensional profile data to obtain risk assessment information; Step 205: Based on the risk assessment information, predict the solution to obtain the current adjustment solution, and send the current adjustment solution to the slave node.

[0026] Steps 201 to 205 of this application embodiment involve obtaining multi-source heterogeneous data by acquiring local insurance data from the master node and medical and health data from the slave nodes. This multi-source heterogeneous data is then anonymized, and a profile is extracted based on the anonymized data, resulting in multi-dimensional profile data including health risk profiles, health behavior profiles, and health improvement profiles. Risk assessment is then performed based on this multi-dimensional profile data to obtain risk assessment information. Based on this risk assessment information, a solution prediction is made to obtain a current adjustment plan. This current adjustment plan includes at least one of the following: a rights allocation plan, an intervention service plan, an underwriting conclusion, and a claims conclusion. The current adjustment plan is then promptly sent to the slave nodes. This application can process massive amounts of multi-source heterogeneous data, break down information silos, and mine multi-dimensional user profiles, thereby improving the efficiency and accuracy of insurance plan generation.

[0027] In step 201 of some embodiments, the insurance data in the multi-source heterogeneous data can originate from the insurance company's internal business system, such as the insurance company's claims system and underwriting system. The medical data and health data in the multi-source heterogeneous data originate from external data sources, such as medical devices and equipment. The medical device may include, but is not limited to, hospitals, physical examination institutions, and chronic disease management platforms. The medical data may include, but is not limited to, electronic medical records, follow-up visit records, physical examination reports, and chronic disease indicators. Chronic disease indicators may include, for example, blood pressure and blood sugar. The medical data can be structured or unstructured data, and this application embodiment does not limit it. Structured data may include, for example, data such as name, gender, diagnosis code, medication records, and chronic disease indicator values ​​in electronic medical records. Unstructured data may include, for example, data such as CT images, ultrasound reports, and physical examination summary texts. The equipment may include, but is not limited to, smart bracelets. Wearable devices or IoT devices such as blood glucose meters and blood pressure monitors can be used. The device end can also include a user terminal with an application installed. Health data includes device monitoring data and user behavior data. Device monitoring data can include, for example, real-time data uploaded by wearable devices or IoT devices such as smart bracelets, blood glucose meters, and blood pressure monitors, or data uploaded by the application such as exercise plans and check-in records. For example, dynamic data such as heart rate, sleep duration, steps, and physiological indicator trends can be included. In one application scenario, device monitoring data can include, for example, heart rate data collected by a smart bracelet every 5 minutes: resting heart rate 68 beats / min, heart rate during exercise 120 beats / min, and sleep data synchronized daily: 3.5 hours of deep sleep, 4 hours of light sleep, and 2 awakenings. The smart blood glucose meter collects fasting and 2-hour postprandial blood glucose data daily: fasting 6.8 mmol / L, and 2-hour postprandial 8.5 mmol / L. User behavior data can be collected through insurance apps / mini-programs, including customer health management behaviors and lifestyle questionnaires. Health management behaviors include things like smoking cessation check-ins, nutritionist consultations, and completion rates of exercise plans. Lifestyle questionnaires include data such as dietary structure and sleep patterns. In one application scenario, user behavior data could include a smart bracelet collecting 12,000 steps per day, 30 minutes of aerobic exercise, and running as the type of exercise.

[0028] In one application scenario, unstructured medical data can be transformed into structured features using Optical Character Recognition (OCR) technology and Natural Language Processing (NLP). For example, the text "[slight inflammation in the lower lobes of both lungs]" in a CT report can be extracted using NLP to extract the keywords "lower lobes of both lungs" and "inflammation," and converted into standardized feature values. Then, federated learning can be used to perform joint analysis of the image data across institutions without the need to transmit the original image files.

[0029] In one application scenario, tracking code can be deployed on key pages of insurance apps / mini-programs, such as health check-in pages, nutritionist consultation pages, and exercise plan pages. This allows for the collection of basic user behavior data through full tracking, such as page dwell time, click count, and navigation path. For example, a user might enter the smoking cessation check-in page from the homepage, stay for 10 minutes, and complete the check-in. Custom tracking is also possible. Specifically, for core health management behaviors, dedicated tracking points can be set, such as the check-in time and consecutive check-in days for smoking cessation check-ins, the completion rate and actual exercise duration for exercise plans, and the consultation duration and content category for nutritionist consultations. Consultation content categories could include, for example, dietary consultations and chronic disease management consultations.

[0030] In some embodiments, multi-source heterogeneous data is acquired through federated learning technology. The insurance company acts as the model trainer and master node; the medical and device ends act as data holders and slave nodes, retaining the original data on their own servers and not transmitting any original data to the insurance company. The insurance company, as the model trainer, deploys the federated learning master node and distributes encrypted model parameters to each data holder. Each data holder updates the model parameters locally using homomorphic encryption based on its local data, only feeding back the updated encrypted parameters to the master node. The master node aggregates the encrypted parameters from all data holders, completing model iteration and ultimately achieving virtual integration of multi-source data. This application, through federated learning technology, eliminates the need for insurance companies to directly obtain the original data from hospitals and devices; it integrates multi-terminal data simply through model parameter sharing. This breaks down medical data silos, avoids data privacy leaks, and complies with regulatory requirements.

[0031] In some embodiments, step 201 may include, but is not limited to: Initialize the original federated learning model and encrypt the initial model parameters of the original federated learning model to obtain encrypted model parameters; The encryption model parameters and the preset feature alignment standard are sent to the slave node; the preset feature alignment standard is used by the slave node to perform feature alignment on the local data to obtain locally aligned data. Obtain the model update parameters obtained from the node's model training based on encrypted model parameters and local alignment data to obtain medical and health data.

[0032] In some embodiments, step 201 may also include, but is not limited to: The original federated learning model is updated based on the model update parameters to obtain the target federated learning model, which is then sent to the slave node.

[0033] In one application scenario, step 201 may include, but is not limited to: Step 2011, Node Deployment; Specifically, this may include: deploying federated learning nodes at the insurance company, hospital, and device ends respectively, with the insurance company as the master node and the medical end and device end as slave nodes; Step 2022, Node Identity Authentication; specifically, this may include: clarifying the permissions of each node through digital certificates to ensure the security of node access. For example, the hospital, as a slave node, can only upload model parameters related to medical data and cannot access data from other nodes.

[0034] Step 2023, data preprocessing; specifically, it may include: each slave node, such as the hospital end and the device end, preprocessing the local data, where the local data includes medical data and health data, such as data cleaning and missing value filling for medical data and health data; the specific technical details of data cleaning and missing value filling are not limited in this application embodiment; Step 2024, Feature Alignment; specifically, it may include: the master node defining a unified preset feature standard, and each slave node performing feature alignment on the preprocessed local data according to the preset feature standard to obtain locally aligned data. For example, blood pressure indicators from different hospitals are uniformly converted to the [systolic / diastolic] format, and heart rate data from different devices are uniformly converted to the [beats / minute] format to ensure the consistency of model training. Step 2025, Model Initialization and Local Training; specifically, this may include: the master node initializing the original federated learning model and encrypting the initial model parameters of the original federated learning model to obtain encrypted model parameters, and distributing the encrypted model parameters to each slave node; each slave node training the model based on the local aligned data and the encrypted model parameters, calculating the parameter gradients, generating locally updated model update parameters, encrypting the model update parameters to obtain encrypted update parameters, and sending the encrypted update parameters to the master node; thus, the master node does not need to obtain the original data from the slave nodes, but can integrate multi-terminal data only through model parameter sharing, which breaks down data silos, avoids data privacy leaks, and complies with regulatory requirements; Step 2026, parameter aggregation and model iteration; specifically, it may include: the master node using a secure aggregation algorithm (SA) to aggregate the encrypted update parameters of all slave nodes, eliminating the risk of privacy leakage of individual parameters; based on the aggregated parameters, the master node completes model iteration, sends the new model parameters to each slave node again, and repeats steps 2025 and 2026 until the model converges, finally obtaining the target federated learning model.

[0035] Step 2027, Model Application and Data Feedback; specifically, it may include: after the model converges, the master node will deploy the final target federated learning model to the insurance company, so that it can be used to assess customers' health risks, predict chronic diseases, and other scenarios; at the same time, the feedback data generated during the application of the target federated learning model, such as the deviation between the prediction results and the actual health status, will be synchronized to each slave node through federated learning for continuous model optimization.

[0036] Please see Figure 3 In some embodiments, step 202 may include, but is not limited to: Step 301: Perform sensitive classification on the multi-source heterogeneous data to obtain multi-source classification data with sensitive categories; Step 302: Determine the desensitization method based on the sensitive category; wherein the desensitization method includes at least one of the following: static desensitization and dynamic desensitization; Step 303: De-identify the multi-source classification data based on the de-identification method to obtain multi-source de-identified data.

[0037] In steps 301 to 303 of some embodiments, multi-source heterogeneous data is sensitively classified to obtain multi-source classified data with sensitive categories. The sensitive categories can be used to characterize the information type of the multi-source classified data, such as characterizing the multi-source classified data as personal sensitive information. For personal sensitive information, the desensitization method used is static desensitization. Static desensitization can include, but is not limited to, at least one of permanent desensitization, obfuscation, etc. Taking the financial scenario as an example, core sensitive fields such as name, ID number, mobile phone number, and home address in personal sensitive information are permanently desensitized. Taking the medical scenario as an example, information such as medical record number and hospital name in personal sensitive information is obfuscated. For example, the hospital name "XX City First People's Hospital" is replaced with "XX City Tertiary Hospital".

[0038] Sensitive categories can also be used to characterize the use cases of multi-source classification data. For different use cases, dynamic desensitization is adopted. For example, when using data for internal personnel of insurance companies, if the use case is a health assessment scenario, only the desensitized physiological indicator data is displayed, and no personal sensitive information is displayed. If the use case is a claims review scenario, after authorization, some necessary desensitized information can be displayed, such as the date of medical treatment and diagnosis results, but core sensitive fields are still not disclosed.

[0039] In some embodiments, step 202 may further include: generating a de-identification log. Specifically, all de-identification related information is recorded to generate a de-identification log. The de-identification related information may include, but is not limited to, de-identification method, de-identification time, de-identified fields, operator information, etc., for subsequent blockchain evidence storage and traceability.

[0040] In some embodiments, after step 202, the present application may further include: performing on-chain evidence storage processing on the multi-source de-identified data. For details, please refer to... Figure 4 This may include, but is not limited to: Step 401: Convert the format of the multi-source de-identified data through a smart contract to obtain a de-identified data block; wherein, the de-identified data block includes at least a hash value; Step 402: Associate the de-identified data blocks based on hash values ​​to obtain chain-structured data; Step 403: Obtain data collection information for multi-source heterogeneous data, wherein the data collection information includes at least the collection time and data source identifier; Step 404: Perform on-chain evidence storage processing based on chain-structured data and data collection information.

[0041] In steps 401 to 404 of some embodiments, multi-source de-identified data is format-converted through smart contracts into a blockchain-storable format, thereby obtaining de-identified data blocks. These de-identified data blocks include at least a hash value. In some application scenarios, to ensure immutability, hash algorithms and digital signature technology are used. Each de-identified data block's hash value is associated with the hash value of the previous de-identified data block, forming a chain-like data structure. If the data of a node is tampered with, its hash value changes and becomes inconsistent with the hash values ​​of other nodes, immediately triggering an anomaly alert. Furthermore, the entire chain cannot be tampered with by modifying a single node's data, thus ensuring data authenticity. In another application scenario, to achieve traceability, the de-identified data block also includes at least a timestamp and node signature information, allowing traceability of data collection information for multi-source heterogeneous data, such as collection time and data source identifier. This data source identifier can be used to identify the collection source, for example, identifying blood pressure data as coming from a smart blood pressure monitor manufactured by Company N, or identifying medical data as coming from Hospital M. Taking insurance companies, hospitals, and devices as participating nodes in the blockchain as an example, the chain-structured data and data collection information are processed for on-chain storage and uploaded to the consortium blockchain nodes. All nodes (insurance companies, hospitals, and devices) store data synchronously, eliminating the need for a single data center and preventing data loss or tampering.

[0042] In one application scenario, customer C1's fasting blood glucose data of 7.2 mmol / L is collected from a smart blood glucose meter and statically anonymized (e.g., the customer's phone number and device serial number are removed and replaced with a hash identifier). This data is then uploaded to the consortium blockchain via a smart contract. The insurance company, the device itself, and the regulatory agency act as blockchain nodes, synchronously storing this data with a timestamp of 2026-04-20 08:30:00. The data source is verified through node signature information. If customer C1's fasting blood glucose data of 7.2 mmol / L is subsequently tampered with, for example, changed to 6.0 mmol / L, its hash value changes and does not match the hash values ​​of other nodes, immediately triggering an anomaly. The tampering node and the time of tampering can be traced through the blockchain, ensuring the data's immutability. Regulatory agencies can trace the entire process of data collection, anonymization, and storage through the consortium blockchain nodes to verify data compliance.

[0043] In this embodiment, all data is anonymized and stored using blockchain technology to ensure data privacy and security. Furthermore, through federated learning, insurance companies can integrate multi-terminal data without obtaining raw data from hospitals or devices, simply by sharing model parameters. This breaks down medical data silos, avoids data privacy leaks, and complies with regulatory requirements.

[0044] Please see Figure 5 In some embodiments, step 203 may specifically include, but is not limited to: Step 501: Perform verification processing on the multi-source de-identified data to obtain multi-source verification data; Step 502: Clean the multi-source verification data to obtain cleaned multi-source data; Step 503: Extract features from the multi-source cleaned data to obtain the original multidimensional features; Step 504: Perform feature fusion on the original multidimensional features to obtain multidimensional portrait data.

[0045] In steps 501 and 502 of some embodiments, due to potential data distortion and format errors during the desensitization process, verification and cleaning processes are performed. This can employ a dual mechanism of hash verification and logical verification. Hash verification compares the hash values ​​of data before and after desensitization to ensure that core data characteristics, such as physiological indicator values ​​and behavioral record sequences, have not been tampered with during the desensitization process. Logical verification checks for logical inconsistencies in the desensitized data, such as discrepancies between desensitized blood glucose data and historical trends, or conflicts between behavioral check-in records and timestamps. Abnormal data is marked, and corrections are made using interpolation or mean replacement methods. Cleaning processes may include: removing redundant data, standardizing data formats, and filling missing data. Removing redundant data may include removing duplicate uploaded physiological indicators and invalid check-in records. Standardizing data formats may include converting heart rate data uploaded from different devices to [beats / minute] and timestamps to [YYYY-MM-DDHH:MM:SS] format. Filling missing data can be done by predicting missing values ​​based on historical data after desensitization and filling them in to ensure data integrity.

[0046] In step 503 of some embodiments, different feature extraction methods are used for different types of data in the multi-source cleaned data. For example, for data such as blood pressure and blood sugar, key numerical features and classification features can be extracted through feature normalization and encoding transformation methods, such as one-hot encoding and label encoding methods. For example, the blood sugar value of 7.2 mmol / L is normalized into a feature vector in the [0,1] interval. For data such as questionnaire scale scores and attendance records, trend features and behavioral features can be extracted through time-series feature extraction algorithms, such as LSTM. For example, features such as weekly exercise frequency and continuous attendance duration can be extracted from exercise attendance records for 6 consecutive months. For data such as image features and consultation text, image features can be extracted through convolutional neural networks (CNN) and text semantic features can be extracted through NLP. For example, the size of the inflammatory area can be extracted from CT images, and semantic features such as dietary preferences and chronic disease management needs can be extracted from nutritionist consultation texts.

[0047] In step 504 of some embodiments, an attention mechanism is employed to achieve deep fusion of multimodal features. This mechanism automatically assigns feature weights, prioritizing core features that significantly impact the health profile while downplaying secondary features. Core features include, for example, chronic disease indicators and imaging features related to serious illnesses, while secondary features include, for example, occasional records of interrupted physical activity. During feature fusion, feature alignment and feature fusion matrix operations convert feature vectors from different modalities into a unified-dimensional fused feature vector, resulting in multidimensional profile data. This multidimensional profile data can represent dimensions such as health risk, health behavior, and health improvement. These dimensions are interconnected and mutually validated, ensuring the consistency of the profile.

[0048] In step 204 of some embodiments, quantifiable judgment thresholds can be preset based on multidimensional profile data to form standardized judgment rules, thereby eliminating the need for manual intervention and adjustment. Furthermore, the standardized judgment rules can be iteratively optimized based on historical data. Specifically, quantifiable judgment thresholds can include reduced health risk, compliance with health behavior standards, etc. Reduced health risk can include a risk score that decreases by ≥15% from the baseline value, or chronic disease indicators that change from abnormal to normal and remain within the target range. Compliance with health behavior standards can include an exercise check-in rate of ≥80%, normal chronic disease indicators for 3 consecutive months, and a health management behavior completion rate of ≥90%, etc.

[0049] In this embodiment, by reading the latest multidimensional profile data in real time, core features such as the change in risk score in the health risk dimension, the duration of achieving chronic disease indicators, the completion rate of check-in in the health behavior dimension, and the persistence of behavior are extracted and automatically compared with the judgment threshold to obtain risk assessment information.

[0050] In some embodiments, the risk assessment information includes health improvement information, and the current adjustment plan includes a rights allocation plan. Please refer to [link / reference]. Figure 6 In some embodiments, step 205 may specifically include, but is not limited to: Step 601: Evaluate the improvement based on the health improvement information to obtain the improvement status; Step 602: Classify the improvements based on their effectiveness to obtain improvement categories; Step 603: Generate a rights allocation plan based on the improvement category; Step 604: Push the rights and interests allocation plan to the slave nodes.

[0051] In steps 601 to 604 of some embodiments, the improvement status includes the rate of decrease, the achievement of targets, and detailed improvement information. The improvement status is determined based on the health improvement information. For example, a decrease of 15%-25% indicates a mild improvement, a decrease of 26%-40% indicates a moderate improvement, a decrease of 41%-50% indicates a moderate to severe improvement, and a decrease of ≥51% indicates a severe improvement. Achieving targets for 3 consecutive months is considered basic achievement, achieving targets for 6 consecutive months or more is considered stable achievement, and continuous check-ins for ≥90 days are considered stable achievement. Detailed improvement information includes improvements in blood glucose indicators, achievement of hypertension control targets, and standardized exercise behavior, thereby obtaining a standardized quantitative report of improvement information, which serves as the core basis for the allocation of benefits. In some embodiments, the benefits allocation scheme includes upgraded coverage and premium discounts. In one application scenario, differentiated benefit tiers can be set based on the degree of health improvement. For example, a 15%-25% reduction in health risk with satisfactory behavior results in a 3%-5% premium reduction in the following year; a 26%-40% reduction in health risk with satisfactory behavior results in a 6%-8% premium reduction in the following year; a ≥41% reduction in health risk with satisfactory behavior results in a 5%-10% increase in critical illness coverage; and a ≥50% reduction in health risk with satisfactory behavior results in an 11%-20% increase in critical illness coverage or a 9%-10% premium reduction. In some application scenarios, benefits that customers have not yet enjoyed can be prioritized to avoid duplicate redemption. For example, if a customer has already enjoyed a premium discount in the previous year, an increase in coverage can be prioritized this time.

[0052] In other embodiments, the risk assessment information also includes health deterioration information, the current adjustment plan also includes an intervention service plan, and step 205 in some embodiments may also include, but is not limited to: Intervention service plans are generated based on health deterioration information and then pushed to slave nodes.

[0053] In some embodiments, intervention service programs include free nutritionist guidance, free specialist doctor consultations, chronic disease management courses, customized diet plans, and customized exercise plans. In one application scenario, if risk assessment information indicates an increased health risk, such as persistently high blood sugar, the coverage is not directly reduced. Instead, targeted health intervention service programs, such as free nutritionist guidance and chronic disease management courses, are first pushed. If several prognostic health indicators still do not improve, the coverage scope is then optimized, such as increasing reimbursement for the diagnosis and treatment of chronic disease complications, to avoid damage to the customer's insurance rights.

[0054] In some embodiments, the current adjustment scheme further includes at least one underwriting conclusion or claim conclusion. After step 203 in some embodiments, the user profile-based scheme generation method further includes generating at least one underwriting conclusion or claim conclusion based on risk assessment information. In practical applications, when a customer renews their policy, an underwriting conclusion is generated based on the latest multi-dimensional profile data, eliminating the need to resubmit a medical examination report, thus achieving seamless renewal and personalized protection.

[0055] Following step 205 in some embodiments, the user profile-based scheme generation method further includes: profile updating; specifically, please refer to Figure [Figure Number]. Figure 7 This may include, but is not limited to: Step 701: Obtain post-intervention health data based on a preset assessment period; Step 702: Based on risk assessment information and post-intervention health data, conduct an intervention assessment to obtain intervention effect data; Step 703: Update the multidimensional profile data based on the intervention effect data.

[0056] In steps 701 to 703 of some embodiments, the preset evaluation period can be set based on actual needs, such as one month after intervention or three months after intervention. Health data after intervention is obtained, and the risk assessment information before intervention and the health data after intervention are compared. The resulting intervention effect data includes health risk scores, changes in chronic disease indicators, etc. The intervention effect data can also include effect categories, such as effective, ineffective, and worsening. Specifically, a decrease in risk score of ≥5% and indicators tending to normalize can be considered effective, no change in risk score and indicators still exceeding the standard can be considered ineffective, and a continued increase in risk score can be considered worsening. The system updates the multidimensional profile data based on the intervention effect data. Steps 204 and 205 can then be executed again based on the updated multidimensional profile data to update the risk assessment information. For example, if the intervention effect data is ineffective, the coverage optimization is triggered. Based on the customer's chronic disease type and the reason for the increased risk, a corresponding coverage optimization plan is matched. For example, for diabetic customers, reimbursement for the treatment of diabetic complications is added, and for hypertensive customers, reimbursement for the treatment of stroke is added. If the intervention effect data is effective, intervention services are continuously pushed until health indicators return to normal. If several pre-effect data indicate deterioration, intervention services are upgraded, such as increasing the frequency of follow-up visits by specialist doctors, while triggering coverage optimization to prevent damage to the customer's protection rights.

[0057] The closed-loop linkage system for health intervention and coverage adjustment in this application embodiment is not limited to coverage adjustment. It also reduces claims risk through a closed loop of intervention, evaluation, and adjustment. For high-risk customers, it automatically matches medical resources in cooperation with insurance companies, such as online consultations with family doctors and home visits by chronic disease managers. The service fees are borne by the insurance special fund. After the customer completes the intervention service, the intervention effect is evaluated and incorporated into the health profile as the basis for subsequent coverage adjustments, forming a positive cycle of intervention service, health improvement, coverage upgrade, and customer retention.

[0058] The user profile-based solution generation method provided in this application obtains local insurance data from the master node, medical data from the medical end, and health data from the device end. This health data includes device monitoring data and user behavior data, resulting in massive amounts of multi-source heterogeneous data. The method then performs static and dynamic desensitization on this multi-source heterogeneous data, and extracts profiles based on the desensitized data. This yields multi-dimensional profile data including health risk profiles, health behavior profiles, and health improvement profiles. Risk assessment is then performed based on this multi-dimensional profile data to obtain risk assessment information. Based on this risk assessment information, a solution prediction is made to obtain a current adjustment plan. This current adjustment plan includes at least one of the following: a rights allocation plan, an intervention service plan, an underwriting conclusion, and a claims conclusion. The current adjustment plan is promptly sent to slave nodes and pushed to users, enabling them to promptly obtain information about their health status, underwriting results, and claims results. This application can process massive amounts of multi-source heterogeneous data, break down information silos, and mine multi-dimensional user profiles, thereby improving the efficiency and accuracy of solution generation.

[0059] This application embodiment improves the health status of high-risk customers in advance through health intervention, which can reduce the incidence of high-cost claims such as complications of chronic diseases and critical illnesses, thereby reducing claims risk and operating costs. The implementation of this application can also realize automated underwriting and renewal, thereby reducing the workload of manual underwriting, shortening the underwriting cycle, and improving operational efficiency.

[0060] This application's embodiments, by employing incentives to enhance coverage when health improves, can retain more healthy customers, thereby avoiding the adverse selection of losing healthy customers and retaining high-risk customers. At the same time, the addition of health intervention services can increase customers' willingness to pay, promote the upgrading of insurance products from basic coverage to coverage and services, and increase premium income sources.

[0061] In this embodiment, healthy customers can obtain increased coverage and premium discounts through their own health management, avoiding the unfairness of healthy and unhealthy customers enjoying the same protection. High-risk customers can receive free health intervention services instead of directly losing coverage, ensuring their rights are not compromised and increasing customer trust in insurance. Furthermore, customers can enjoy services such as family doctor and nutritionist guidance provided by the insurance without additional fees. In particular, patients with chronic diseases can reduce the risk of their condition worsening through real-time monitoring and intervention by the system, thereby enjoying health management through insurance and improving their quality of life.

[0062] This application embodiment makes policy renewal more convenient and provides a better experience. There's no need to repeatedly submit medical reports or undergo manual underwriting during renewal. This application embodiment can automatically complete underwriting and coverage adjustments based on the latest multi-dimensional profile data, achieving one-click renewal and personalized coverage. This avoids the pain points of cumbersome processes and long waiting times in traditional policy renewal, thus improving the customer experience.

[0063] In the medical context, this application's embodiments can promote prevention as a priority and optimize resource allocation. By incentivizing customers to participate in health management and accept preventive interventions through insurance, it can reduce the likelihood of minor illnesses developing into major ones, alleviate the pressure on hospital emergency and inpatient departments, shift medical resources from treatment to prevention, and improve the efficiency of medical resource utilization.

[0064] Please see Figure 8 This application also provides a solution generation apparatus based on user profiles, which can implement the above-described solution generation method based on user profiles. The apparatus includes: The multi-source heterogeneous data acquisition module is used to acquire multi-source heterogeneous data; the multi-source heterogeneous data comes from the master node and the slave node; The data desensitization module is used to desensitize multi-source heterogeneous data to obtain multi-source desensitized data; The multidimensional profile extraction module is used to extract profiles from multi-source de-identified data to obtain multidimensional profile data; among which, the multidimensional profile data includes at least two of the following: health risk profile, health behavior profile, and health improvement profile; The risk assessment module is used to perform risk assessments based on multi-dimensional profile data and obtain risk assessment information. The scheme generation module is used to predict schemes based on risk assessment information, obtain the current adjustment scheme, and send the current adjustment scheme to the slave node.

[0065] In some embodiments, the data desensitization module can specifically be used to implement: Sensitive classification is performed on multi-source heterogeneous data to obtain multi-source classification data with sensitive categories; The desensitization method is determined based on the sensitive category; the desensitization method includes at least one of the following: static desensitization and dynamic desensitization; The multi-source classification data is anonymized using an anonymization method to obtain anonymized multi-source data. Specifically, the data anonymization module can be used to implement steps 301 to 303 above, which will not be elaborated here.

[0066] In some embodiments, the device further includes an on-chain evidence storage module, which can be specifically used to implement: The data is processed through a smart contract to convert the format of multi-source de-identified data, resulting in de-identified data blocks; each de-identified data block includes at least a hash value. By associating de-identified data blocks based on hash values, a chain-structured data structure is obtained; Acquire data collection information from multi-source heterogeneous data, including at least the collection time and data source identifier; On-chain evidence storage is performed based on chain-structured data and data collection information.

[0067] Specifically, the on-chain evidence storage module can be used to implement steps 401 to 404 above, which will not be elaborated here.

[0068] In some embodiments, the multi-dimensional image extraction module can be specifically used to implement: The multi-source de-identified data is validated to obtain multi-source validation data. The multi-source verification data is cleaned to obtain multi-source cleaned data. Feature extraction is performed on multi-source cleaned data to obtain the original multidimensional features; Feature fusion is performed on the original multidimensional features to obtain multidimensional profile data.

[0069] Specifically, the multi-dimensional image extraction module can be used to implement steps 501 to 504 above, which will not be described in detail here.

[0070] In some embodiments, the solution generation module can specifically be used to implement: An assessment is conducted based on health improvement information to determine the extent of improvement. Based on the degree of improvement, improvement categories are obtained; Generate a rights allocation plan based on improvement categories; The rights and interests allocation plan is pushed to the slave nodes.

[0071] Specifically, the scheme generation module can be used to implement steps 601 to 604 above, which will not be described in detail here.

[0072] In some embodiments, the device further includes an image update module, which can specifically be used to implement: Obtain post-intervention health data based on a pre-set assessment cycle; Intervention effectiveness data are obtained by evaluating the intervention based on risk assessment information and post-intervention health data. The multidimensional profile data is updated based on the intervention effect data.

[0073] Specifically, the image update module can be used to implement steps 701 to 703 above, which will not be described in detail here.

[0074] The specific implementation of the solution generation device based on user profiles is basically the same as the specific implementation of the solution generation method based on user profiles described above, and will not be repeated here.

[0075] This application also provides a computer device, which can be any smart terminal including tablet computers, in-vehicle computers, etc. The computer device includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement: Acquire multi-source heterogeneous data; where the multi-source heterogeneous data comes from master nodes and slave nodes; De-identification processing is performed on multi-source heterogeneous data to obtain multi-source de-identified data; Profiling is performed on multi-source de-identified data to obtain multi-dimensional profile data; among which, multi-dimensional profile data includes at least two of the following: health risk profile, health behavior profile, and health improvement profile; Risk assessment is conducted based on multi-dimensional profile data to obtain risk assessment information; Based on risk assessment information, a solution is predicted to obtain the current adjustment solution, and the current adjustment solution is sent to the slave node.

[0076] Please see Figure 9 , Figure 9 The hardware structure of a computer device according to another embodiment is illustrated. The computer device includes: The processor 901 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application. The memory 902 can be implemented as a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 902 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 902 and is called and executed by the processor 901 to execute the user profile-based solution generation method of the embodiments of this application. The input / output interface 903 is used to implement information input and output; The communication interface 904 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). Bus 905 transmits information between various components of the device (e.g., processor 901, memory 902, input / output interface 903, and communication interface 904); The processor 901, memory 902, input / output interface 903, and communication interface 904 are connected to each other within the device via bus 905.

[0077] This application embodiment also provides a storage medium, which is a computer-readable storage medium, storing a computer program that is implemented when executed by a processor. Acquire multi-source heterogeneous data; among which, multi-source heterogeneous data includes insurance data, which comes from master nodes and slave nodes; De-identification processing is performed on multi-source heterogeneous data to obtain multi-source de-identified data; Profiling is performed on multi-source de-identified data to obtain multi-dimensional profile data; among which, multi-dimensional profile data includes at least two of the following: health risk profile, health behavior profile, and health improvement profile; Risk assessment is conducted based on multi-dimensional profile data to obtain risk assessment information; Based on risk assessment information, a solution is predicted to obtain the current adjustment solution, and the current adjustment solution is sent to the slave node.

[0078] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0079] The user profile-based solution generation method, apparatus, device, and storage medium provided in this application embodiment obtain massive amounts of multi-source heterogeneous data by acquiring local insurance data from the master node, medical data from the medical end, and health data from the device end. This health data includes device monitoring data and user behavior data. The multi-source heterogeneous data undergoes static and dynamic desensitization processing, and profile extraction is performed based on the desensitized multi-source data to obtain multi-dimensional profile data including health risk profiles, health behavior profiles, and health improvement profiles. Risk assessment is then performed based on this multi-dimensional profile data to obtain risk assessment information. Solution prediction is then performed based on this risk assessment information to obtain a current adjustment plan. This current adjustment plan includes at least one of the following: a rights allocation plan, an intervention service plan, an underwriting conclusion, and a claims conclusion. The current adjustment plan is promptly sent to slave nodes and pushed to users, enabling users to promptly obtain information about their health status, underwriting results, and claims results. This application embodiment can process massive amounts of multi-source heterogeneous data, break down information silos, and mine multi-dimensional user profiles, thereby improving the efficiency and accuracy of solution generation.

[0080] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0081] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0082] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0083] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0084] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0085] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0086] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0087] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0088] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0089] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0090] It should be noted that any AI models, software tools, or components not belonging to this company appearing in the embodiments of this application are merely illustrative examples and do not represent actual use. All user personal information involved in the embodiments of this application has been authorized (with the knowledge and consent) by the relevant parties or has been fully authorized by all parties, and the executing entity may obtain it through various legal and compliant means. The collection, storage, use, processing, transmission, provision, and disclosure of the information, data, and signals involved all comply with relevant laws and regulations and do not violate public order and good morals.

[0091] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A solution generation method based on user profiles, characterized in that, The method is applied to a master node, which is communicatively connected to at least one slave node. The method includes: Acquire multi-source heterogeneous data; wherein the multi-source heterogeneous data originates from the master node and the slave node; The multi-source heterogeneous data is de-identified to obtain multi-source de-identified data; The multi-source de-identified data is used to extract profiles to obtain multi-dimensional profile data; wherein, the multi-dimensional profile data includes at least two of the following: health risk profile, health behavior profile, and health improvement profile; Risk assessment is performed based on the multi-dimensional profile data to obtain risk assessment information; Based on the risk assessment information, a solution prediction is performed to obtain the current adjustment solution, and the current adjustment solution is sent to the slave node.

2. The method according to claim 1, characterized in that, After performing desensitization processing on the multi-source heterogeneous data to obtain multi-source desensitized data, the method further includes: The multi-source de-identified data is converted into a format using a smart contract to obtain a de-identified data block; wherein the de-identified data block includes at least a hash value; The de-identified data blocks are associated based on the hash values ​​to obtain chain-structured data; Acquire data collection information of the multi-source heterogeneous data, wherein the data collection information includes at least the collection time and the data source identifier; On-chain evidence storage is performed based on the chain-structured data and the data collection information.

3. The method according to claim 1, characterized in that, The process of extracting profiles from the multi-source de-identified data to obtain multi-dimensional profile data includes: The multi-source de-identified data is subjected to verification processing to obtain multi-source verification data; The multi-source verification data is cleaned to obtain multi-source cleaned data; Feature extraction is performed on the multi-source cleaned data to obtain the original multidimensional features; The original multidimensional features are fused to obtain the multidimensional portrait data.

4. The method according to claim 1, characterized in that, The multi-source heterogeneous data includes medical data and health data. The acquisition of multi-source heterogeneous data includes: Initialize the original federated learning model and encrypt the initial model parameters of the original federated learning model to obtain encrypted model parameters; The encryption model parameters and the preset feature alignment standard are sent to the slave node; wherein, the preset feature alignment standard is used by the slave node to perform feature alignment on local data to obtain locally aligned data; The medical data and the health data are obtained by acquiring the parameters obtained by the slave node through model training based on the encrypted model parameters and the local alignment data.

5. The method according to claim 1, characterized in that, The method further includes: updating the multidimensional profile data, specifically including: Obtain post-intervention health data based on a pre-set assessment cycle; Based on the risk assessment information and the post-intervention health data, an intervention evaluation is conducted to obtain intervention effect data; The multidimensional profile data is updated based on the intervention effect data.

6. The method according to any one of claims 1 to 5, characterized in that, The risk assessment information includes health improvement information, the current adjustment plan includes a rights allocation plan, and the process of predicting the plan based on the risk assessment information, obtaining the current adjustment plan, and sending the current adjustment plan to the slave node includes: Based on the aforementioned health improvement information, the improvement is categorized to obtain the improvement status; Based on the improvements described, improvement categories are obtained; A rights allocation scheme is generated based on the aforementioned improvement categories; The rights and interests allocation plan is pushed to the slave node.

7. The method according to any one of claims 1 to 5, characterized in that, The process of desensitizing the multi-source heterogeneous data to obtain multi-source desensitized data includes: Sensitive classification is performed on the multi-source heterogeneous data to obtain multi-source classification data with sensitive categories; The desensitization method is determined based on the aforementioned sensitivity category; wherein, the desensitization method includes at least one of the following: static desensitization and dynamic desensitization; The multi-source classification data is desensitized based on the aforementioned desensitization method to obtain the desensitized multi-source data.

8. A solution generation device based on user profiles, characterized in that, The device is applied to a master node, which is communicatively connected to at least one slave node, and the device includes: A multi-source heterogeneous data acquisition module is used to acquire multi-source heterogeneous data; wherein, the multi-source heterogeneous data originates from the master node and the slave node; The data desensitization module is used to desensitize the multi-source heterogeneous data to obtain multi-source desensitized data; The multi-dimensional profile extraction module is used to extract profiles from the multi-source de-identified data to obtain multi-dimensional profile data; wherein, the multi-dimensional profile data includes at least two of the following: health risk profile, health behavior profile, and health improvement profile; The risk assessment module is used to perform risk assessment based on the multidimensional profile data to obtain risk assessment information; The scheme generation module is used to predict the scheme based on the risk assessment information, obtain the current adjustment scheme, and send the current adjustment scheme to the slave node.

9. A computer device, characterized in that, The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.

Images