Intelligent Health Consultation Processing Method and Device Based on Private Domain Scenarios

Abstract

This invention proposes an intelligent health consultation processing method and device based on private domain scenarios, relating to the field of artificial intelligence technology. The method includes: real-time monitoring of consultation messages initiated by customers in private domain scenarios; identifying the consultation intent of the consultation messages and generating a query instruction corresponding to the consultation intent; according to the query instruction, calling the medical management system to obtain customer health data and calling the professional knowledge base to retrieve the professional knowledge corresponding to the query instruction; inputting the consultation message, customer health data, and professional knowledge into an artificial intelligence model to generate initial response content; correcting and verifying the initial response content according to the verification information in the professional knowledge base; if the verification passes and the consultation message is identified as not belonging to a preset consultation type, generating final response content; if the verification fails or the consultation message is identified as belonging to a preset consultation type, transferring the consultation to manual processing, and generating final response content after manual processing.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and more particularly to an intelligent health consultation processing method and device based on private domain scenarios. Background Technology

[0002] With increasing health awareness and growing demand for medical services, intelligent health consultation systems are being used more and more widely in private settings. Currently, health consultation services are mainly provided through online Q&A systems, health management platforms, and insurance service channels, with a particular emphasis on health management and consultation services for those already suffering from illness.

[0003] Existing technologies have significant shortcomings in scenarios involving private health services and insurance integration for patients with pre-existing conditions. First, existing general-purpose health Q&A robots cannot link to individual customer health data, resulting in standardized and mechanical responses that fail to meet the personalized needs of patients. Meanwhile, single-disease management systems only support data collection and knowledge delivery, lacking real-time Q&A interaction capabilities. Second, existing solutions often only cover a single stage of consultation, data processing, or conversion, lacking a comprehensive collaborative mechanism encompassing private domain interaction, artificial intelligence (AI) analysis, health data, and business conversion, thus failing to support a closed-loop business process. Expanding existing single-disease systems to other diseases requires reconstructing data collection forms, knowledge templates, and system logic, a process that takes 3 to 6 months and is costly, hindering low-cost and rapid disease expansion.

[0004] Especially in private domain scenarios, existing technologies struggle to effectively integrate health consultations and insurance services. On one hand, health consultation systems lack the ability to understand and integrate insurance rules; on the other hand, insurance service systems underutilize customer health data, failing to provide accurate health advice and insurance recommendations. Furthermore, existing systems suffer from inadequate collaboration mechanisms between artificial intelligence and human services when handling complex consultation scenarios, making it difficult to guarantee consultation quality and professionalism.

[0005] Therefore, there is an urgent need for an intelligent health consultation processing solution that can link health data and intelligent question answering in private domain scenarios, coordinate multiple scenarios and modules, and expand disease coverage quickly and at low cost, in order to meet the personalized health consultation needs of the already ill population and achieve effective linkage between health services and insurance business. Summary of the Invention

[0006] To address the problems existing in the prior art, this invention proposes an intelligent health consultation processing method and device based on private domain scenarios.

[0007] In a first aspect of the present invention, a method for intelligent health consultation processing based on private domain scenarios is proposed, the method comprising: Real-time monitoring of customer inquiries in private domain scenarios; identification of inquiry intent in the inquiry messages; and generation of query instructions corresponding to the inquiry intent. Based on the query instruction, the medical management system is invoked to obtain customer health data, and the professional knowledge base is invoked to retrieve the professional knowledge corresponding to the query instruction. The consultation message, customer health data, and professional knowledge are input into the artificial intelligence model to generate an initial response. The initial response content was corrected and verified according to the verification information in the professional knowledge base. If the verification passes and the inquiry message is identified as not belonging to the preset inquiry type, the final response content is generated. If the verification fails or the inquiry message is identified as belonging to a preset inquiry type, the inquiry will be transferred to manual processing, and the final response will be generated after manual processing.

[0008] In a second aspect of the present invention, an intelligent health consultation processing device based on a private domain scenario is provided, the device comprising: The message processing module is used to monitor customer consultation messages initiated in private domain scenarios in real time, identify the consultation intent of the consultation messages, and generate query instructions corresponding to the consultation intent. The data retrieval module is used to retrieve customer health data from the medical management system and retrieve professional knowledge corresponding to the query instruction from the professional knowledge base according to the query instruction. The intelligent reply module is used to input the consultation message, customer health data and professional knowledge into the artificial intelligence model to generate initial reply content; The correction and verification module is used to correct and verify the initial response content according to the verification information in the professional knowledge base. If the verification passes and the inquiry message is identified as not belonging to the preset inquiry type, the final response content is generated. The manual processing module is used to transfer the consultation to manual processing if the verification fails or the consultation message is identified as belonging to a preset consultation type. After manual processing, the final reply content is generated and pushed to the customer.

[0009] In a third aspect of the present invention, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement an intelligent health consultation processing method based on a private domain scenario.

[0010] In a fourth aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing a computer program, which, when executed by a processor, implements an intelligent health consultation processing method based on a private domain scenario.

[0011] In a fifth aspect of the present invention, a computer program product is provided, the computer program product comprising a computer program, which, when executed by a processor, implements an intelligent health consultation processing method based on a private domain scenario.

[0012] The intelligent health consultation processing method and device based on private domain scenarios proposed in this invention achieves automated consultation processing by real-time monitoring of consultation messages, identifying intent, calling health data and professional knowledge, generating responses through artificial intelligence, verification and correction, and a manual backup mechanism. The overall solution can significantly improve consultation response efficiency, and ensure the professionalism and accuracy of responses through the linkage of data and knowledge, providing strong technical support for private domain health services and insurance services. Attached Figure Description

[0013] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 This is a schematic diagram of a smart health consultation processing method based on a private domain scenario according to an embodiment of the present invention.

[0015] Figure 2 This is a flowchart illustrating the consultation intent recognition process according to a specific embodiment of the present invention.

[0016] Figure 3 This is a schematic diagram of the process of generating initial response content based on an artificial intelligence model according to a specific embodiment of the present invention.

[0017] Figure 4 This is a flowchart illustrating the emotion recognition process according to a specific embodiment of the present invention.

[0018] Figure 5 This is a schematic diagram of the manual processing flow according to a specific embodiment of the present invention.

[0019] Figure 6 This is a schematic diagram of the process for mining potential customers according to an embodiment of the present invention.

[0020] Figure 7 This is a schematic diagram of the architecture of an intelligent health consultation processing device based on a private domain scenario according to an embodiment of the present invention.

[0021] Figure 8 This is a schematic diagram of a computer device structure according to an embodiment of the present invention.

[0022] Figure 9 This is a schematic diagram of the system architecture of a specific embodiment of the present invention.

[0023] Figure 10 This is a schematic diagram of the processing logic of a specific embodiment of the present invention. Detailed Implementation

[0024] The principles and spirit of the invention will now be described with reference to several exemplary embodiments. It should be understood that these embodiments are given merely to enable those skilled in the art to better understand and implement the invention, and are not intended to limit the scope of the invention in any way. Rather, these embodiments are provided to make this disclosure more thorough and complete, and to fully convey the scope of this disclosure to those skilled in the art.

[0025] Those skilled in the art will recognize that embodiments of the present invention can be implemented as a system, apparatus, device, method, or computer program product. Therefore, this disclosure can be specifically implemented in the following forms: entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software.

[0026] According to an embodiment of the present invention, a method and apparatus for intelligent health consultation processing based on private domain scenarios are proposed, relating to the field of artificial intelligence technology.

[0027] The principles and spirit of the present invention will be explained in detail below with reference to several representative embodiments.

[0028] Figure 1 This is a schematic diagram of a smart health consultation processing method based on a private domain scenario according to an embodiment of the present invention. Figure 1 As shown, the method includes: S101, Real-time monitoring of customer consultation messages initiated in private domain scenarios, identification of consultation intent in the consultation messages, and generation of query instructions corresponding to the consultation intent; S102, according to the query instruction, call the medical management system to obtain customer health data, and call the professional knowledge base to retrieve the professional knowledge corresponding to the query instruction; S103, Input the consultation message, customer health data and professional knowledge into the artificial intelligence model to generate the initial response content; S104, Correct and verify the initial response content according to the verification information in the professional knowledge base; If the verification passes and the inquiry message is identified as not belonging to the preset inquiry type, the final response content is generated. If the verification fails or the inquiry message is identified as belonging to a preset inquiry type, the inquiry will be transferred to manual processing, and the final response will be generated after manual processing.

[0029] The intelligent health consultation processing method proposed in this invention, based on private domain scenarios, achieves end-to-end automation of consultation processing by real-time monitoring of consultation messages, identifying intent, accessing health data and professional knowledge, generating responses based on artificial intelligence (AI), and implementing verification, correction, and human fallback mechanisms. Compared to existing technologies that rely on human intervention or generic robotic responses, this invention significantly improves consultation response efficiency. Furthermore, the integration of data and knowledge ensures the professionalism and accuracy of responses, providing strong technical support for health consultation and insurance service scenarios.

[0030] To provide a clearer explanation of the above-mentioned intelligent health consultation processing method based on private domain scenarios, each step will be explained in detail below.

[0031] In one embodiment, for S101, real-time monitoring of customer consultation messages initiated in private domain scenarios, identification of consultation intent in the consultation messages, and generation of query instructions corresponding to the consultation intent.

[0032] This invention possesses the ability to subdivide and identify consultation intent, supporting the simultaneous breakdown of customer inquiries into health-related intents and insurance-related intents, and generating corresponding query commands for each. This process overcomes the limitation of traditional question-and-answer robots that can only handle single intents, enabling the system to accurately respond to customers' complex consultation needs regarding health and insurance in private domain scenarios. This ensures the integrity of the linkage between health consultation services and insurance business, avoiding service disconnects caused by incomplete intent recognition.

[0033] refer to Figure 2 This is a schematic diagram of a consultation intent recognition process according to a specific embodiment of the present invention. Figure 2 As shown, the specific process is as follows: S1011, Identify the consultation intent of the consultation message through a semantic analysis algorithm, wherein the consultation intent includes at least a health need intent and an insurance service need intent; S1012, based on the stated health need intention and insurance service need intention, generate a first query instruction corresponding to the stated health need intention and a second query instruction corresponding to the stated insurance service need intention, respectively.

[0034] This solution uses composite intent recognition to handle cross-domain questions about health and insurance simultaneously in a single conversation flow, eliminating the need for customers to switch between different systems or groups, thus improving customer experience and insurance conversion efficiency.

[0035] In one embodiment, for S102, according to the query instruction, the medical management system is invoked to obtain customer health data, and the professional knowledge base is invoked to retrieve the professional knowledge corresponding to the query instruction.

[0036] Specifically, according to the first query instruction, customer health data is retrieved from the medical management system. The customer health data includes at least basic customer information, medical history data, and insurance records. According to the second query instruction, the corresponding professional knowledge is retrieved from the professional knowledge base, which includes at least medical knowledge and insurance rule information.

[0037] This invention proposes a method of using different query commands to invoke different data sources. Specifically, a first query command retrieves customer health data, while a second query command retrieves professional knowledge. This achieves precise matching of health data and professional knowledge, enabling the AI ​​model to simultaneously consider the customer's individual treatment background (such as medical history and medication) and authoritative medical and insurance regulations when generating responses. This effectively improves the individualization of responses and avoids the professional risks associated with generic answers in different customer scenarios.

[0038] In this embodiment, the professional knowledge base adopts a basic knowledge framework and is equipped with disease-specific plugins; wherein, the basic knowledge framework is configured with knowledge classification standards, question-and-answer matching rules and access control logic; the disease-specific plugins encapsulate the knowledge and data mapping rules of a single disease, and support knowledge updates and disease-specific plugin replacements.

[0039] For knowledge classification standards, medical knowledge (cause, treatment, rehabilitation) and insurance rules (coverage scope, claims process) can be stored by disease type.

[0040] The professional knowledge base adopts a modular architecture design with a basic knowledge framework and disease-specific plugins. By decoupling general logic from specific knowledge, when adding a new disease, only the corresponding plugin needs to be replaced or loaded, shortening the expansion cycle to 1-2 weeks (the traditional single-disease system expansion requires code reconstruction, with an expansion cycle of 3-6 months). This greatly reduces the development cost and technical complexity of adapting the system to new businesses, and supports the rapid and large-scale expansion of insurance business from single diseases to multiple diseases.

[0041] In this embodiment, the Medical Management System (DMS) stores basic customer information (age, disease type), medical history data (diagnosis time, treatment plan), and insurance records (insured products, claims status) to provide personalized data for the artificial intelligence model.

[0042] Specifically, the method also includes: The data access permissions of the medical management system and professional knowledge base are controlled; in particular, when processing consultation messages, customer health data and professional knowledge are temporarily retrieved according to the data access permissions.

[0043] This invention establishes a data access control mechanism for the medical management system and professional knowledge base. By adhering to the principle of temporary data retrieval, it ensures highly personalized responses while maximizing the protection of customer privacy and data security. By implementing a data access minimization strategy, it prevents the unnecessary storage and misuse of sensitive customer health data, ensuring that the technical solution strictly complies with relevant laws and regulations in the process of implementing intelligent services, thus eliminating data compliance risks from the architectural design level.

[0044] In one embodiment, for S103, the consultation message, customer health data, and professional knowledge are input into an artificial intelligence model to generate initial response content.

[0045] refer to Figure 3 This is a schematic diagram illustrating the process of generating initial response content based on an artificial intelligence model according to a specific embodiment of the present invention. Figure 3 As shown, the specific process is as follows: S1031, Obtain customer historical consultation text, and supplement the consultation message with context based on the customer historical consultation text to obtain consultation text; S1032, Identify the emotion tags in the consultation text and call the corresponding empathy script template; S1033, integrate the consultation text, customer health data and professional knowledge, and generate initial response content using the empathic dialogue template.

[0046] This invention incorporates mechanisms for supplementing historical dialogue context and emotional adaptation during the response generation process. Through multi-round dialogue association, it ensures interactive continuity in follow-up questioning scenarios, eliminating the need for customers to repeatedly describe their symptoms, thus improving consultation efficiency and experience. Furthermore, by recognizing customer emotions and employing empathetic language, the machine response is upgraded from mechanical notification to warm and caring, significantly reducing anxiety among those already ill during consultations and improving customer satisfaction and engagement.

[0047] Specifically, for S1031, in practical application scenarios, consultation messages can include user-input text, voice, and uploaded images. For images, key information (e.g., inspection indicators, numerical values, and conclusion descriptions) is extracted through recognition and intelligent analysis functions and converted into structured text. When supplementing the context, the continuity of the dialogue is ensured when visual materials such as inspection reports and images are involved, based on related historical consultation texts (which can refer to the image analysis results from previous consultations).

[0048] refer to Figure 4 In S1032, the process of identifying emotion tags in the consultation text and calling the corresponding empathy script template is as follows: S1032-1, Identify emotional tags (e.g., anxiety, panic) in the consultation text through keyword matching and semantic tendency analysis. S1032-2, Based on the emotion tag, call the preset empathy script template to adjust the expression of the initial response content.

[0049] By using keyword matching and semantic tendency analysis, we can identify the implicit emotional tags in the current round of consultation text (including text descriptions extracted from images). For example, when words such as "abnormally high" or "critical value" appear in the text description or image analysis results, we can identify the emotions of "anxiety" or "worry" and then call up the corresponding reassurance and care script templates.

[0050] For S1033, the current consultation text, which integrates image information, the client's long-term health record data, and relevant knowledge retrieved from a professional knowledge base, are input into the artificial intelligence model. Within the framework of empathetic communication, the model generates an initial response that includes a targeted interpretation of the current examination results. For example, combining the client's "diabetes history" and the "blood glucose level of 15 mmol / L" extracted from the consultation image, a personalized, data-driven response is generated, such as, "We see that your blood glucose level is a bit high this time, but don't worry too much. We suggest you pay a little attention to your diet."

[0051] This invention identifies emotions such as anxiety and panic and adjusts responses accordingly through keyword matching and semantic tendency analysis. The artificial intelligence model possesses affective computing capabilities, enabling gentle intervention targeting the negative emotions commonly found in patients. Compared to conventional, indiscriminate, and rigid text responses, this invention effectively reduces the psychological burden on clients when obtaining medical information. Practical verification has shown that it significantly reduces client anxiety feedback rates and enhances the humanistic level of service.

[0052] In one embodiment, for S104, the initial response content is corrected and verified according to the verification information in the professional knowledge base; If the verification passes and the inquiry message is identified as not belonging to the preset inquiry type, the final response content is generated. If the verification fails or the inquiry message is identified as belonging to a preset inquiry type, the inquiry will be transferred to manual processing, and the final response will be generated after manual processing.

[0053] After generating the final response, it is pushed to the customer, thereby providing precise services and improving customer loyalty.

[0054] The preset consultation types include at least the following: consultations involving adjustments to personalized treatment plans, emergency medical warnings, or the use of high-risk drugs.

[0055] By automatically identifying complex or high-risk issues that exceed the processing scope of artificial intelligence through preset consultation types and judgment logic, and forcibly transferring them to human agents, the system avoids service and legal risks caused by inappropriate medical advice due to misjudgment by artificial intelligence processing methods, thus ensuring the security and compliance of intelligent responses from a technical perspective.

[0056] Among them, reference Figure 5 This is a schematic diagram of the emotion recognition process according to a specific embodiment of the present invention. Figure 5 As shown, the specific process for manual processing includes: S501, the initial response content is pushed to the manual backup interaction interface, and the processing instructions input by the health consultant or insurance agent through the manual backup interaction interface are received to generate the final response content; S502, mark the final response content after manual processing, and the marked information should include at least the disease classification, consultation scenario and knowledge points; The labeled information can be used as training labels to assist in the training of professional knowledge bases and artificial intelligence models.

[0057] S503 synchronizes the annotated final response content to the case library of the professional knowledge base and to the training set of the artificial intelligence model.

[0058] This invention performs structured annotation on high-quality manually processed responses and feeds them back into the training set of the professional knowledge base and artificial intelligence model. This breaks through the shortcomings of the static and non-iterative knowledge base in the prior art, thereby giving the system architecture the ability to evolve on its own. This allows the accuracy of question answering to continuously improve as the system runs, achieving a process where the more services are provided, the more accurate the knowledge becomes and the higher the efficiency.

[0059] In another embodiment, to achieve pre-service delivery and precise marketing, the present invention also proposes a mechanism for identifying potential customers. (See reference...) Figure 6 The method includes: S601, record the customer's consultation behavior, which includes at least the frequency of consultations initiated by the customer within a preset period, the consultation content, and the consultation interaction duration; S602, when the consultation behavior meets the preset conditions, the customer is marked as a high-intent customer and a follow-up reminder is sent to the insurance agent.

[0060] In practical applications, the preset conditions can be: consultation frequency ≥ 3 times / month, consultation content includes insurance-related questions, and interaction time ≥ 5 minutes / time.

[0061] When a customer's consultation behavior meets one or more preset conditions, the customer will be automatically marked as a "high-intent customer" and the information will be pushed to the agent. Specific preset conditions can be set according to actual application scenarios. For example, different preset conditions can be set for different types of diseases or insurance types; each health consultant / insurance agent can also set different preset conditions.

[0062] This invention applies artificial intelligence technology to customer value mining. By setting preset conditions and judgment logic, it automatically identifies potential customers who frequently inquire about insurance terms and are interested in them, helping insurance agents to accurately capture sales leads and achieve precise conversion from blind sales to service-based sales, thereby improving insurance conversion rates.

[0063] It should be noted that although the operation of the method of the present invention has been described in a specific order in the above embodiments and figures, this does not require or imply that the operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0064] After introducing the method of exemplary embodiments of the present invention, the following references are made. Figure 7 This invention provides an exemplary embodiment of an intelligent health consultation processing device based on a private domain scenario.

[0065] The implementation of the intelligent health consultation processing device based on private domain scenarios can refer to the implementation of the above methods, and the repeated parts will not be described again. The term "module" or "unit" used below can be a combination of software and / or hardware to achieve a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0066] Based on the same inventive concept, this invention also proposes an intelligent health consultation processing device based on a private domain scenario, such as... Figure 7 As shown, the device includes: The message processing module 710 is used to monitor customer consultation messages initiated in private domain scenarios in real time, identify consultation intent in the consultation messages, and generate query instructions corresponding to the consultation intent. The data retrieval module 720 is used to retrieve customer health data from the medical management system and retrieve professional knowledge corresponding to the query instruction from the professional knowledge base according to the query instruction. The intelligent reply module 730 is used to input the consultation message, customer health data and professional knowledge into the artificial intelligence model to generate initial reply content; The correction and verification module 740 is used to correct and verify the initial response content according to the verification information in the professional knowledge base. If the verification passes and the inquiry message is identified as not belonging to the preset inquiry type, the final response content is generated. The manual processing module 750 is used to transfer the consultation to manual processing if the verification fails or the consultation message is identified as belonging to a preset consultation type. After manual processing, the final reply content is generated and pushed to the customer.

[0067] It should be noted that although several modules of the intelligent health consultation processing device based on private domain scenarios have been mentioned in the detailed description above, this division is merely exemplary and not mandatory. In fact, according to embodiments of the present invention, the features and functions of two or more modules described above can be embodied in one module. Conversely, the features and functions of one module described above can be further divided and embodied by multiple modules.

[0068] Based on the aforementioned inventive concept, such as Figure 8 As shown, the present invention also proposes a computer device 800, including a memory 810, a processor 820, and a computer program 830 stored in the memory 810 and executable on the processor 820. When the processor 820 executes the computer program 830, it implements the aforementioned intelligent health consultation processing method based on private domain scenarios.

[0069] Based on the aforementioned inventive concept, the present invention proposes a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the aforementioned intelligent health consultation processing method based on a private domain scenario.

[0070] Based on the aforementioned inventive concept, this invention proposes a computer program product, which includes a computer program that, when executed by a processor, implements an intelligent health consultation processing method based on a private domain scenario.

[0071] The following describes the intelligent health consultation processing method and device based on private domain scenarios proposed in this invention, with reference to a specific embodiment.

[0072] To achieve intelligent health consultation processing based on private domain scenarios, a "three-layer, four-domain" system architecture is constructed. This architecture, consisting of an "access layer," a core processing layer, and a support layer, enables cross-domain collaboration among the "customer domain," "business domain," "data domain," and "knowledge domain." (Reference) Figure 9 This is a schematic diagram of the system architecture of a specific embodiment of the present invention. Figure 9As shown, the access layer 910 serves as the interaction point between the customer and the system, including: chatbots (e.g., enterprise WeChat RPA robots), human-assisted interactive interfaces, and data statistics dashboards.

[0073] Chatbot: Connects to private communities (chat groups) of patients with existing conditions to monitor customer inquiries in real time. Customer inquiries support multiple input formats, including text, voice, and images. For text and voice messages, it automatically captures or converts them to text; for images such as medical examination reports uploaded by users, it can recognize and intelligently analyze them, extracting key information (such as indicator names, values, and units) and converting them into structured text. It also supports message flow with a human assistant interface.

[0074] Human-assisted interactive interface: Provides health consultants or insurance agents with solutions to complex inquiries that AI cannot answer. Supports viewing customer health records, historical conversations, and the initial responses from AI, and then feeds the results back to the chatbot after processing.

[0075] Data statistics dashboard: Real-time display of data such as inquiry volume, AI response accuracy, manual guarantee rate, premium conversion rate, etc., to support operational decision-making.

[0076] The core processing layer 920, with the artificial intelligence model module at its core, undertakes the functions of consultation analysis, response generation, and quality control, including: multi-turn dialogue unit, emotion adaptation unit, professional verification unit, and intent recognition unit.

[0077] Multi-turn dialogue units: link customer's historical consultation content to achieve seamless interaction; for example, there is no need to repeat background information when the customer asks follow-up questions.

[0078] Emotional Adaptation Unit: Generates empathic responses through emotion recognition algorithms; for example, it calls up reassuring phrases after recognizing the emotion of "anxiety".

[0079] Professional verification unit: Verify the accuracy of the responses against the professional knowledge base and correct any discrepancies.

[0080] Intent recognition unit: Determines the type of customer inquiry (health issues, insurance services, emergency needs) and retrieves corresponding knowledge and data accordingly.

[0081] For the support layer 930, data and knowledge support is provided, including: medical management system, professional knowledge base, and customer access management module.

[0082] Medical management system: Stores basic customer information (age, disease type), medical history data (diagnosis time, treatment plan), and insurance records (insured products, claims status), providing personalized data for artificial intelligence models; Professional knowledge base: Stores medical knowledge (cause, treatment, rehabilitation) and insurance rules (coverage, claims process) categorized by disease type, and supports knowledge updates and disease-specific plugin replacement; Customer access management module: Controls data access permissions to ensure privacy and compliance; for example, only allowing AI models to access a customer's health data when processing a corresponding customer's inquiry.

[0083] based on Figure 9 The architecture provides a detailed explanation of the intelligent health consultation process.

[0084] When a customer sends an inquiry in a private community (chat group), the chatbot captures the message and identifies the customer, retrieves the customer's health data from the medical management system, and synchronizes the inquiry message and health data to an artificial intelligence model (e.g., an AI large language model). The artificial intelligence model calls on a professional knowledge base to generate a response, and a professional verification unit verifies the response.

[0085] If the verification passes (and it is not a complex inquiry), the chatbot pushes a reply to the customer, and the data statistics dashboard records the service data. If the verification fails (or the consultation is complex), it will be transferred to the human assistant interface. After manual processing, feedback will be sent to the customer, and the human reply will be synchronized to the professional knowledge base for model training.

[0086] This invention covers the entire process logic from consultation reception, intelligent processing, human backup, and knowledge accumulation. Its core feature is a collaborative model that prioritizes artificial intelligence (AI) while supplementing human intervention. Routine consultations (e.g., dietary precautions after chemotherapy) are processed end-to-end by AI, with a response time of ≤1 minute. Complex consultations (e.g., personalized treatment plan adjustment suggestions) or urgent needs (e.g., sudden pain) are automatically transferred to human agents, ensuring service professionalism and safety. Furthermore, human responses are annotated and synchronized to a knowledge base to continuously optimize the AI ​​model, forming a closed-loop processing logic of service-learning-iteration.

[0087] The following is combined Figure 10 The diagram shown illustrates the processing logic, and the processing logic is explained in detail.

[0088] S1: Input: Customer inquiry messages, which can include text, voice (after being transcribed), or images.

[0089] Source: Private Domain Community.

[0090] Executor: Chatbot.

[0091] Execution content: For text or voice messages, directly capture or convert them; for images (e.g., examination reports), invoke the image recognition and intelligent analysis module to perform OCR recognition and content interpretation, extract key information, and generate structured text descriptions. Subsequently, match the customer's health record ID in the medical management system using the customer ID.

[0092] Output: {Consultation text, Customer ID, Health record ID}, output to the artificial intelligence model. The output consultation text may include text generated from image analysis.

[0093] S2: Input: {Consultation text, Customer ID, Health record ID}.

[0094] Source: Chatbot.

[0095] Execution entity: Artificial intelligence model (intent recognition unit).

[0096] Execution content: Identify the consultation intent (e.g., health consultation, insurance claim) and generate query instructions.

[0097] Output: {Consultation Intent, Query Command, Health Record ID}, output to the medical management system and professional knowledge base.

[0098] In one specific embodiment, S2 is used for intent recognition and targeted querying to solve the problem of general question answering lacking specificity.

[0099] For example, the input information includes customer inquiry messages (e.g., I am currently undergoing chemotherapy and cannot eat. Can my insurance cover nutritional supplements?), customer ID, and health record ID.

[0100] The executing entity is the intent recognition unit of the artificial intelligence model.

[0101] The execution content is as follows: through semantic analysis, the consultation is broken down into "health needs (chemotherapy diet) and insurance needs (nutritional supplement reimbursement)", and two targeted query instructions are generated: 1. retrieve the current chemotherapy stage data from the medical management system; 2. retrieve chemotherapy diet advice and nutritional supplement reimbursement rules from the professional knowledge base.

[0102] This step can overcome the limitations of general robots' "single intent recognition", support the splitting of complex intents, ensure that AI can respond to health and insurance needs at the same time, and solve the pain point of "lack of business linkage" in existing technologies.

[0103] S3: Input: {Query command, Health Record ID}.

[0104] Source: Artificial intelligence model.

[0105] Implementing entity: Medical Management System.

[0106] Execution content: Retrieve personalized customer data based on health record ID (e.g., 3 months post-breast cancer surgery, chemotherapy regimen AC).

[0107] Output: {Customer health data}, output to the artificial intelligence model.

[0108] S4: Input: {Inquiry Intent, Query Command}.

[0109] Source: Artificial intelligence model.

[0110] Implementing entity: Professional knowledge base.

[0111] Execution content: Retrieve relevant knowledge based on consultation intent (e.g., nursing care for side effects of chemotherapy in breast cancer).

[0112] Output: {Related knowledge fragments}, output to the artificial intelligence model.

[0113] S5: Input: {Consultation text, customer health data, related knowledge fragments}.

[0114] Source: Medical Management System, Knowledge Base, Chatbot.

[0115] Execution entity: Artificial intelligence model (multi-turn dialogue unit, emotion adaptation unit).

[0116] Execution content: Customize knowledge segments (e.g., rehabilitation exercise recommendations 3 months after surgery) based on the client's health data (e.g., 3 months after surgery).

[0117] Identify customer emotions (e.g., anxiety) and generate responses using empathic dialogue templates.

[0118] Output: {Initial response content}, output to the professional verification unit.

[0119] In this embodiment, the "input" consultation text may include the user's original input text and a structured description of key information (e.g., indicators, values) extracted from uploaded images (e.g., check-up forms) through intelligent analysis.

[0120] Through data fusion, combining customer health data (e.g., medical history) with current examination data extracted from images, related knowledge fragments are deeply customized. For example, based on the customer's "diabetes history" and "current blood sugar level" obtained from image analysis, specific dietary recommendations are selected. Emotion recognition is performed to identify the emotions implied in the consultation text (including image captions). Response generation utilizes empathetic language templates to generate initial responses that integrate the customer's personalized health data, professional knowledge, and targeted interpretation of the image examination results.

[0121] In one specific embodiment, S5 is used to generate personalized and emotional responses to address the problems of mechanical and impersonal responses.

[0122] Taking the input information as customer consultation text, the data returned by the medical management system as "3 months after breast cancer surgery, AC chemotherapy regimen", and the knowledge base returned as "dietary restrictions of AC regimen" as examples.

[0123] The main implementers are: the multi-turn dialogue unit and the emotion adaptation unit of the artificial intelligence model.

[0124] The execution content is as follows: Personalized customization: Based on the data from "3 months post-surgery", "mid-post-operative dietary recommendations" are selected from the knowledge (excluding the initial prohibitions) and it is suggested that "high-sugar foods should be avoided in the AC regimen".

[0125] Emotional Adaptation: Recognizing the anxiety implied in the customer's text regarding "not being able to eat," the template is used to generate: "We understand the hardship of poor appetite during chemotherapy. This is a common reaction to the AC regimen. You can try these light recipes suitable for 3 months after surgery (with specific suggestions), which can both supplement nutrition and meet your needs at this stage of treatment."

[0126] This step upgrades responses from "general and standardized" to "personalized and emotional" through triple input of health data, knowledge fragments, and emotion tags, addressing the core pain point that existing technologies cannot adapt to the emotions and individual needs of the already ill population.

[0127] S6: Input: {Initial response content, related knowledge fragments}.

[0128] Source: Artificial intelligence model.

[0129] Execution entity: Artificial intelligence model (professional verification unit).

[0130] Tasks include: verifying whether the response complies with medical standards (e.g., whether any side effect risk warnings are omitted), and correcting any errors.

[0131] Output: {Final reply content / marker requiring manual processing}, output to the chatbot / human backup interaction interface.

[0132] In this embodiment, content verification is performed by comparing the initial response with a professional knowledge base to check its medical accuracy and compliance with insurance rules. This verification specifically includes reviewing the data and conclusions cited in the response and derived from image analysis to ensure that their interpretation is accurate and conforms to medical standards. Further risk assessment is conducted. If the initial response is primarily based on the interpretation of a complex image (e.g., a CT scan) or involves urgent treatment suggestions for abnormal examination results, the system can determine it as high-risk or beyond the scope of the consultation, generating a "manual processing required" flag.

[0133] In one specific embodiment, S6 is used for professional verification and manual processing to address the issues of response accuracy and risk control.

[0134] Taking the initial response generated by the artificial intelligence model based on the input information as an example, such as "A certain nutritional supplement can be taken during chemotherapy, and insurance will reimburse 100% of it."

[0135] Execution entity: Professional verification unit for artificial intelligence models.

[0136] Execution content: 1. Medical accuracy verification: Comparing with the "applicable scenarios of a certain nutritional supplement" in the professional knowledge base, it was found that "it can only be used for severe malnutrition". The statement was corrected to "if the patient is assessed by a doctor as having severe malnutrition, the nutritional supplement can be included in the reimbursement."

[0137] 2. Insurance rule verification: Check the "reimbursement ratio" in the knowledge base and find that "the actual reimbursement ratio is 80%", then correct the ratio information.

[0138] 3. Complex Problem Judgment: If a customer inquires about "whether the chemotherapy regimen can be changed", it is judged as a "complex problem requiring a doctor's decision" and a "manual processing required" label is generated.

[0139] This step avoids professional errors through a dual verification mechanism and establishes "AI boundary recognition" rules to ensure that complex / high-risk issues are handled manually, thus addressing the pain points of existing technologies being "insufficient in professionalism and uncontrollable in risk".

[0140] S7: Input: {Manual processing required, inquiry text, customer health data, initial response}.

[0141] Source: Artificial intelligence model.

[0142] Implementation entity: Manual support interface (health consultant).

[0143] Task: Manually edit replies and resolve complex issues.

[0144] Output: {human response content}, output to chatbot and knowledge base.

[0145] In this embodiment, after manual review and editing, the health consultant can view the user's original consultation (including uploaded images), the system's intelligent analysis text of the images, the AI-generated initial response, and related health records on the interface, and conduct final review, correction, or re-editing to form a professional and secure final response.

[0146] In practical applications, knowledge can also be accumulated; high-quality responses processed manually can be annotated, with annotation information including disease classification, consultation scenario (e.g., "interpretation of imaging reports," "analysis of laboratory reports"), and relevant knowledge points. The annotated cases are then synchronized to the case library and AI training set in the professional knowledge base.

[0147] In one specific embodiment, S7 is used for manual reply to knowledge accumulation to solve the problem of the system's lack of self-evolution capability.

[0148] For example, if the input information is a manually processed response, such as, "Given your history of diabetes, you need to pay extra attention to certain aspects of your diet during chemotherapy."

[0149] The implementing entity is: the professional knowledge base.

[0150] The execution involves automatically labeling the response as a case of "diabetes combined with breast cancer chemotherapy diet," storing it in the knowledge base according to the disease type, and synchronizing it to the training set of the artificial intelligence model to optimize the response logic for subsequent similar questions.

[0151] This step transforms human experience into system knowledge, enabling the accuracy of AI models to improve over time (expected to increase by 3-5% per quarter), thus addressing the limitation of existing technologies where "static knowledge bases cannot be iterated."

[0152] Through the above architecture and process design, this invention achieves deep collaboration between "data, knowledge, AI and business", comprehensively solves the pain points of existing technologies in terms of personalization, professionalism and scalability, and supports the intelligent upgrade of private domain services for patients.

[0153] S8: Input: {Final reply content / Manual reply content}.

[0154] Source: Artificial intelligence model / human-assisted interactive interface.

[0155] Executor: Chatbot.

[0156] Execution content: Push the reply to the customer and record service data (response time, type).

[0157] Output: {Service record data}, output to the data statistics dashboard.

[0158] Based on the intelligent health consultation system architecture and processing flow proposed in this invention, at least the following technical effects can be achieved: 1. Breaking through the technological gap in "data-question-answering" linkage, achieving personalized data-driven intelligent question-and-answering. Through real-time interaction between the medical management system and the artificial intelligence model, the problem of the separation between health data and question-and-answer functions in existing technologies is solved. When generating responses, artificial intelligence can automatically call upon personalized data such as the client's medical history and treatment stage, upgrading the question-and-answer process from general to customized, greatly improving data utilization and avoiding professional bias caused by the lack of individual data.

[0159] 2. This invention achieves multi-module collaboration and process automation, filling the gap in end-to-end technology collaboration in private domain scenarios. By connecting the entire process of "consultation retrieval - data retrieval - response generation - result push" through a chatbot, it solves the problem of "independent operation of the front-end, middle-end, and back-end" in existing solutions. This invention completes end-to-end processing of basic consultations without human intervention, achieving a 90% automation rate. It also supports seamless transition between AI and human intervention (automatically triggering human intervention for complex issues), improving system response efficiency by 10 times.

[0160] 3. Construct a modular technical architecture adapted to specific diseases to overcome the technical constraints of scaling up business operations. The professional knowledge base adopts a decoupled design of "basic framework + disease-specific plugins," solving the technical limitation of "refactoring required for expansion" in single-disease systems. When adding a new disease, only the dedicated knowledge base plugin needs to be replaced, without modifying the core code. The disease expansion cycle is shortened from 3-6 months to 1-2 weeks, and the technical adaptation cost is reduced by 70%, supporting the system to quickly cover multiple groups of patients with existing diseases.

[0161] 4. Enhance the credibility and security of AI responses by establishing a professional verification technology barrier. Through a dual control mechanism of "real-time verification by a professional knowledge base + manual review," the professional error rate of AI responses has been reduced from over 15% to below 8% using existing technologies. Simultaneously, the system can automatically identify high-risk consultations (e.g., adjustments to treatment plans) and trigger manual intervention, avoiding risks to medical advice due to AI misjudgments and ensuring service security from a technical perspective.

[0162] 5. Enables dynamic iteration of knowledge and models, possesses the ability for technological self-evolution, and constructs a closed-loop mechanism of "human response - knowledge accumulation - AI training". High-quality responses processed by humans are automatically marked as knowledge base cases and synchronized to the AI ​​training set, so that the accuracy of AI question answering continues to improve with the length of use (3-5% increase per quarter). This solves the problem of "lagging updates and inability to iterate" in the existing static knowledge base, and the system's technical capabilities can be optimized in the long term.

[0163] 6. Ensure customer data privacy and compliance by building an access control technology system. The customer access management module achieves "minimum data access" control: AI can only temporarily access a customer's health data when processing their inquiry and cannot store the original data; human staff can only view information about customers they are responsible for, eliminating the risk of data leakage. Technically, it meets relevant regulatory requirements, achieving 100% data compliance.

[0164] At the business level, at least the following improvements can be achieved: 1. Cost Reduction and Efficiency Improvement: Significantly reduces labor costs for private domain services and improves operational efficiency; with a human substitution rate exceeding 90%, only a small number of staff are needed to support the operation of a private domain community of thousands, eliminating the need to hire a large number of specialist doctors and saving significant labor costs. At the same time, the average customer consultation waiting time has been reduced from 10 minutes to less than 1 minute, and the service complaint rate has decreased by 42%, resulting in a significant improvement in operational efficiency.

[0165] 2. Service Upgrade: The transformation from "basic consultation" to "emotionally-oriented professional service" has added "24-hour AI health consultant" and "disease-specific knowledge push" services, increasing the average daily consultation frequency of customers from 0.8 times to 2.3 times; the emotional adaptation function has reduced the customer anxiety feedback rate by 37%, and the customer satisfaction rate has reached 91%. The service experience has been upgraded from "mechanical response" to "warm professional service", significantly enhancing customer stickiness.

[0166] 3. Business Growth: Driving a closed-loop "service-sales" system and improving premium conversion rates. AI identifies high-intent customers and sends notifications, assisting agents in precise follow-up and effectively increasing insurance premiums. Simultaneously, the ability to rapidly expand disease coverage supports the expansion of insurance business from "single disease" to "multi-disease" coverage, with an expected coverage of 5 core diseases within one year, resulting in a 50% increase in business scale.

[0167] 4. Ecosystem Building: This invention creates a reusable technical template to support horizontal business expansion. Its technical architecture can be reused in scenarios such as hospital patient follow-up and community services for health management companies. It can collaborate with hospitals to adapt the system for "post-operative rehabilitation consultation services," creating a new revenue model of "technology output + service monetization" and expanding the boundaries of the business ecosystem.

[0168] The acquisition, storage, use, and processing of data in the technical solution of this invention all comply with the relevant provisions of laws and regulations.

[0169] The intelligent health consultation processing method and device based on private domain scenarios proposed in this invention achieves automated consultation processing by real-time monitoring of consultation messages, identifying intent, calling health data and professional knowledge, generating responses through artificial intelligence, verification and correction, and a manual backup mechanism. The overall solution can significantly improve consultation response efficiency, and ensure the professionalism and accuracy of responses through the linkage of data and knowledge, providing strong technical support for private domain health services and insurance services.

[0170] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0171] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0172] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0173] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0174] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for processing intelligent health consultations based on private domain scenarios, characterized in that, The method includes: Real-time monitoring of customer inquiries in private domain scenarios; identification of inquiry intent in the inquiry messages; and generation of query instructions corresponding to the inquiry intent. Based on the query instruction, the medical management system is invoked to obtain customer health data, and the professional knowledge base is invoked to retrieve the professional knowledge corresponding to the query instruction. The consultation message, customer health data, and professional knowledge are input into the artificial intelligence model to generate an initial response. The initial response content was corrected and verified according to the verification information in the professional knowledge base. If the verification passes and the inquiry message is identified as not belonging to the preset inquiry type, the final response content is generated. If the verification fails or the inquiry message is identified as belonging to a preset inquiry type, the inquiry will be transferred to manual processing, and the final response will be generated after manual processing.

2. The method according to claim 1, characterized in that, Real-time monitoring of customer inquiries initiated in private domain scenarios, identification of the inquiry intent in the inquiries, and generation of query instructions corresponding to the inquiry intent, including: The consultation intent of the consultation message is identified by a semantic analysis algorithm, wherein the consultation intent includes at least the intent to seek health services and the intent to seek insurance services. Based on the stated health needs intent and insurance service needs intent, a first query instruction corresponding to the stated health needs intent and a second query instruction corresponding to the stated insurance service needs intent are generated respectively.

3. The method according to claim 2, characterized in that, Based on the query instruction, the system retrieves customer health data from the medical management system and searches the professional knowledge base for the professional knowledge corresponding to the query instruction, including: According to the first query instruction, customer health data is retrieved from the medical management system. The customer health data includes at least basic customer information, medical history data, and insurance records. According to the second query instruction, the corresponding professional knowledge is retrieved from the professional knowledge base, which includes at least medical knowledge and insurance rule information.

4. The method according to claim 1, characterized in that, The consultation message, customer health data, and professional knowledge are input into an artificial intelligence model to generate an initial response, including: Obtain the customer's historical consultation text, and supplement the consultation message with context based on the customer's historical consultation text to obtain the consultation text; Identify the emotion tags in the consultation text and call up the corresponding empathy script template; By integrating the consultation text, customer health data, and professional knowledge, the initial response content is generated using the empathic dialogue template.

5. The method according to claim 4, characterized in that, Identify the emotion tags in the consultation text and call the corresponding empathy script template, including: By using keyword matching and semantic tendency analysis, the emotional tags in the consultation text were identified; The initial response content is adjusted by calling a preset empathy script template based on the emotion tag.

6. The method according to claim 1, characterized in that, The professional knowledge base adopts a basic knowledge framework and includes disease-specific plugins. The basic knowledge framework is configured with knowledge classification standards, question-and-answer matching rules, and access control logic. The disease-specific plugins encapsulate the knowledge and data mapping rules for a single disease.

7. The method according to claim 1, characterized in that, The preset consultation types include at least: consultations involving adjustments to personalized treatment plans, emergency medical warnings, or the use of high-risk drugs; If the verification fails or the inquiry message is identified as belonging to a preset inquiry type, the inquiry will be transferred to manual processing. After manual processing, a final response will be generated; the final response will then be sent to the customer, including: The initial response content is pushed to the human-assisted interactive interface, and the processing instructions input by the health consultant or insurance agent through the human-assisted interactive interface are received to generate the final response content; The final response after manual processing is annotated, and the annotated information should include at least the disease classification, consultation scenario, and knowledge points. The annotated final response content will be synchronized to the case library of the professional knowledge base and to the training set of the artificial intelligence model.

8. The method according to claim 1, characterized in that, The method also includes: Record customer consultation behavior, which includes at least the frequency of consultations initiated by the customer within a preset period, the content of the consultations, and the duration of the consultation interactions; When the consultation behavior meets the preset conditions, the customer is marked as a high-intent customer, and a follow-up reminder is sent to the insurance agent.

9. The method according to claim 1, characterized in that, The method also includes: The data access permissions of the medical management system and professional knowledge base are controlled; in particular, when processing consultation messages, customer health data and professional knowledge are temporarily retrieved according to the data access permissions.

10. A smart health consultation processing device based on a private domain scenario, characterized in that, The device includes: The message processing module is used to monitor customer consultation messages initiated in private domain scenarios in real time, identify the consultation intent of the consultation messages, and generate query instructions corresponding to the consultation intent. The data retrieval module is used to retrieve customer health data from the medical management system and retrieve professional knowledge corresponding to the query instruction from the professional knowledge base according to the query instruction. The intelligent reply module is used to input the consultation message, customer health data and professional knowledge into the artificial intelligence model to generate initial reply content; The correction and verification module is used to correct and verify the initial response content according to the verification information in the professional knowledge base. If the verification passes and the inquiry message is identified as not belonging to the preset inquiry type, the final response content is generated. The manual processing module is used to transfer the consultation to manual processing if the verification fails or the consultation message is identified as belonging to a preset consultation type. After manual processing, the final reply content is generated and pushed to the customer.

Images