Methods, apparatuses, devices, media, and products for intelligent interaction

Abstract

The present disclosure relates to methods, apparatuses, devices, media and products for intelligent interaction. The method comprises: in response to a scanning operation on an identification code on a patient's auxiliary examination application form, opening a dialogue interface in an instant messaging application program for the user to interact with an auxiliary health intelligent agent, a plurality of auxiliary examination application forms in the same visit of the patient having the same identification code; the auxiliary health intelligent agent monitors or receives the result issuing state of the examination items associated with the identification code, and when one or more examination results are monitored or received, the one or more examination results are pushed to the dialogue interface; and displaying the interaction content between the auxiliary health intelligent agent and the user related to the one or more examination results in the dialogue interface, the interaction content including first interaction content, which includes the auxiliary health reference information sent by the auxiliary health intelligent agent to the user based on all the examination results in response to all the examination results of the examination items associated with the identification code having been issued.

Description

Technical Field

[0001] This disclosure relates to the field of intelligent interaction, and more specifically, to a method, apparatus, device, medium, and product for intelligent interaction. Background Technology

[0002] With the rapid development of medical informatization, electronic medical records (EMR) and medical information systems (HIS) have been widely used in medical institutions at all levels. However, in actual diagnosis and treatment, there are still many technical bottlenecks in the acquisition, management and utilization of patient data, making it difficult to meet patients' needs for continuity, convenience and intelligence in medical services.

[0003] First, data acquisition in the hospital's treatment process is fragmented and delayed. In modern medical models, doctors often order multiple tests for patients based on their condition, covering imaging (such as CT scans and MRIs), ultrasound (B-mode ultrasound), and laboratory tests (such as complete blood counts and comprehensive biochemical assays). These tests are usually performed by different departments or with different equipment, resulting in data generation times and storage in different subsystems within the hospital. Patients often need to repeatedly query data on the hospital system or self-service terminals, or wait for printed reports. This fragmented data acquisition method increases patients' waiting time and anxiety.

[0004] Secondly, health data spanning multiple institutions and time periods is difficult to aggregate into a continuous record. When patients visit different medical institutions at different times, their historical examination data is often isolated in the independent databases of each hospital. Existing technological solutions lack effective cross-institutional data aggregation mechanisms, resulting in fragmented patient health data. Doctors find it difficult to access long-term historical data to compare disease progression, and patients themselves cannot grasp long-term health trends. This is particularly detrimental to scenarios requiring continuous data support, such as chronic disease management and long-term follow-up.

[0005] Furthermore, data management for special groups faces barriers and obstacles. For groups such as the elderly and children who have difficulty operating smart devices independently, managing their health data presents a double challenge. On the one hand, the internet service authorization mechanisms across hospitals are inconsistent, making it cumbersome for family members to access information on behalf of others or for cross-hospital authorization. On the other hand, the technical terminology and interpretation of medical examination reports are often complex, making it difficult for non-professionals to understand the meaning of abnormal indicators. Existing technological solutions fail to provide intelligent, assisted interpretation and proactive push services, hindering this group from effectively utilizing their own health data for health management.

[0006] In summary, existing technologies lack a solution capable of automatically aggregating multi-source heterogeneous medical data, achieving cross-institutional data standardization and versioning management, and proactively serving patients through intelligent interaction. Therefore, a medical data interaction method based on an auxiliary health agent is urgently needed to address these issues. Summary of the Invention

[0007] To address the aforementioned technical issues, this disclosure provides a method for intelligent interaction. This method generates a QR code containing patient identity and context information on the test form, guiding the patient or their family to scan it via an instant messaging application to establish a connection with an auxiliary health agent. The auxiliary health agent synchronizes the test results to the dialogue interface in real time, interprets them in conjunction with historical health records, and constructs a health profile. Furthermore, identity authorization verification is implemented for scenarios where the user is not the account holder to ensure data compliance. After all test results are issued, the auxiliary health agent performs comprehensive analysis based on all data and historical records to generate supplementary opinions, thereby guiding the user to register for an appointment or providing reminders for medical visits. This achieves real-time push, intelligent interpretation, and closed-loop management of the entire medical process.

[0008] According to one aspect of this disclosure, a method for intelligent interaction is provided, the method comprising: in response to a scanning operation of an identification code on a patient's auxiliary examination request form, opening a dialogue interface in an instant messaging application for interaction between a user and an auxiliary health agent, wherein multiple auxiliary examination request forms of the patient during the same medical visit have the same identification code; the auxiliary health agent monitoring or receiving the result issuance status of examination items associated with the identification code, and pushing one or more examination results to the dialogue interface when monitoring or receiving the issuance of one or more examination results; and displaying in the dialogue interface interactive content between the auxiliary health agent and the user associated with one or more examination results, wherein the interactive content includes first interactive content, the first interactive content including first auxiliary health reference information sent by the auxiliary health agent to the user based on all examination results in response to the issuance of all examination results of the examination items associated with the identification code.

[0009] According to embodiments of this disclosure, the method further includes: performing an identity consistency check before pushing one or more examination results to the dialog interface, wherein, in response to the user's identity information being consistent with the patient's identity information, the assisting health agent pushes one or more examination results to the dialog interface; and in response to the user's identity information being inconsistent with the patient's identity information, the assisting health agent prompts the user to perform a patient authorization operation, and after the user completes the patient authorization operation, pushes one or more examination results to the dialog interface.

[0010] According to embodiments of this disclosure, patient authorization operations include at least one of SMS verification, facial biometric verification, electronic signature verification, guardianship relationship verification, or verification by inputting relevant information from the patient's identification documents.

[0011] According to embodiments of this disclosure, performing identity consistency verification before pushing one or more examination results to the dialog interface includes: when a user scans an identification code and adds an auxiliary health agent as a friend in an instant messaging application, determining whether the mobile phone number of the user bound to the instant messaging application is consistent with the mobile phone number stored in the patient's health record.

[0012] According to embodiments of this disclosure, performing identity consistency verification before pushing one or more examination results to the dialog interface includes: when a user scans an identification code and adds an auxiliary health agent as a friend in an instant messaging application, collecting at least one of the user's name and identification information, and determining whether the user's name and / or identification information matches the name and / or identification information of the patient associated with the identification code.

[0013] According to embodiments of this disclosure, performing identity consistency verification before pushing one or more inspection results to the chat interface includes: performing identity consistency verification when a user scans an identification code and adds the auxiliary health agent as a friend in an instant messaging application; or performing identity consistency verification when the auxiliary health agent monitors or receives one or more inspection results.

[0014] According to embodiments of this disclosure, the method further includes updating the patient's health record based on one or more examination results in a dialog interface and interactive content associated with the one or more examination results.

[0015] According to embodiments of this disclosure, the assistive health agent sends first assistive health reference information to the user based on all the patient's examination results and relevant information from the patient's health record.

[0016] According to embodiments of this disclosure, the identification code is associated with at least one of the patient's identity information and application scenario information, wherein the patient's identity information includes at least one of the patient's mobile phone number, name, and identification information, and wherein the application scenario information includes at least one of the department where the patient seeks medical treatment and one or more examination items.

[0017] According to embodiments of this disclosure, opening a dialogue interface in an instant messaging application for a user to interact with an auxiliary health agent includes: in response to a scan, based on the auxiliary health agent being a friend of the user, redirecting to a dialogue interface in the instant messaging application for interacting with the auxiliary health agent; and in response to a scan, based on the auxiliary health agent not being a friend of the user, redirecting to a request to add a friend interface in the instant messaging application for requesting to add the auxiliary health agent as a friend, and in response to the auxiliary health agent adding the user as a friend through the request to add a friend interface, redirecting to a dialogue interface in the instant messaging application for interacting with the auxiliary health agent.

[0018] According to embodiments of this disclosure, the interactive content further includes second interactive content, which includes second auxiliary health reference information related to each examination result sent to the user by the auxiliary health agent after each examination result is pushed.

[0019] According to embodiments of this disclosure, the interaction content further includes third interaction content and fourth interaction content. The third interaction content includes inquiry information related to the patient's condition sent by the assistive health agent, and the fourth interaction content includes reply information sent by the user in response to the inquiry information.

[0020] According to embodiments of this disclosure, the interaction content further includes fifth interaction content, which includes consultation assistance information sent by the health-assisting intelligent agent to assist patients in communicating with doctors.

[0021] According to another aspect of this disclosure, an apparatus for intelligent interaction is provided, the apparatus comprising: an interaction module configured to open a dialogue interface in an instant messaging application for user interaction with an auxiliary health agent in response to a scanning operation of an identification code on a patient's auxiliary examination request form, wherein multiple auxiliary examination request forms of the patient during the same medical visit have the same identification code; a result monitoring and push module configured to monitor or receive the result issuance status of examination items associated with the identification code, and push one or more examination results to the dialogue interface when monitoring or receiving the issuance of one or more examination results; and a display module configured to display in the dialogue interface interactive content between the auxiliary health agent and the user associated with one or more examination results, wherein the interactive content includes first interactive content, the first interactive content including first auxiliary health reference information sent by the auxiliary health agent to the user based on all examination results in response to the issuance of all examination results of the examination items associated with the identification code.

[0022] According to embodiments of this disclosure, the device further includes: an identity verification module configured to perform an identity consistency verification before pushing one or more examination results to the dialog interface, wherein, in response to the user's identity information being consistent with the patient's identity information, the assisting health agent pushes one or more examination results to the dialog interface; and in response to the user's identity information being inconsistent with the patient's identity information, the assisting health agent prompts the user to perform a patient authorization operation, and after the user completes the patient authorization operation, pushes one or more examination results to the dialog interface.

[0023] According to embodiments of this disclosure, the device further includes an archiving module configured to update a patient's health record based on one or more examination results in a dialog interface and interactive content associated with the one or more examination results.

[0024] According to another aspect of this disclosure, a device for intelligent interaction is provided, the device comprising: a processor and a memory storing computer-executable instructions, which, when executed by the processor, cause the processor to perform the aforementioned method for intelligent interaction.

[0025] According to another aspect of this disclosure, a computer-readable recording medium is provided that stores computer-executable instructions, wherein the computer-executable instructions, when executed by a processor, cause the processor to perform the above-described method for intelligent interaction.

[0026] According to another aspect of this disclosure, a computer program product is provided, the computer program product including computer-executable instructions, wherein the computer-executable instructions, when executed by a processor, cause the processor to perform the above-described method for intelligent interaction.

[0027] Therefore, the methods, apparatus, devices, media, and products for intelligent interaction according to embodiments of this disclosure associate multiple examination items from a single medical visit using same-code binding technology, and combine this with a result readiness detection mechanism to achieve instant push of examination results, significantly reducing the cost of repeated inquiries for patients. Furthermore, identity consistency verification supports guardians or family members to handle matters on behalf of patients while ensuring the compliance of medical data. Structured archiving and cross-hospital data aggregation technologies are used to construct long-term, continuous health records to improve the efficiency of subsequent follow-up visits. In addition, task orchestration automatically guides patients to follow-up appointment registration and provides reminders of important information, thereby reducing patient anxiety and comprehensively improving the efficiency of the medical process. Attached Figure Description

[0028] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are merely some exemplary embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0029] Figure 1 A flowchart of a method for intelligent interaction according to an embodiment of the present disclosure is shown;

[0030] Figure 2 A schematic diagram of a system architecture according to an embodiment of the present disclosure is shown;

[0031] Figure 3 A schematic diagram of the interaction flow according to an embodiment of the present disclosure is shown;

[0032] Figure 4 A schematic diagram of a multi-item aggregation process with the same code according to an embodiment of the present disclosure is shown;

[0033] Figure 5 A schematic diagram of a non-personalized QR code authorization process according to an embodiment of the present disclosure is shown;

[0034] Figure 6 A schematic diagram of a result readiness detection and push process according to an embodiment of the present disclosure is shown;

[0035] Figure 7 A schematic diagram illustrating the results of structured archiving of health records and cross-hospital comparison according to an embodiment of the present disclosure is shown.

[0036] Figure 8 A block diagram of an apparatus for intelligent interaction according to an embodiment of the present disclosure is shown;

[0037] Figure 9 A block diagram of a device for intelligent interaction according to an embodiment of the present disclosure is shown;

[0038] Figure 10 A schematic diagram of a recording medium according to an embodiment of the present disclosure is shown. Detailed Implementation

[0039] To make the objectives, technical solutions, and advantages of this disclosure more apparent, exemplary embodiments according to this disclosure will now be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this disclosure, and not all embodiments of this disclosure. It should be understood that this disclosure is not limited to the exemplary embodiments described herein.

[0040] In this specification and accompanying drawings, substantially the same or similar steps and elements are indicated by the same or similar reference numerals, and repeated descriptions of these steps and elements will be omitted. Furthermore, in the description of this disclosure, the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance or order.

[0041] In this specification and accompanying drawings, elements are described in singular or plural forms according to embodiments. However, the singular and plural forms have been suitably chosen for the presented cases merely for ease of explanation and are not intended to limit this disclosure. Thus, a singular form may include a plural form, and a plural form may include a singular form, unless the context clearly indicates otherwise.

[0042] When visiting a hospital, doctors often order various tests. After each test, patients need to wait a sufficient amount of time to view the results on the hospital's equipment. However, some patients may not have enough time to wait for the results before leaving the hospital, resulting in delays and inconvenience in obtaining the results. Furthermore, after receiving the results, patients usually need to wait a certain amount of time before seeing a doctor. For emergency patients, this can lead to missing the optimal treatment window or causing prolonged anxiety. Moreover, if a patient visits different hospitals multiple times over a long period, it is difficult to effectively save these test results from different times and hospitals, making it very inconvenient to access them later and hindering doctors' comprehensive assessment of the patient's condition. In addition, for elderly patients or children who lack the ability to independently operate electronic devices, obtaining and managing test results is even more inconvenient: different hospitals require different user authorizations to access the elderly or children's data; and even after obtaining the results, timely and professional interpretation of the results is impossible.

[0043] To address one or more of the aforementioned issues, this disclosure provides a method for intelligent interaction. This method uses a code-binding technology to associate multiple examination items from a single medical visit, combined with a result readiness detection mechanism, to achieve instant push of examination results, significantly reducing the cost of repeated inquiries for patients. Furthermore, this method performs a comprehensive analysis of the patient's condition based on all examination results, providing supplementary health advice. In addition, through identity consistency verification, it supports guardians or family members acting on behalf of patients while ensuring the compliance of medical data. Utilizing structured archiving and cross-hospital data aggregation technologies, it constructs long-term, continuous health records to improve the efficiency of subsequent follow-up visits.

[0044] The following will describe in detail the content related to the method for intelligent interaction provided in this disclosure with reference to the accompanying drawings.

[0045] Figure 1 This is a flowchart illustrating a method for intelligent interaction according to an embodiment of the present disclosure.

[0046] As an example, this method can be executed by any suitable processor or processing unit. The processor or processing unit can be located on any suitable user-specified device, such as a user's mobile device. The processor or processing unit can implement the method in software, in hardware, or any combination of software and hardware.

[0047] Reference Figure 1 In step S110, in response to the scanning operation of the identification code on the patient's auxiliary examination request form, a dialogue interface for the user to interact with the auxiliary health agent can be opened in the instant messaging application, wherein multiple auxiliary examination request forms of the patient in the same medical visit have the same identification code.

[0048] As an example, auxiliary examination request forms include various types of forms such as laboratory tests, imaging tests, functional tests, endoscopy, and pathology. Common examples include laboratory test request forms such as complete blood count, urinalysis, complete biochemistry panel, coagulation function test, and infectious disease screening; imaging test request forms such as chest X-ray, CT scan, MRI, ultrasound, and bone density test; functional test request forms such as electrocardiogram, Holter monitoring, pulmonary function test, and electroencephalogram; endoscopic test request forms such as gastroscopy, colonoscopy, and laryngoscopy; and pathological test request forms such as puncture biopsy and postoperative specimen submission.

[0049] As an example, the identification code may include a QR code or any other suitable code. This identification code can be placed in a suitable location on the auxiliary inspection request form, allowing the user to scan it.

[0050] According to embodiments of this disclosure, the identification code may be associated with at least one of patient identity information and application scenario information, wherein the patient identity information includes at least one of the patient's mobile phone number, name, and identification information, and wherein the application scenario information includes at least one of the department where the patient seeks medical treatment and one or more examination items.

[0051] As an example, the patient's identity information may specifically include one or more of the patient's mobile phone number, real name, ID card number, medical insurance information, etc. The application scenario information specifically covers the clinical department where the patient is visiting, such as internal medicine, surgery, laboratory, radiology, etc., as well as one or more auxiliary examination items such as blood routine, CT imaging, ultrasound examination, electrocardiogram, pathological biopsy, etc.

[0052] In addition, this disclosure utilizes the same code multi-item aggregation technology (as will be combined later). Figure 4 As described, this technology enables the establishment of a unified identifier across "visit event - examination item - result document - conversation interface" in scenarios involving multiple examinations during a single visit, facilitating result aggregation and delivery. Applying this technology to such scenarios, by using the same identifier for multiple examination items within the same visit, not only achieves automatic aggregation and intelligent delivery of multimodal examination results, ensuring patients can promptly obtain complete examination results on their mobile devices, but also significantly improves diagnostic and treatment efficiency and optimizes the patient experience.

[0053] As an example, the instant messaging application can include any suitable existing instant messaging application, such as WeChat, DingTalk, etc. The method provided in this disclosure is convenient because it allows users to interact with the user by redirecting them to the instant messaging application rather than downloading the application.

[0054] It should be noted that using an instant messaging application as the interaction carrier for the auxiliary health intelligent agent in step S110 effectively addresses the technical pain points of insufficient convenience of service interaction and low user stickiness in existing technologies. Specifically, in existing technologies, similar service tools such as Ant AQ (formerly AQ, an AI health application) exist as independent apps or mini-programs, and the auxiliary health intelligent agent interaction scenarios they provide, such as consultation dialogues, are limited to within the independent app or mini-program. These independent apps or mini-programs are usually low-frequency applications used by users. When users have service needs, they often find it difficult to remember the existence of the application tool, and each use requires actively searching for and opening the independent application, which is cumbersome. This not only reduces users' willingness to use the service but also results in a low degree of relevance between users and service products, making it impossible to achieve efficient service access.

[0055] This disclosure embeds the health-assistant intelligent agent interaction service into frequently used instant messaging applications such as WeChat and DingTalk, offering significant technical advantages over existing technologies. Firstly, instant messaging applications are frequently used tools, with users opening and using them multiple times daily. Embedding the service significantly enhances the user's connection to the application, enabling proactive service delivery and preventing user forgetting due to infrequent application use. Secondly, the core function of instant messaging applications is chat interaction. Designing the health-assistant intelligent agent's interface to fit chat scenarios perfectly aligns with users' long-established chat habits. Users can quickly learn and interact with the health-assistant intelligent agent without needing to learn new operational logic, effectively reducing operational costs and significantly improving the user experience.

[0056] As an example, instant messaging applications may include any suitable existing high-frequency instant messaging applications, such as QQ and WeChat Work, in addition to WeChat and DingTalk mentioned above. As long as they can achieve the embedding and redirection of the auxiliary health intelligent agent interaction interface, they are all within the protection scope of this disclosure.

[0057] In summary, by redirecting users to an instant messaging application to interact with the health-assistant agent, rather than requiring users to download and install a separate application, this method not only solves the technical problems of cumbersome user experience and low relevance in existing technologies, but also further enhances the interactive experience through an interface that aligns with user habits, making the method disclosed in this publication significantly practical.

[0058] As an example, an intelligent agent is an intelligent, autonomous entity possessing the capabilities of perception, memory, planning, tool invocation, and autonomous reasoning and execution. Its core foundation is a large language model. Under a general architecture, an intelligent agent is not merely capable of simple dialogue using a large model, but rather an intelligent system that autonomously completes complex tasks, with the large model as its core brain and multiple layers of capability modules superimposed. An intelligent agent typically includes a large model, a perception module, a planning module, a tool invocation module, and a memory module.

[0059] Specifically, the large model serves as the core foundation and reasoning hub of the entire intelligent agent, undertaking the core roles of natural language understanding, semantic parsing, logical reasoning, clinical knowledge deduction, multi-turn dialogue generation, and task intent recognition, providing underlying cognitive and decision-making support for the other modules. The perception module receives multi-source inputs from external sources, including text interactions, structured data, interface messages, and device push notifications, and then the large model performs semantic understanding and information extraction. The memory module, relying on a vector database and conversation memory mechanism, stores the dialogue context. The large model retrieves the memory context and integrates it into subsequent reasoning and response generation, achieving vertical association and continuous interaction. The planning module, led by the large model, decomposes tasks and orchestrates processes. For complex objectives such as multi-turn dialogues, information aggregation, and content generation, it autonomously breaks them down into step-by-step sub-tasks, determining the processing order and interaction logic. The tool invocation module, under the semantic decision-making and instruction issuance of the large model, calls external tools such as RAG historical retrieval, knowledge graph query, feature extraction, and cross-system interface data retrieval as needed. This breaks through the knowledge boundaries of the large model itself, obtains real-time, personalized, and professional domain external data, and supports the intelligent agent to complete business processes beyond pure dialogue capabilities.

[0060] The auxiliary health intelligence agent involved in this application, which is used in the medical field, is designed to meet the professional, compliant, and clinically adaptable needs of the medical field. It makes targeted modifications, optimizations, and additions to each module of the large model, while strengthening the medical field capabilities of the large model, and ultimately realizes the full-chain capability of "interpreting test results - multi-round consultation dialogue - auxiliary health information reference - health management planning".

[0061] First, specific optimizations are made to the core foundation (large model) for the medical field:

[0062] The large-scale model of the general-purpose intelligent agent is based on general semantic understanding and basic reasoning. The auxiliary health intelligent agent, however, needs to undergo deep adaptation of the large-scale model to the medical field, enabling it to possess medical-grade cognitive and reasoning capabilities. This is the foundation for all medical-related capabilities, specifically including:

[0063] Evidence-based fine-tuning in the medical field: Abandoning general and messy internet data, we use millions of real clinical cases, authoritative medical textbooks, and clinical practice guidelines (such as NCCN recommended protocols) as exclusive training data. Through supervised fine-tuning (SFT), we optimize large models to ensure their accurate understanding of medical terminology, examination indicators, and clinical logic. The output content strictly conforms to clinical norms and avoids non-professional expressions and compliance risks.

[0064] Enhanced Clinical Thinking Modeling: By modeling clinical logic through reinforcement learning (RLHF), the large model can simulate the standardized consultation path of doctors: "chief complaint → present illness history → past medical history → accompanying symptoms". It has the ability to actively and progressively ask follow-up questions (such as asking about the nature of symptoms, duration, and triggering factors), rather than simply responding passively.

[0065] Compliance boundary embedding: Medical compliance rules are embedded in the reasoning logic of the large model to explicitly prohibit the generation of content beyond the scope of assistance, such as disease diagnosis conclusions and prescription plans. All outputs are limited to "reference interpretation, risk warning, and medical treatment guidance" to ensure compliance with medical data privacy and treatment standards.

[0066] Secondly, the sensing module is specifically adapted for the medical field:

[0067] The perception module of the general-purpose intelligent agent mainly processes general text and simple commands. The auxiliary health intelligent agent needs to modify its perception module to suit the input characteristics of medical scenarios, ensuring accurate capture and parsing of medical-related information. Specifically, this includes:

[0068] Medical data parsing and adaptation: Optimize the text extraction and structuring capabilities of the perception module to accurately identify and parse various test reports (blood routine, biochemistry, CT, MRI, etc.) input by users, automatically extract core fields such as test items, values, units, reference ranges, and abnormal markers, and support the text conversion of image reports (after extracting image features with CV tools, the perception module passes them to the large model).

[0069] Multi-source medical input compatibility: The input interface of the extended sensing module supports connection to medical systems such as LIS / PACS / HIS, ​​and can directly receive structured medical data and examination results pushed by the device. At the same time, it is compatible with unstructured information such as test values, symptom descriptions, and consultation responses manually entered by users, so as to realize the unified reception and preliminary analysis of multi-source medical information.

[0070] Medical semantic error correction and normalization: The new medical terminology normalization capability maps non-standard expressions input by users (such as "platelet", "PLT", "platelet count") to standard medical terms, while correcting typos and expression deviations in the input (such as incorrect numerical units) to ensure that the information passed to the large model is accurate.

[0071] Secondly, the memory module is specifically modified for the medical field:

[0072] The memory modules of general-purpose intelligent agents primarily store short-term session information. Assistive health intelligent agents need to enhance the long-term storage, accurate retrieval, and adaptation to medical scenarios of their memory modules, supporting personalized consultations and historical data comparisons. Specific tasks include:

[0073] Personal health record storage: For example, using the patient's unique identifier (PatientID) as an index, the storage structure of the memory module is modified to store the user's long-term health data, including the results of each examination, historical consultation records, basic medical history, medication history, allergy history, and previous auxiliary health reference information, forming a complete personal health record that can be retrieved by the large model at any time.

[0074] Medical data vectorization and precise retrieval: By deeply integrating the memory module with the vector database, stored health records and historical examination data are vectorized. This allows large models to quickly retrieve historical data of similar examinations through RAG longitudinal historical retrieval technology, enabling comparison of indicator trends (such as cross-cycle changes in platelet counts), and providing support for diagnosis and auxiliary interpretation.

[0075] Memory Priority Ranking: Optimize the calling logic of the memory module, with the large model leading the priority ranking of memory content, and prioritize the retrieval of memory information related to the current test results and consultation scenario (such as prioritizing the retrieval of historical liver function data and liver disease history when interpreting liver function indicators), to ensure that the memory content is accurately matched with the business scenario.

[0076] Then, for the medical field, the tool calling module is extended specifically for medical use:

[0077] General-purpose intelligent agents primarily focus on using general-purpose tools (such as search and computation). Assistive health intelligent agents require the addition and adaptation of medical-specific tools, with a large model leading the tool invocation decisions to handle complex tasks in medical scenarios, specifically including:

[0078] New medical-specific tools have been integrated: RAG historical retrieval tool (for retrieving personal historical examination data), medical knowledge graph query tool (for associating the correspondence between indicators, symptoms, and clinical guidelines), and CV image feature extraction tool (for parsing CT, chest X-ray, and other image data to extract lesion features). All tools are linked with the large model through standardized interfaces, and the large model autonomously decides whether to call them and the order of calling them based on task requirements (such as report interpretation and consultation completion).

[0079] Medical system interface adaptation: The modified tool calling module supports interface with hospital LIS / PACS / HIS and other systems. The large model can issue commands to retrieve real-time examination data, historical medical records and other information through the tool calling module, eliminating the need for manual input by the user and improving interaction efficiency and data accuracy.

[0080] Tool call logic optimization: Based on the big model and medical clinical logic, the trigger conditions and process of tool calls are optimized. For example, when interpreting an image report, the CV tool is automatically called to extract lesion features, and then the medical knowledge graph tool is called to match the clinical references corresponding to the features. Finally, combined with the historical data of the memory module, complete auxiliary interpretation content is generated.

[0081] In addition, the planning module has been upgraded specifically for the medical field:

[0082] The general-purpose intelligent agent's planning module primarily handles simple task decomposition. The auxiliary health intelligent agent requires an upgraded planning module to enable it to autonomously plan the entire task chain—"test result interpretation—consultation—auxiliary health information reference—health management"—based on the core needs of the medical scenario. Specifically, this includes:

[0083] Optimization of Medical Task Decomposition Capabilities: Led by a large model, the complex core task of "generating health management plans based on test results" is broken down into step-by-step executable sub-tasks: First, structured interpretation and anomaly identification of test results; second, based on abnormal indicators / test results, initiating multiple rounds of consultations using simulated clinical thinking to supplement information such as symptoms and medical history; third, integrating test results, consultation information, and historical data to generate auxiliary health reference information; and finally, based on the auxiliary health reference information, planning short-term and long-term health management plans (follow-up recommendations, lifestyle adjustments, and medical consultation guidance, etc.).

[0084] Autonomous planning of the consultation process: The large model is linked to the planning module, which autonomously plans the consultation path based on the abnormality of the current test results. For example, when an elevated white blood cell count is detected, the consultation sequence is planned as "chief complaint → fever symptoms → duration of fever → accompanying symptoms → medication history", and the follow-up questions are dynamically adjusted to ensure that the consultation information can accurately support the subsequent generation of auxiliary health reference information and health management plans.

[0085] Health management planning adaptation: The planning module combines the clinical knowledge reasoning ability of the big model to autonomously plan personalized health management plans based on the test results, medical history and symptoms of different users, and specify the follow-up time, follow-up items, lifestyle adjustment suggestions (such as diet and exercise) and emergency medical treatment indications. At the same time, the big model can drive the planning module to dynamically update the health management plan based on the user's subsequent test results and consultation feedback.

[0086] In summary, the core difference between general-purpose intelligent agents and health-assisted intelligent agents lies in their "medical adaptation"—the modification of all modules, the integration of tools, and the planning of processes are all centered around medical clinical standards and user health needs. The realization of all this relies on the fine-tuning of a large-scale model within the medical field and the modeling of clinical thinking. The large-scale model endows the health-assisted intelligent agent with medical-grade cognitive and reasoning capabilities, while each dedicated module applies these capabilities to specific medical scenarios such as "test interpretation, consultation, auxiliary health information reference, and health management." Ultimately, this results in a professional, autonomous intelligent agent capable of automatically performing structured interpretation of medical reports, vertical retrieval of historical data, multi-round consultations simulating clinical thinking, multi-source data aggregation and analysis, and automatic generation of auxiliary health reference information and / or doctor-patient communication assistance information.

[0087] As an example, an assistive health intelligent agent can refer to an intelligent functional entity that can achieve a closed-loop process of "perceiving the external environment - processing information - generating decisions - executing responses - receiving feedback - iterative optimization" through preset or adaptive information interaction mechanisms, decision-making logic, and action execution capabilities to achieve specific task objectives or adapt to dynamic scene changes. This entity is not limited to hardware, software, or a combination of both. Its core components may include, but are not limited to, an information perception module (used to acquire external or internal information such as environmental data, user commands, and system status; perception methods include, but are not limited to, sensors, data interfaces, signal receiving devices, text / voice / image recognition, etc.). The system consists of a sensory information processing and decision-making module (used to analyze, calculate, reason, or learn from the perceived information; the decision logic can be constructed based on preset rules, algorithm models (such as machine learning, deep learning, reinforcement learning, logical reasoning algorithms, etc.) or a combination of both), an action execution module (used to transform the decision results into actionable responses; execution methods include, but are not limited to, controlling the operation of hardware devices, outputting data / text / voice / image information, adjusting system parameters, triggering process instructions, etc.), and an optional feedback iteration module (used to receive the result information after the action execution and optimize the perception strategy, decision logic, or execution method based on this information to improve the accuracy of task achievement or environmental adaptability).

[0088] According to embodiments of this disclosure, opening a dialog interface in an instant messaging application for a user to interact with an auxiliary health agent may include: in response to a scan, based on the user's friendship with the auxiliary health agent, redirecting to a dialog interface in the instant messaging application for interaction with the auxiliary health agent.

[0089] As an example, a friend relationship between an assistive health agent and a user may include: the assistive health agent having obtained the user's identification information (such as an ID); and the assistive health agent communicating with the user via the user's identification information to interact with content. A friend relationship between the assistive health agent and the user indicates an authenticated / bound user relationship. This suggests that the user's interaction with the assistive health agent is not a first-time one.

[0090] According to embodiments of this disclosure, opening a dialogue interface in an instant messaging application for a user to interact with an auxiliary health agent may include: in response to a scan, based on the fact that the auxiliary health agent and the user are not friends, redirecting to a request to add the auxiliary health agent as a friend in the instant messaging application; and in response to the auxiliary health agent adding the user as a friend through the request to add the auxiliary health agent as a friend, redirecting to the dialogue interface in the instant messaging application for interaction with the auxiliary health agent.

[0091] As an example, if the health assistant and the user are not friends, it indicates that this is the first interaction between the health assistant and the user. In this case, the user can first be redirected to the friend request screen in the instant messaging application to request to add the health assistant as a friend. In this friend request screen, the user can request to add the health assistant as a friend. Then, in response to the health assistant and the user becoming friends via the friend request screen, the user is redirected to the user interface in the instant messaging application where they can interact with the health assistant.

[0092] As can be seen, the method provided in this disclosure can implement different interface display processes for first-time and non-first-time interactions, thereby facilitating better subsequent interactions with users.

[0093] In step S120, the assistive health agent can monitor or receive the result issuance status of the inspection items associated with the identification code, and push one or more inspection results to the dialogue interface when monitoring or receiving one or more inspection results.

[0094] As an example, as will be in the following Figure 6 The description of assisting health intelligent agents in monitoring or receiving the result issuance status of examination items associated with identification codes in medical information systems may include: monitoring or receiving the result issuance status of examination items by listening to result issuance events in medical information systems, polling the result status of examination items, receiving server message pushes, calling preset callback interfaces, subscribing to business event messages, and capturing database data changes.

[0095] According to embodiments of this disclosure, an identity consistency check can be performed before pushing one or more examination results to the dialog interface. Specifically, in response to the user's identity information matching the patient's identity information, the assistive health agent pushes one or more examination results to the dialog interface; and in response to the user's identity information not matching the patient's identity information, the assistive health agent prompts the user to perform a patient authorization operation, and after the user completes the patient authorization operation, pushes one or more examination results to the dialog interface (as follows). Figure 5 (will be described).

[0096] As an example, when the current user's identity information is verified to be completely consistent with the patient's name, mobile phone number, or identification information, the health-assisting AI can directly push one or more examination results, such as blood routine tests, CT images, and ultrasound reports, to the dialogue interface for the user to view. If the verification determines that the current user's identity information does not match the patient's identity information, such as in a scenario where a family member or agent is viewing someone else's examination report, the health-assisting AI will immediately prompt the user to initiate and complete the patient authorization process. After the user completes the patient authorization online and the identity verification is successful, the health-assisting AI will then push the corresponding examination results to the dialogue interface for display, thereby ensuring the access security and privacy compliance of medical examination data.

[0097] According to embodiments of this disclosure, patient authorization operations may include at least one of the following: SMS verification, facial biometric verification, electronic signature verification, guardianship relationship verification, or verification by inputting relevant information from the patient's identification documents.

[0098] As an example, patient authorization can be completed using at least one of several compliant verification methods. For instance, it could involve sending a verification code to the patient's registered mobile phone number for SMS verification, collecting and comparing the patient's facial image for facial biometric verification, having the patient sign an electronic signature online for electronic signature verification, verifying kinship and guardianship relationships for minors or patients without independent capacity for action, or having the operator manually enter the patient's ID card, medical card number, or other relevant document information for identity matching verification. By combining any one or more of the above methods to complete the authorization verification, the user can legally obtain the right to view the patient's examination results.

[0099] According to embodiments of this disclosure, performing identity consistency verification before pushing one or more examination results to the dialog interface may include: when a user scans an identification code and adds an auxiliary health agent as a friend in an instant messaging application, determining whether the mobile phone number of the user bound to the instant messaging application is consistent with the mobile phone number stored in the patient's health record.

[0100] As an example, when a user (such as the patient or their family member) scans the patient's identification code (such as a medical appointment QR code) with their mobile phone and adds the auxiliary health agent as a friend in instant messaging (such as WeChat or DingTalk), the system automatically triggers verification logic. It extracts the mobile phone number bound to the currently logged-in account in the instant messaging application and compares it with the reserved mobile phone number stored in the patient's health record database. Taking the scenario of an "elderly patient's follow-up visit" as an example, if the patient's children scan the identification code and add the auxiliary health agent as a friend, and the system detects that the mobile phone number bound to the children's account matches the mobile phone number registered in the patient's record, it directly passes the verification and establishes a data channel, allowing the auxiliary health agent to subsequently push examination results. Conversely, if the scanner uses an unrelated person's mobile phone number, the system will determine that the identity is inconsistent and refuse to establish an association. Thus, at the very first moment the user establishes a connection with the auxiliary health agent, the system utilizes the inherent attributes of the communication account to accurately verify the visitor's identity, effectively preventing unauthorized access.

[0101] According to embodiments of this disclosure, performing identity consistency verification before pushing one or more examination results to the dialog interface may include: when a user scans an identification code and adds an auxiliary health agent as a friend in an instant messaging application, collecting at least one of the user's name and identification information, and determining whether the user's name and / or identification information matches the name and / or identification information of the patient associated with the identification code.

[0102] As an example, when a user scans a patient identification code and adds the auxiliary health agent as an instant messaging friend, the system prompts the user to enter at least one of their real name and ID number. This input information is then matched against the patient's name and ID information associated with the identification code. Taking the scenario of "family members checking medical records on behalf of patients in other locations" as an example, if the patient is unable to operate the system, their children, when adding the auxiliary health agent as a friend by scanning the identification code, need to enter "Zhang San" and ID number "110101...". The system then compares the "Zhang San" associated with the identification code with the corresponding ID number. If the information matches perfectly, the verification is successful, a data channel is established, and subsequent test results are allowed. Conversely, if the input information does not match the records, the system will intercept the data and indicate an identity mismatch. This rigorous verification of the user's biometrics or legal identity information ensures that only authorized personnel with verified identities can access the patient's medical data.

[0103] According to embodiments of this disclosure, performing identity consistency verification before pushing one or more inspection results to the chat interface may include: performing identity consistency verification when a user scans an identification code and adds the auxiliary health agent as a friend in an instant messaging application; or performing identity consistency verification when the auxiliary health agent monitors or receives one or more inspection results.

[0104] As an example, in the identity consistency verification loop of pushing examination results to the dialogue interface, the method disclosed herein supports flexible triggering timing to adapt to different business scenarios: On the one hand, verification can be embedded in the initial stage of user interaction, that is, it is executed immediately when the user scans the patient identification code and adds the auxiliary health agent as an instant messaging friend. For example, when a patient's family member scans the medical appointment QR code to add a friend, the system immediately compares the account's real-name information with the patient's file. If it fails, the subsequent connection is blocked. On the other hand, verification can also be deployed in the real-time stage of data flow, that is, it is triggered when the auxiliary health agent monitors or receives the generation of the examination result report. For example, when an emergency examination report is issued, the system automatically verifies whether the identity of the user currently communicating with the auxiliary health agent is consistent with the patient associated with the report. If it is found that an unauthorized person is trying to view the report, the push is immediately intercepted and a prompt is made to perform the patient authorization operation. This "pre-binding verification" and "in-process dynamic verification" can be used alone or in combination, so that when a user initiates access at the beginning of the connection or when the result is generated, the patient's private medical data can be strictly protected.

[0105] In step S130, the interactive content between the auxiliary health agent and the user associated with one or more examination results can be displayed in the dialogue interface. The interactive content includes first interactive content, which includes first auxiliary health reference information sent by the auxiliary health agent to the user based on all examination results in response to the issuance of all examination results of the examination items associated with the identification code.

[0106] As an example, the health-assisting agent not only displays raw data in the dialogue interface but also presents interactive content generated based on the patient's examination results. Specifically, once the system has monitored and completed the review and issuance of all examination items associated with the patient's identification code (such as complete blood count, CT images, and biochemical indicators), the health-assisting agent will automatically trigger aggregate analysis to generate a comprehensive interpretation report based on all examination results and push it to the dialogue interface as a message. Taking the "acute appendicitis screening" scenario as an example, when the patient's complete blood count, white blood cell count, abdominal CT images, and C-reactive protein test results all show abnormalities and are ready, the health-assisting agent will not simply list the data but will send a "first auxiliary health reference information" message in the dialogue window containing "suspected acute appendicitis, immediate surgical consultation recommended," along with correlation analysis of abnormal values ​​for each indicator and risk warnings. This helps users intuitively obtain intelligently integrated treatment suggestions in the dialogue interface, rather than facing fragmented raw data.

[0107] According to embodiments of this disclosure, the interactive content may further include second interactive content, which includes second auxiliary health reference information related to each examination result sent to the user by the auxiliary health agent after each examination result is pushed.

[0108] As an example, the method disclosed herein can also assist the health agent in generating and sending "secondary auxiliary health reference information" for each individual test result after it is pushed. Taking the scenario of "step-by-step examination for chest pain patients" as an example, when the patient's first test result, "myocardial enzyme spectrum," is first issued and pushed to the dialogue interface, the auxiliary health agent then sends "secondary auxiliary health reference information," indicating that "troponin levels are slightly elevated, and the risk of myocardial damage should be alerted in conjunction with clinical symptoms." Subsequently, when the second result, "ECG electrocardiogram," is pushed, it is further supplemented with "The ECG shows ST segment depression, and it is recommended to prioritize the investigation of angina pectoris." This mode of interpreting each result in real time allows users to obtain real-time professional guidance for each abnormal or normal result during the examination process without waiting for all reports to be summarized, significantly improving the timeliness and guidance value of diagnostic and treatment information.

[0109] As an example, in the aforementioned interactive mode of real-time interpretation of each result, the auxiliary health agent can automatically interpret each test result and push the analysis results to the dialogue interface after pushing each test result. Optionally, after the auxiliary health agent pushes each test result, it will not proactively push the analysis results. Instead, it will only push the analysis results to the dialogue interface based on the user's indication in the interactive content pushed in the interactive interface that analysis results are required.

[0110] By providing this flexible interactive mode that combines "automatic interpretation" and "on-demand interpretation," the automatic mode enhances the timeliness and proactivity of information acquisition. For example, in emergency situations or when a patient's condition is critical, the system's automatic push of interpretations ensures that users receive key risk warnings (such as critical value alerts) immediately, without requiring users to have professional medical knowledge to actively inquire. This greatly reduces the risk of misjudgment due to information lag, reflecting the "proactive care" of the health-assisting intelligent agent. In the on-demand mode, the user experience and interface simplicity are optimized. For example, in routine examinations or scenarios where users already possess a certain level of judgment, on-demand interpretation avoids the dialogue interface being overwhelmed by a large amount of redundant information, maintaining a clean and focused interaction. At the same time, it gives users the initiative in interpretation, respecting the personalized needs of different users for information depth and timing, and reducing the anxiety caused by information overload.

[0111] According to embodiments of this disclosure, the interactive content may further include third interactive content and fourth interactive content. The third interactive content includes inquiry information related to the patient's condition sent by the assistive health agent, and the fourth interactive content includes reply information sent by the user in response to the inquiry information.

[0112] As an example, after analyzing examination results or combining them with past medical history, the health-assisting intelligent agent initiates a "third interaction," which involves sending users inquiries about the patient's current condition to supplement key clinical details needed to provide auxiliary health reference information. Taking the scenario of "fever of unknown cause" as an example, when the health-assisting intelligent agent detects that the patient's blood routine shows elevated white blood cell count but lacks specific clues about the source of infection, it will push an inquiry into the dialogue interface: "Has the patient recently experienced accompanying symptoms such as chills, cough, or rash?" Subsequently, the user (such as a family member) replies with "fourth interaction," namely, "The patient experienced chills yesterday, accompanied by a mild dry cough." Based on this new feedback, the health-assisting intelligent agent then dynamically adjusts the logic for providing auxiliary health reference information and updates its auxiliary suggestions. This real-time interactive mechanism of "intelligent inquiry—user feedback" effectively compensates for the limitations of static examination data, making the provided auxiliary health information more accurate, comprehensive, and relevant to the patient's actual condition.

[0113] According to embodiments of this disclosure, the interactive content may further include fifth interactive content, which includes consultation assistance information sent by the health-assisting intelligent agent to assist patients in communicating with doctors.

[0114] As an example, in the scenario of influenza A / upper respiratory tract infection, a patient's test results show a positive result for influenza A. Based on this test result, the health-assisting AI, combined with information obtained through inquiries in the interactive interface (highest fever temperature, onset time, cough / sore throat, dyspnea, underlying conditions (asthma / cardiovascular), medication history, allergy history, pregnancy / postpartum status, etc.), generates a "communication question list (for the patient)" and a "visit summary (for the doctor)" for communication with the doctor. For example, the communication question list for the patient might include: Am I at high risk and need to start antiviral medication (such as oseltamivir) as early as possible? Has the optimal window for medication been missed? Do I currently need further testing (complete blood count / chest X-ray / blood oxygen monitoring) to rule out pneumonia or complications? How should I choose, what is the dosage and interval for antipyretics (acetaminophen / ibuprofen)? Does it conflict with my existing chronic disease medications? What symptoms require immediate follow-up / emergency care (e.g., dyspnea, persistent high fever, decreased blood oxygen, etc.)? If there are elderly people / children at home, are preventative measures or isolation recommendations necessary? A summary of your medical visit for your doctor's reference may include: Chief complaint: Fever and cough for 2 days, highest temperature 39.2℃; Tests: Test report shows positive for influenza A; Symptoms: Significant cough / sore throat, no chest pain; Shortness of breath; Blood oxygen saturation (not measured / 98%); Past medical history: Mild asthma, denies immunosuppression; Drug allergy: Penicillin (rash); Previous medication: Acetaminophen 0.5g, short-lived fever reduction; Concerns: Whether early antiviral treatment is needed; Whether pneumonia / complications need to be ruled out; Interactions between antipyretics / cough suppressants and asthma medications.

[0115] In other words, the health-assisting intelligent agent can, based on the examination results and relevant information extracted from the interactive interface, help users prepare a list of questions to discuss with doctors in advance, and even directly generate a summary of the consultation information for doctors, so that doctors and patients can have efficient and high-quality communication within a limited time.

[0116] According to embodiments of this disclosure, a patient's health record can be constructed based on the patient's basic information data, wherein the basic information data includes the patient's identity and name and demographic information; and the patient's health record can be updated based on one or more examination results in a dialog interface and interactive content associated with one or more examination results.

[0117] As an example, the method disclosed herein first uses the patient's identification (such as ID card number or medical card number) as the core, integrating demographic information such as name, age, and gender to construct an initial personal health record. Subsequently, by dynamically monitoring the interactive content in the dialogue interface, it parses and extracts the "one or more test results" (such as a positive H1N1 report, blood routine values) and their associated "interactive content" (such as the health assistant's inquiries about symptoms, the patient's responses regarding medication history, etc.), automatically merging and updating this dynamically generated clinical data with the initial record. Taking a "flu patient follow-up visit" as an example, when the patient completes antigen testing and replies with information such as "history of asthma" and "previously took antipyretics" during the dialogue with the health assistant, the system not only archives the positive test result but also automatically writes these key medical history details and medication feedback into the "past history" and "medication record" modules of the health record. This transforms a one-time dialogue interaction into a continuously evolving, structurally complete personal life-cycle health data asset, providing comprehensive background support for providing accurate auxiliary health information subsequently.

[0118] As an example, one or more examination results may include examination results from multiple visits by a patient at the same or different hospitals. Therefore, the method disclosed herein not only enables the archiving of single-time data but also has the capability for dynamic evolution of health records across periods and institutions. Specifically, through a unified patient identification system, the system automatically aggregates and structures the results of multiple examinations from different hospitals and time points (such as previous blood routine tests, CT images, and biochemical indicators). Key clinical details generated during dialogue (such as symptom evolution, medication response, and inquiries and responses from the health assistant) are deeply correlated as contextual data to the corresponding examination records. Based on this, the system uses timeline technology to standardize, clean, and align multi-source heterogeneous data, automatically constructing a longitudinal comparison view to intuitively display the long-term trends and fluctuations of key patient indicators (such as blood glucose, tumor markers, and inflammatory factors). For example, the system can automatically perform pixel-level registration and difference annotation between a patient's lung CT image taken three years ago at one hospital and an image taken at another hospital this time. Combined with interactive information recorded in the dialogue, such as "smoking history" and "worsening cough," a "full life-cycle health record" is generated, including historical baselines, recent changes, and risk warnings, thus providing doctors with continuous diagnostic and treatment evidence across hospitals and time periods.

[0119] According to embodiments of this disclosure, the assistive health agent can also send first assistive health reference information to the user based on relevant information in the patient's health record.

[0120] As an example, when the health-assisting agent generates and pushes "first auxiliary health reference information" based on all examination results, it not only relies on the current single test data, but also deeply correlates with and calls upon the patient's historical health record information to construct more comprehensive and in-depth recommendations. Specifically, after the health-assisting agent receives all the latest "chest CT" and "complete blood count" results from the patient, before generating auxiliary health reference information, it automatically retrieves the patient's past medical history (such as "had tuberculosis three years ago" or "long-term smoking history") and historical imaging baseline data stored in the patient's health record, and dynamically compares and analyzes the current results with historical data. Taking the "lung nodule follow-up" scenario as an example, if the current CT scan shows a new nodule, when the health-assisting AI pushes the first auxiliary health reference information, it will not only state "nodule found", but will combine the records of "no history of tuberculosis" and "nodule stable within five years" in the file to generate personalized suggestions such as "the new nodule is significantly larger than the baseline five years ago. Combined with the patient's long-term smoking history, it is recommended to prioritize screening for early lung cancer risk and perform three-dimensional reconstruction by comparing with the CT images three years ago". In this way, by utilizing the continuity of historical data, the accuracy of the auxiliary health information provided and the pertinence of risk warnings can be significantly improved.

[0121] The above has been referred to Figure 1This disclosure describes a method for intelligent interaction according to embodiments of the present disclosure. The method provided by the present disclosure facilitates interaction with the user by redirecting them to an instant messaging application instead of downloading a related application. Furthermore, the method provides binding multiple examination items from a single medical visit with the same code, enabling aggregated push notifications and full completion determination, reducing the cost of repeated inquiries for patients. In addition, the method provides a result readiness detection mechanism to ensure immediate delivery of results after they are issued, reducing waiting time and missed examinations. Furthermore, the method provides supplementary health reference information based on examination results, thereby alleviating prolonged patient anxiety or enabling timely treatment of illnesses. Moreover, the method provides identity mismatch detection and authorization token control to support compliant proxy procedures by guardians / family members. Finally, the method provides structured archiving and cross-hospital comparison to form long-term health records, improving the efficiency of subsequent follow-up visits.

[0122] Figure 2 A schematic diagram of a system architecture according to an embodiment of the present disclosure is shown.

[0123] like Figure 2 As shown, the system architecture for intelligent interaction includes a medical device layer, a data processing layer, an agent intelligence layer, and an interactive terminal layer. Each layer communicates with the other through standardized interfaces to achieve data flow and command interaction, forming a complete closed loop from raw data collection to user-end information push. At the same time, a built-in authorization verification mechanism ensures the security of medical data access and privacy compliance.

[0124] The medical equipment layer is the bottom data acquisition end of the system, including various clinical auxiliary examination equipment such as CT equipment, MRI equipment, and biochemical analyzers, which are used to collect and generate raw examination data such as images and laboratory tests. The equipment in this layer communicates directly with the medical information system (e.g., laboratory information system (LIS)) in the data processing layer through data interfaces, and uploads the raw examination data to the LIS system for unified collection.

[0125] The data processing layer receives raw examination data from the medical device layer, performs data parsing, format conversion and standardization, generates structured examination result data, and pushes the processed structured data to the Agent brain of the Agent intelligence layer, providing a compliant and usable data foundation for subsequent intelligent analysis and user interaction.

[0126] The Agent intelligence layer is the core control and intelligent processing unit of the system, including the Agent brain, the auxiliary health knowledge base, and the authorization verification module. The auxiliary health knowledge base stores data such as medical knowledge, consultation process templates, and risk assessment rules, providing knowledge support for the information processing and generation of auxiliary health reference information for the auxiliary health intelligence agent. The authorization verification module is responsible for verifying user identity consistency and agent authorization, ensuring data access security. This layer receives structured examination data pushed by the LIS system through the Agent brain, calls the auxiliary health knowledge base for information processing and auxiliary health reference information generation, and provides dynamic QR code and instant messaging (IM) interface support to the interactive terminal layer. It also establishes bidirectional interaction with the user's mobile terminal: receiving verification requests initiated by the user via mobile phone, the authorization verification module compares the user's identity information with the patient's identity information. If the identities match, data push is directly allowed; if the identities do not match, the module receives the agent authorization information submitted by the user and completes the verification. Only after successful verification is data access granted to the user.

[0127] The interactive terminal layer serves as the user-side interaction entry point, including dynamic QR codes, an IM interface, and the user's mobile terminal. Users scan the dynamic QR code generated by the Agent intelligent layer to add an auxiliary health agent in the instant messaging application and send a verification request to the system. After the authorization verification module completes identity consistency verification or agent authorization verification, the Agent intelligent layer pushes the processed check results and auxiliary health reference information to the user's mobile terminal via the IM interface for viewing and use.

[0128] The system architecture disclosed herein automates the entire process of examination data flow from device collection to the user terminal without manual intervention, significantly improving the efficiency of examination result distribution and shortening the time for patients to obtain reports. Through identity verification and authorization mechanisms, it achieves refined control over access permissions to examination results, effectively preventing unauthorized access and data leakage, and complying with relevant regulations on medical data security and personal information protection. Relying on knowledge graph support, it enables structured presentation of examination results and generation of auxiliary health reference information, enhancing the utilization value of examination results and the patient's medical experience. Dynamic QR codes and IM interfaces provide convenient and low-barrier user interaction entry points, eliminating the need to download a dedicated app and lowering the barrier to entry for users.

[0129] Figure 3 A schematic diagram of the interaction flow according to an embodiment of the present disclosure is shown.

[0130] like Figure 3 As shown, this interaction process constructs a closed-loop medical data interaction system from user-triggered QR code scanning to precise push notifications from the auxiliary health intelligence agent. The specific steps are described in detail below:

[0131] 1. Patient-side QR code scanning and parsing

[0132] The patient scans a QR code on the examination form using their smartphone. This QR code contains encrypted unique identification (such as a patient ID) and metadata related to the current examination. The scanning action triggers an instant messaging application on the phone (such as WeChat or DingTalk), which parses the QR code data into structured instructions and automatically initiates a connection request to the server.

[0133] 2. Server-side core processing

[0134] Once the request reaches the server, it is processed collaboratively by the following five core modules:

[0135] QR code parsing and mapping module: First, the QR code data is decrypted and verified to extract the patient ID and examination item code, and then mapped in the database to confirm the specific department and execution status of the examination.

[0136] Identity verification and authorization module: The system compares the identity information of the current user scanning the code (or logged-in account) with the preset information in the patient's health record (such as mobile phone number, ID card number, or biometrics). If the identities match, the verification is passed directly; if they do not match (e.g., a family member is checking on behalf of the user), a dynamic authorization process is triggered (e.g., requiring a verification code to be entered) to ensure that only authorized personnel can access the data.

[0137] Result Acquisition and Synchronization Module: Once authorization is granted, this module initiates a data retrieval request to the medical information system. It monitors the examination status in real time, and once the relevant examination results (such as blood reports and imaging films) have been reviewed, it standardizes and encapsulates the multimodal data (text, numerical values, DICOM images) for push.

[0138] Health record archiving module: While synchronizing data, the system automatically extracts the structured results of this examination, the generation time, the examination items, and the context generated by this interaction (such as the auxiliary health agent inquiry record and patient response), and writes them into the patient's long-term health record according to a unified standard, so as to realize the continuous accumulation and version update of data.

[0139] Task scheduling and reminder module: Based on preset business rules (such as "critical value", "all results issued", or "after patient scans code"), this module triggers task scheduling logic. It determines the timing, method (instant push or timed reminder) of the push notification, and the recipients, and generates the corresponding push task queue.

[0140] 3. Assisting health-related intelligent agents in interaction and decision-making

[0141] The server will push the processed task queue and aggregated inspection data to the terminal interface of the health-assisting intelligent agent through the message interface.

[0142] Intelligent agent assistance: The intelligent agent assists health by automatically performing preliminary analysis of data and generating "first auxiliary health reference information".

[0143] As an example, the health-assisting intelligent agent relies on technologies such as fine-tuning of medical evidence-based large models, RAG vector memory retrieval, chain-thinking CoT reasoning, clinical reasoning RLHF modeling, and multi-tool collaborative invocation to achieve professional interpretation of test results, complete a comprehensive assessment of the patient's health status by combining consultation information and personal historical health records, and generate compliant auxiliary health reference information.

[0144] Specifically, the auxiliary health intelligence agent uses a large-scale model with SFT fine-tuned under high-quality evidence-based medicine supervision as its core reasoning hub. The training corpus is limited to authoritative clinical guidelines, professional medical textbooks, and a large number of real desensitized cases, enabling the large-scale model to incorporate a standardized medical terminology system, indicator reference logic, and evidence-based medicine knowledge benchmarks, thus possessing a professional understanding of laboratory and imaging reports from the ground up. At the same time, the model is modeled through clinical logic RLHF reinforcement learning, learning the standard clinical consultation path and progressive questioning logic of chief complaint, present medical history, past medical history, and accompanying symptoms. Combined with the chain thinking CoT reasoning mechanism, it simulates the step-by-step decomposition and layer-by-layer connection thinking paradigm of professional doctors, providing a reasoning foundation for subsequent comprehensive assessment and reference information generation.

[0145] First, the health-assisting intelligent agent uses a perception module to extract medical entities and perform structured parsing on unstructured data such as lab reports and imaging reports input by users. It then standardizes and outputs machine-recognizable fields such as examination items, test values, units, reference ranges, abnormal markers, and key features of imaging lesions. Simultaneously, it is compatible with receiving natural language information such as user symptom descriptions and consultation responses, completing medical semantic normalization and information regularization, and providing a standardized input source for subsequent analysis.

[0146] Secondly, the auxiliary health intelligence agent calls the memory module and RAG vector retrieval tool, using the patient's unique identifier PatientID as an index, to retrieve the patient's historical similar examination records, previous health records and past consultation interaction information in the vector database, recalling the temporal historical context. This not only enables horizontal comparison between the current test indicators and the standard reference range, but also completes vertical trend analysis across cycles and medical institutions, making up for the information limitations of interpreting only a single report.

[0147] Furthermore, under the scheduling of the planning module, the auxiliary health agent conducts multiple rounds of autonomous consultations following clinical thinking paths: based on existing abnormal test results and missing information in historical records, the large model autonomously identifies information gaps, dynamically generates progressive consultation questions, and supplements key clinical information such as symptom onset and end time, pain nature, triggering and relieving factors, basic medical history, medication history, and allergy history; and stores the real-time consultation interaction content in a structured long-term memory to form a full-dimensional health dataset covering the current test results, imaging features, historical health records, and real-time consultation symptoms and medical history.

[0148] Finally, the health-assisting intelligent agent relies on a large-scale CoT chain reasoning model, linking medical knowledge graph tools and clinical rule bases to fuse, correlate, logically deduce, and comprehensively analyze multi-source information. On the one hand, it provides both professional and accessible interpretations of various test and imaging results, objectively explaining the meaning of indicators, abnormal alerts, and time-series trends. On the other hand, it integrates test data, historical health trends, current symptoms, basic medical history, and medication usage to complete a comprehensive assessment of overall health status. The health-assisting intelligent agent generates evidence-based and compliant auxiliary health reference information, including professional interpretations of test results, health risk warnings, recommendations for relevant departments, follow-up examination schedules, and daily health management precautions, providing professional reference for patients' subsequent medical treatment and long-term health management.

[0149] As an example, the first set of auxiliary health reference information may include: interpretation of abnormal indicators and risk warnings, such as a white blood cell count (WBC) of 12.5 × 10⁻⁶ in this blood routine test. 9 / L, neutrophil percentage 82%, exceeding the upper limit of the reference range, suggesting the possibility of acute infection. It is recommended to consult a doctor to assess the site of infection and anti-infection regimen in conjunction with clinical symptoms; (cross-hospital) historical data comparison trend, for example, comparing the results of the last 3 liver function tests, the ALT index has increased from 35U / L to 58U / L, showing a continuous upward trend. Although it is still within the reference range, it is recommended to inform the doctor of this change during the consultation to rule out drug-induced liver injury or other factors affecting liver function; matching test results with symptoms suggests, for example, the positive H1N1 antigen test combined with the recorded high fever, Muscle soreness symptoms suggest a high degree of consistency with influenza A infection. It is recommended to inform the doctor of the antigen result and duration of symptoms during the consultation. For subsequent departmental recommendations and appointments, for example, given the pulmonary nodules seen on the chest CT scan, it is suggested to prioritize a visit to the respiratory medicine or thoracic surgery department, bringing the CT report along with previous imaging data for comparison and evaluation. Regarding follow-up visits and re-examinations, for example, with a glycated hemoglobin result of 7.2%, combined with a history of diabetes, it is recommended to have fasting blood glucose and glycated hemoglobin re-examined within 1-3 months, and simultaneously consult an endocrinologist to adjust the blood sugar control regimen; etc.

[0150] 4. Results Push

[0151] The server pushes the examination results, auxiliary health reference information, and suggestions from the health-assisting AI to the user's (patient's or authorized family member's) chat interface via instant messaging. The pushed content includes not only the original report, but also easy-to-understand interpretations, medical advice, and historical data comparisons generated by the health-assisting AI, ensuring that the patient can intuitively and accurately understand the examination results.

[0152] Specifically, the health-assisting intelligent agent can generate auxiliary health reference information based on the patient's examination results and "consultation" dialogue.

[0153] When the patient's test results are obtained, the health-assisting AI will initiate a consultation dialogue with the user based on the test results and the current test items (such as H1N1 influenza). For example, messages such as: "The current results show that you may have been infected with H1N1 influenza. We need to confirm this further based on your symptoms. Have you had a fever in the past two days? What was the temperature?" and "Besides fever, do you have any other symptoms such as runny nose or cough?"

[0154] System-level prompt engineering (PE) can be used to enable the health-assisting agent to collect relevant symptom information from the user when it receives examination results. For example, by setting the following system-level PE: "If you are the chief physician of the respiratory department of a tertiary-level hospital, please conduct necessary consultations based on the patient's examination results to understand the current symptom information, and based on the comprehensive judgment of the examination results and symptom information, provide the user with auxiliary health reference information on whether they are infected with related diseases. The consultation rounds should not exceed 10 times," the health-assisting agent can have the above consultation capabilities.

[0155] The health-assisting intelligent agent can also interpret patients' examination results based on RAG longitudinal historical retrieval.

[0156] Specifically, the health-assisting intelligent agent pre-configures a unique PatientID as an identity index for each patient and stores all of the patient's past examination reports in a vector database after structured and vectorized processing. When a new test result is obtained, the agent uses the PatientID as the primary key to retrieve the patient's historical similar examination data from the vector database, constructing historical context information including each test value, reference range, test time, and testing institution. The agent performs field alignment, unit unification, and value normalization between the current test result and historical similar test results, not only comparing single test values ​​horizontally with standard reference ranges but also enabling longitudinal trend analysis of indicators across periods and batches. For example, the patient's current platelet count is 120 × 10⁻⁶. 9 / L, although within the normal reference range of 100-300×10 9Within / L, but the auxiliary health agent retrieved the patient's platelet count from last month via RAG retrieval, which was 200 × 10⁶. 9 / L, after longitudinal comparison, showed a significant downward trend in the indicator. Based on this, the health-assisting intelligent agent generated in-depth interpretation: "Your platelet count is 120 × 10⁻⁶." 9 / L, although within the normal range, is lower than last month's 200×10 9 If the L / L ratio has decreased significantly, it is recommended to explain this change to the doctor during your consultation.

[0157] The health-assisting intelligent agent can generate auxiliary health reference information for patients based on all the examination results of this visit and information collected from multiple rounds of consultations.

[0158] Specifically, the health-assisting agent can integrate multi-dimensional, comprehensive data, including: aggregated results of multiple examinations, historical RAG examination trend data, CV image extraction features, symptom and medical history information collected from multiple rounds of consultations, and a medical knowledge graph rule base. Through overall reasoning and integration using a large model, it generates compliant auxiliary health reference information, outputting only reference content such as health risk warnings, home care suggestions, guidance on the timing of medical visits, and lifestyle precautions, without generating diagnostic conclusions or prescription treatment plans. For example, after integrating comprehensive information on a patient's decreased platelet count, occasional gingival bleeding, and lack of underlying medical history and medication history, the health-assisting agent generates auxiliary health reference information: "It is recommended to avoid strenuous exercise and external impacts in the near future, and pay attention to skin and gingival bleeding in daily life; if subsequent bruising increases and bleeding becomes more frequent, it is necessary to go to the clinic in time to further investigate the cause."

[0159] The health-assisting intelligent agent can also generate auxiliary consultation information for doctor-patient communication based on all the examination results of the patient's current visit and the information collected from multiple rounds of consultations.

[0160] Specifically, the health-assisting intelligent agent extracts key information from comprehensive test results, historical trends, and consultation information using a large model. Following a standardized medical narrative structure, it automatically generates two types of supplementary content adapted for doctor-patient communication: First, a consultation summary, which concisely summarizes the chief complaint, core abnormal test results, symptom timeline, past medical history, medication allergies, and points of concern during the consultation. This summary is formatted neatly and concisely, and can be directly copied and provided to the attending physician. Second, a patient consultation question list, which automatically generates standardized questions that patients should ask their doctors during consultations, focusing on dimensions such as abnormal indicators, trend changes, necessity of examinations, medication contraindications, and indications for follow-up visits. For example, the consultation summary summarizes changes in the patient's platelet count, accompanying symptoms, and health concerns; it also generates a question list, including whether further specialized examinations are needed, whether intervention and conditioning are required, any medication contraindications, and which symptoms require urgent follow-up visits, effectively assisting patients in precise communication with doctors.

[0161] Furthermore, the health-assisting intelligent agent relies on a high-quality evidence-based medicine-specific fine-tuning mechanism, a clinical thinking simulation mechanism, and a chain-like reasoning layer to achieve professional and compliant intelligent consultation and report interpretation. The pre-trained model is not trained using general, messy internet data, but rather uses millions of real clinical cases, authoritative medical textbooks, and various clinical treatment guidelines (such as NCCN treatment recommendations) as its dedicated training corpus. Supervised fine-tuning (SFT) is used to optimize domain adaptation, ensuring that the model's output strictly conforms to authoritative clinical guidelines, avoiding generalization errors and unprofessional expressions. Simultaneously, the health-assisting intelligent agent replicates real clinical diagnostic thinking paths during consultations, strictly adhering to the standard clinical consultation logic framework of chief complaint, present illness, past medical history, and accompanying symptoms. This proactive and precise follow-up questioning capability is learned through reinforcement learning based on human feedback (RLHF) modeling of standardized clinical logic. It can dynamically initiate targeted, progressive questions based on the user's existing information, such as automatically asking whether the pain is persistent or intermittent, the onset time of symptoms, and triggering and relieving factors, among other key clinical information. In addition, the built-in reasoning layer of the health-assisting intelligent agent introduces chain thinking (CoT) and intelligent agent collaboration mechanism, which simulates the diagnostic thinking of professional doctors to analyze step by step and deduce layer by layer. Through multiple rounds of information completion, correlation reasoning and logical verification, it completes multi-dimensional information fusion and judgment. Without making a disease diagnosis, it realizes intelligent consultation, result interpretation and auxiliary health reference information generation that conforms to the norms of evidence-based medicine and fits the real clinical thinking.

[0162] As an example, the health-assisting intelligent agent performs structured interpretation of raw examination reports from LIS, PACS, or HIS systems: Using pre-defined medical entity extraction rules or pre-trained models, it extracts key information from unstructured text and maps it to standardized fields, including examination item name, test value, unit, reference range, anomaly markers, examination time, and institution. Based on the reference range, it generates anomaly labels and risk level markers; for example, it might interpret "white blood cell count 12.5 × 10⁻⁶" as an example. 9 / L, reference value 3.5-9.5×10 9 The " / L" string is parsed as structured data and labeled as "slightly elevated," providing a calculable and comparable standardized data foundation for subsequent processing.

[0163] Then, the health-assisting intelligent agent performs aggregated analysis on multiple examination results from a single visit or across time periods based on the same patient_id or encounter_id: it aggregates all examination data using the single visit identifier as an index, completes unit unification and deduplication, and performs cross-item association analysis and cross-time trend comparison based on clinical rules in the knowledge graph. For example, for patients who are positive for influenza A, it aggregates influenza A antigen, blood routine and chest X-ray results, identifies the combination pattern of "antigen positive + elevated white blood cell count + no consolidation on chest X-ray", and compares the trend of white blood cell count changes with historical health records to generate a multi-dimensional data view, providing a complete basis for subsequent consultation and auxiliary information generation.

[0164] The health-assisting intelligent agent also conducts structured consultations with patients based on examination results: it calls up preset standardized consultation templates according to the type of examination items, identifies information gaps by comparing existing examination data, and then asks targeted questions to patients through the interactive interface to supplement key information such as symptom presentation, basic medical history, medication history, allergy history, and pregnancy and childbirth status. For example, for patients who test positive for influenza A, the health-assisting intelligent agent will automatically ask about the highest temperature of fever, duration, accompanying symptoms, and underlying diseases to improve the patient's medical information.

[0165] Finally, based on the complete data collected through structured parsing, aggregation analysis, and consultation, the intelligent agent generates two types of consultation assistance information: one is a consultation summary for doctors to read quickly, which extracts key information according to a fixed structure of "chief complaint - examination results - symptoms - past medical history - concerns". For example, it integrates information such as "fever with cough for 2 days, positive H1N1 antigen, and history of mild asthma" to form a consultation summary that can be directly copied to doctors; the other is a list of communication questions for patients, which generates a structured list of consultation questions based on abnormal examination points, risk factors, and potential patient questions. For example, it prompts patients to consult on antiviral treatment indications, drug interactions, and emergency medical treatment indications.

[0166] According to embodiments of this disclosure, an interactive process of "scanning and parsing—identity verification—message interaction—result push" is constructed. While ensuring patient privacy and security, scattered examination results and consultation information from dialogue interactions are structured and archived into long-term health records, forming a continuous and complete personal health profile. Furthermore, by leveraging an auxiliary health intelligence agent to perform deep correlation analysis between single results and historical records, "first auxiliary health reference information" is generated, including risk warnings, historical comparisons, and medical treatment suggestions. This promotes the transformation of medical services from "passive query" to "proactive intelligent push," significantly improving the efficiency of doctor-patient communication and the patient's medical experience.

[0167] Figure 4 A schematic diagram of a multi-item aggregation process with the same code according to an embodiment of the present disclosure is shown.

[0168] like Figure 4 As shown, the specific implementation steps of the multi-item aggregation process with the same code are as follows:

[0169] Front-end identification and user scanning: All examination request forms issued by a patient during the same visit are associated with the same unique QR code (i.e., a unique identifier for each visit). Patients can scan this QR code with their mobile phones to trigger subsequent processing on the system server.

[0170] Server-side core processing: The server has two key built-in modules, namely the order status management module and the result aggregation and judgment module.

[0171] Order Status Management Module: Responsible for tracking and updating the execution progress and report generation status of all examination items (such as MRI, CT, B-ultrasound, blood routine, liver function, etc.) under a single visit.

[0172] The results aggregation and judgment module collects and integrates examination results data from different devices and of different types based on the same medical event identifier (encounter_id) to form a structured summary report.

[0173] Multi-device data aggregation: Results data generated by various examination devices such as MRI, CT, B-ultrasound, blood routine, liver function, etc. are all associated with the server through the same medical visit event identifier encounter_id, realizing unified aggregation of cross-device and cross-type data.

[0174] Results push phase: After the results aggregation and judgment module completes the data summary and processing of all examination items, the server pushes the full examination results data of a single visit to the auxiliary health intelligent agent through the "result summary push" interface.

[0175] This process uses the same identification code for multiple items in a single visit, enabling one-click collection, status tracking, and summary push of data from multiple examination items in a single visit, thus avoiding data fragmentation and matching errors caused by multiple items and multiple codes.

[0176] Figure 5 A schematic diagram of a non-personalized QR code authorization process according to an embodiment of the present disclosure is shown.

[0177] like Figure 5 As shown, the specific implementation steps of the authorization process for someone else to scan the code are as follows:

[0178] User-side QR code trigger: Family members or guardians scan the QR code on the examination form with their mobile phones to initiate a request to access the examination report to the system.

[0179] Authorization control module processing: After a request enters the authorization control module, the following verification and processing are performed:

[0180] Identity mismatch detection: The system automatically compares the identity information of the user scanning the code with the identity information of the patient, and identifies that the current user is not the patient.

[0181] Monitoring / Authorization Process: Trigger the monitoring relationship verification or proxy authorization process, such as verifying the monitoring relationship and collecting the patient's authorization confirmation (such as SMS verification code, facial verification, electronic signature, etc.) to ensure that the user has legitimate proxy access rights;

[0182] Generate consent_token: After the authorization verification is successful, the system generates a one-time authorization credential (consent_token) as a valid token for subsequent data access.

[0183] After authorization, the system retrieves the examination results: The authorization control module sends a data access request to the hospital information system (e.g., HIS / LIS / PACS) with the consent_token. After the system verifies the validity of the token, it returns the corresponding examination results.

[0184] Results Feedback: The obtained report results are sent back to the authorization control module, which then pushes them to the family member / guardian's mobile terminal to complete the authorization access process in non-personal scenarios.

[0185] The non-personalized QR code authorization method according to the embodiments of this disclosure solves the problem of compliant access to medical data in scenarios where family members / guardians are not the person in question, thus protecting patient privacy and meeting the actual needs of proxy viewing of reports.

[0186] Figure 6 A schematic diagram of a result readiness detection and push process according to an embodiment of the present disclosure is shown.

[0187] like Figure 6 As shown, the specific steps of the result readiness detection and push process are as follows:

[0188] Medical visit event identifier association: Based on the unique identifier encounter_id of a single medical visit event, all examination items under that visit are associated, serving as a unique index for subsequent status tracking and result aggregation.

[0189] Incomplete Project Status Query: The system queries the execution status of all examination items under this visit based on encounter_id, and identifies items that have not yet been completed or for which no report has been issued.

[0190] Asynchronous status monitoring mechanism: Continuously monitor the status updates of incomplete projects through subscription, polling or message queue. When the result of any checked project is generated and ready, the subsequent process is triggered.

[0191] Result readiness determination: The system determines whether the examination results of this visit meet the push conditions based on preset rules (such as all examination items have been completed, key item reports have been generated, etc.). When the results are ready, a summary notification is pushed to the auxiliary health intelligent agent.

[0192] Assisted Health Agent Processing and Push: After receiving the result-ready notification, the assisted health agent pushes the result information to the user through the dialogue interface, and provides two forms of result delivery: structured parsing, which processes the examination results in a structured manner to generate auxiliary health reference information such as a consultation summary and a list of questions that can be directly used to assist communication; and original report link, which provides a link to access the original report of the HIS / LIS / PACS system, allowing users to view the complete examination report details.

[0193] This process achieves unified status management of multiple examination items in a single visit through encounter_id. Combined with the asynchronous monitoring mechanism of subscription / polling / message queue, it realizes automatic judgment and proactive push of results ready. At the same time, it takes into account the dual delivery form of structured auxiliary health reference information and original report links, which not only improves the automation of examination result distribution, but also provides users with diversified ways to view results.

[0194] Figure 7 A schematic diagram illustrating the results of structured archiving of health records and cross-hospital comparisons according to embodiments of the present disclosure is shown.

[0195] According to this disclosure, the system can receive both the current examination results and historical examination results stored in the database. Based on the patient's unique identifier (patient_id), examination data from different times and hospitals are uniformly collected into the patient's health record and stored in a structured format. Each record in the record includes the date, hospital, examination item, test value, reference range, and result label (such as "normal"), forming a standardized comparison chart of examination results. For example, the example records multiple data points, such as ALT measured at hospital A on May 10, 2025, being 42 U / L, and AST measured at hospital B on April 18, 2025, being 36 U / L.

[0196] Furthermore, embodiments of this disclosure also support cross-hospital result comparison: based on structured data in patient health records, a cross-hospital result comparison table is automatically generated. The comparison table integrates test results from different hospitals and corresponding test dates, using the test item as the dimension. It also extracts the most recent test value for each item and calculates its trend. For example, in the example, the ALT item shows two test results from Hospital A and Hospital B (42 U / L and 38 U / L), with the most recent value being 42 U / L from Hospital A, and the trend is marked as "↑" (increasing); the two AST item results are 36 U / L and 39 U / L respectively, with the most recent value being 39 U / L, and the trend is marked as "→" (stable); the two WBC item results are 6.2 × 10⁻⁶. 9 / L and 6.0×10 9 / L, the most recent value is 6.2×10 9 / L, the trend is marked as "↑" (rising).

[0197] This process solves the problem of data silos between different hospitals by structuring and archiving examination results from multiple sources and in multiple formats. It enables cross-hospital and cross-time comparison of examination indicators and visualization of trends, providing a continuous and complete data foundation for the subsequent generation of auxiliary health reference information. It also makes it easier for patients and doctors to intuitively understand the changes in indicators.

[0198] Figure 8 A block diagram of an apparatus for intelligent interaction according to an embodiment of the present disclosure is shown.

[0199] like Figure 8 As shown, the device 800 includes an interaction module 810, a result monitoring and push module 820, and a display module 830.

[0200] The interaction module 810 can be configured to open a dialog interface in an instant messaging application in response to scanning an identification code on a patient's auxiliary examination request form, allowing the user to interact with the auxiliary health agent, where multiple auxiliary examination request forms from the patient during the same visit have the same identification code.

[0201] According to embodiments of this disclosure, the identification code may be associated with at least one of patient identity information and application scenario information, wherein the patient identity information includes at least one of the patient's mobile phone number, name, and identification information, and wherein the application scenario information includes at least one of the department where the patient seeks medical treatment and one or more examination items.

[0202] According to embodiments of this disclosure, opening a dialogue interface in an instant messaging application for a user to interact with an auxiliary health agent includes: in response to a scan, based on the auxiliary health agent being a friend of the user, redirecting to a dialogue interface in the instant messaging application for interacting with the auxiliary health agent; and in response to a scan, based on the auxiliary health agent not being a friend of the user, redirecting to a request to add a friend interface in the instant messaging application for requesting to add the auxiliary health agent as a friend, and in response to the auxiliary health agent adding the user as a friend through the request to add a friend interface, redirecting to a dialogue interface in the instant messaging application for interacting with the auxiliary health agent.

[0203] The result monitoring and push module 820 can be configured to monitor or receive the result issuance status of inspection items associated with the identification code, and push one or more inspection results to the dialog interface when monitoring or receiving one or more inspection result issuances.

[0204] The display module 830 can be configured to display interactive content between the auxiliary health agent and the user associated with one or more examination results in a dialogue interface. The interactive content includes first interactive content, which includes first auxiliary health reference information sent by the auxiliary health agent to the user based on all examination results in response to the issuance of all examination results for the examination items associated with the identification code.

[0205] According to embodiments of this disclosure, the interactive content may further include second interactive content, which includes second auxiliary health reference information related to each examination result sent to the user by the auxiliary health agent after each examination result is pushed.

[0206] According to embodiments of this disclosure, the interactive content may further include third interactive content and fourth interactive content. The third interactive content may include inquiry information related to the patient's condition sent by the assistive health agent, and the fourth interactive content may include reply information sent by the user in response to the inquiry information.

[0207] According to embodiments of this disclosure, the interactive content may further include fifth interactive content, which may include consultation assistance information sent by the assistive health agent to assist patients in communicating with doctors.

[0208] The device 800 may further include an identity verification module (not shown). The identity verification module can be configured to perform an identity consistency check before pushing one or more examination results to the dialog interface, wherein, in response to the user's identity information matching the patient's identity information, the assistive health agent pushes one or more examination results to the dialog interface; and in response to the user's identity information not matching the patient's identity information, the assistive health agent prompts the user to perform a patient authorization operation, and after the user completes the patient authorization operation, pushes one or more examination results to the dialog interface.

[0209] According to embodiments of this disclosure, patient authorization operations may include at least one of the following: SMS verification, facial biometric verification, electronic signature verification, guardianship relationship verification, or verification by inputting relevant information from the patient's identification documents.

[0210] According to embodiments of this disclosure, performing identity consistency verification before pushing one or more examination results to the dialog interface may include: when a user scans an identification code and adds an auxiliary health agent as a friend in an instant messaging application, determining whether the mobile phone number of the user bound to the instant messaging application is consistent with the mobile phone number stored in the patient's health record.

[0211] According to embodiments of this disclosure, performing identity consistency verification before pushing one or more examination results to the dialog interface may include: when a user scans an identification code and adds an auxiliary health agent as a friend in an instant messaging application, collecting at least one of the user's name and identification information, and determining whether the user's name and / or identification information matches the name and / or identification information of the patient associated with the identification code.

[0212] According to embodiments of this disclosure, performing identity consistency verification before pushing one or more inspection results to the chat interface may include: performing identity consistency verification when a user scans an identification code and adds the auxiliary health agent as a friend in an instant messaging application; or performing identity consistency verification when the auxiliary health agent monitors or receives one or more inspection results.

[0213] The device 800 may also include an archiving module (not shown). The archiving module can be configured to update the patient's health record based on one or more examination results in the dialog interface and the interactive content associated with the one or more examination results.

[0214] According to embodiments of this disclosure, the assistive health agent can also send first assistive health reference information to the user based on relevant information in the patient's health record.

[0215] Figure 9 A block diagram of a device for intelligent interaction according to an embodiment of the present disclosure is shown.

[0216] The above explanation of the method also applies to Figure 9 The devices shown are for intelligent interaction, unless otherwise explicitly stated.

[0217] See Figure 9 The device may include a processor 910 and a memory 920. Both the processor 910 and the memory 920 can be connected via a bus 930.

[0218] Processor 910 can perform various actions and processes according to the program stored in memory 920. Specifically, processor 910 can be an integrated circuit chip with signal processing capabilities. The processor can be a general-purpose processor, digital signal processor (DSP), application-specific integrated circuit (ASIC), off-the-shelf programmable gate array (FPGA), or other programmable logic device, discrete gate or transistor logic device, or discrete hardware component. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor, and can be based on x86 architecture or ARM architecture.

[0219] Memory 920 stores computer instructions that, when executed by processor 910, implement the methods described above. Memory 920 may be volatile memory or non-volatile memory, or may include both. Non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory may be random access memory (RAM), which serves as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous linked dynamic random access memory (SLDRAM), and direct memory bus random access memory (DR RAM). It should be noted that the memory of the methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0220] According to yet another embodiment of this disclosure, a computer-readable recording medium is also provided. Figure 10 A schematic diagram of a recording medium 1000 according to an embodiment of the present disclosure is shown.

[0221] like Figure 10As shown, computer-executable instructions 1010 are stored on the recording medium 1000. When the computer-executable instructions 1010 are executed by a processor, they can cause the processor to perform the method described with reference to the above figures according to embodiments of the present disclosure. The computer-readable recording medium in the embodiments of the present disclosure may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. Non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory may be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous linked dynamic random access memory (SLDRAM), and direct memory bus random access memory (DR RAM). It should be noted that the memory of the methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0222] This disclosure also provides a computer program product including computer-executable instructions. When executed by a processor, the computer-executable instructions cause the processor to perform any of the methods described above.

[0223] It should be noted that the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0224] In general, the various exemplary embodiments of this disclosure can be implemented in hardware or dedicated circuitry, software, firmware, logic, or any combination thereof. Some aspects can be implemented in hardware, while others can be implemented in firmware or software that can be executed by a controller, microprocessor, or other computing device. When aspects of embodiments of this disclosure are illustrated or described as block diagrams, flowcharts, or using some other graphical representation, it will be understood that the blocks, apparatuses, systems, techniques, or methods described herein can be implemented as non-limiting examples in hardware, software, firmware, dedicated circuitry or logic, general-purpose hardware or controllers or other computing devices, or some combination thereof.

[0225] The exemplary embodiments of the present invention described in detail above are merely illustrative and not restrictive. Those skilled in the art will understand that various modifications and combinations can be made to these embodiments or their features without departing from the principles and spirit of the invention, and such modifications should fall within the scope of the invention.

Claims

1. A method for intelligent interaction, comprising: In response to scanning the identification code on the patient's auxiliary examination request form, a dialog interface for the user to interact with the auxiliary health agent is opened in an instant messaging application, wherein multiple auxiliary examination request forms of the patient in the same medical visit have the same identification code. The auxiliary health intelligent agent monitors or receives the result issuance status of the examination items associated with the identification code, and pushes the one or more examination results to the dialogue interface when monitoring or receiving one or more examination results. as well as The dialogue interface displays the interaction content between the auxiliary health agent and the user, which is associated with the one or more examination results. The interaction content includes first interaction content, which includes first auxiliary health reference information sent by the auxiliary health agent to the user based on all examination results in response to the issuance of all examination results for the examination items associated with the identification code.

2. The method according to claim 1, further comprising: Before pushing the one or more check results to the dialog interface, perform an identity consistency check. Specifically, in response to the user's identity information matching the patient's identity information, the auxiliary health agent pushes one or more examination results to the dialogue interface; and In response to the inconsistency between the user's identity information and the patient's identity information, the auxiliary health agent prompts the user to perform a patient authorization operation, and after the user completes the patient authorization operation, pushes the one or more examination results to the dialogue interface.

3. The method according to claim 2, wherein, The patient authorization process includes at least one of the following: SMS verification, facial biometric verification, electronic signature verification, guardianship relationship verification, or verification by inputting relevant information from the patient's identification documents.

4. The method according to claim 2, wherein, Performing identity consistency verification before pushing the one or more check results to the dialog interface includes: When the user scans the identification code and adds the health assistant as a friend in the instant messaging application, it is determined whether the mobile phone number of the user bound to the instant messaging application matches the mobile phone number stored in the patient's health record.

5. The method according to claim 2, wherein, Performing identity consistency verification before pushing the one or more check results to the dialog interface includes: When the user scans the identification code and adds the health assistant as a friend in the instant messaging application, at least one of the user's name and identification information is collected, and it is determined whether the user's name and / or identification information matches the name and / or identification information of the patient associated with the identification code.

6. The method according to claim 2, wherein, Performing identity consistency verification before pushing the one or more check results to the dialog interface includes: The identity consistency verification is performed when the user scans the identification code and adds the auxiliary health agent as a friend in the instant messaging application; or When the auxiliary health intelligent agent monitors or receives one or more inspection results, the identity consistency verification is performed.

7. The method according to claim 1, further comprising: The patient's health record is updated based on one or more examination results in the dialog interface and the interactive content associated with the one or more examination results.

8. The method according to claim 7, wherein, The health assistance agent sends the first auxiliary health reference information to the user based on all the patient's examination results and relevant information from the health record.

9. The method according to claim 1, wherein, The identification code is associated with at least one of the patient's identity information and application scenario information, wherein the patient's identity information includes at least one of the patient's mobile phone number, name, and identification information, and wherein the application scenario information includes at least one of the department where the patient sought medical treatment and one or more examination items.

10. The method according to claim 1, wherein, Opening a conversational interface in an instant messaging application for users to interact with an assistive health agent includes: In response to the scan, based on the fact that the health-assistant agent and the user are friends, the user is redirected to the chat interface in the instant messaging application where they can interact with the health-assistant agent; and In response to the scan, based on the fact that the health assistant and the user are not friends, the user is redirected to the friend request interface in the instant messaging application to request to add the health assistant as a friend. In response to the health assistant and the user becoming friends through the friend request interface, the user is redirected to the chat interface in the instant messaging application to interact with the health assistant.

11. The method according to claim 1, wherein, The interactive content also includes second interactive content, which includes second auxiliary health reference information related to each examination result sent by the auxiliary health agent to the user after each examination result is pushed.

12. The method according to claim 1, wherein, The interactive content also includes a third interactive content and a fourth interactive content. The third interactive content includes inquiry information related to the patient's condition sent by the auxiliary health intelligence agent, and the fourth interactive content includes reply information sent by the user in response to the inquiry information.

13. The method according to claim 1, wherein, The interactive content also includes a fifth interactive content, which includes consultation assistance information sent by the auxiliary health agent to assist the patient in communicating with the doctor.

14. A device for intelligent interaction, comprising: The interaction module is configured to open a dialog interface in an instant messaging application in response to a scanning operation of the identification code on the patient's auxiliary examination request form, wherein multiple auxiliary examination request forms of the patient in the same medical visit have the same identification code. The result monitoring and push module is configured to monitor or receive the result issuance status of the inspection items associated with the identification code, and push the one or more inspection results to the dialog interface when monitoring or receiving one or more inspection results. as well as The display module is configured to display, in the dialogue interface, interactive content between the auxiliary health agent and the user associated with one or more examination results, wherein the interactive content includes first interactive content, which includes first auxiliary health reference information sent by the auxiliary health agent to the user based on all examination results in response to the issuance of all examination results for the examination items associated with the identification code.

15. The apparatus of claim 14, further comprising: The identity verification module is configured to perform an identity consistency check before pushing the one or more check results to the dialog interface. Specifically, in response to the user's identity information matching the patient's identity information, the auxiliary health agent pushes one or more examination results to the dialogue interface; and In response to the inconsistency between the user's identity information and the patient's identity information, the auxiliary health agent prompts the user to perform a patient authorization operation, and after the user completes the patient authorization operation, pushes the one or more examination results to the dialogue interface.

16. The apparatus of claim 14, further comprising: The archiving module is configured to update the patient's health record based on one or more examination results in the dialog interface and the interactive content associated with the one or more examination results.

17. A device for intelligent interaction, comprising: processor, and A memory storing computer-executable instructions that, when executed by a processor, cause the processor to perform the method as described in any one of claims 1 to 13.

18. A computer-readable recording medium storing computer-executable instructions, wherein, The computer-executable instructions, when executed by a processor, cause the processor to perform the method as described in any one of claims 1 to 13.

19. A computer program product comprising computer-executable instructions, wherein, The computer-executable instructions, when executed by a processor, cause the processor to perform the method as described in any one of claims 1 to 13.

Images