Multi-agent collaborative hospital guide method, device and equipment and storage medium

Abstract

The embodiment of the invention discloses a multi-agent collaborative hospital guide method and device, equipment and a storage medium. According to the technical scheme provided by the embodiment of the invention, the user inquiry data is obtained through the user interaction agent, the symptom information and the user intention are determined according to the user inquiry data, and when the user intention is the triage request, the candidate department list is generated through the triage decision agent according to the symptom information; determining a key discrimination symptom of each candidate department in the candidate department list through a medical knowledge agent, and performing symptom query processing according to the key discrimination symptom through a user interaction agent to obtain user reply data, the medical knowledge agent determines the department confidence coefficient of each candidate department according to the user inquiry data and the user reply data, the dynamic scheduling agent determines the target department from the candidate department list according to the department confidence coefficient, and the user interaction agent performs hospital guide feedback processing according to the target department. The personalized hospital guide appeal of the user can be accurately responded, and the user experience is improved.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to the technical field of computer, and particularly relate to a multi-agent collaborative triage method and device, equipment and storage medium. BACKGROUND

[0002] The current medical triage system solution relies on manual service or intelligent triage system. The manual service mode is not only inefficient, but also limited by labor cost and working time, and the service coverage and timeliness of response are insufficient. The existing intelligent triage system is mostly based on rules or single language model, and it is difficult to output truly suitable triage feedback for users.

[0003] For example, the existing intelligent triage system performs poorly in aspects such as symptom department matching and medical terminology understanding. The intelligent triage system can only rely on static rules to give a simple triage feedback according to the user's question, and cannot accurately respond to the personalized triage demands of users, resulting in poor user experience. SUMMARY

[0004] Embodiments of the present application provide a multi-agent collaborative triage method, device, equipment and storage medium to solve the technical problem that the intelligent triage solution in related technologies relies on static rules and cannot accurately respond to the personalized triage demands of users, resulting in poor user experience. The personalized triage demands of users can be accurately responded to, and the user experience can be improved.

[0005] In a first aspect, embodiments of the present application provide a multi-agent collaborative triage method, comprising: obtaining user inquiry data through a user interaction agent, performing entity extraction processing and intent analysis processing on the user inquiry data to obtain symptom information and user intent; in the case that the user intent is a triage request, generating a candidate department list according to the symptom information through a triage decision agent, and determining key discriminant symptoms of each candidate department in the candidate department list through a medical knowledge agent; performing symptom follow-up processing according to the key discriminant symptoms through the user interaction agent to obtain user reply data; determining department confidence of each candidate department according to the user inquiry data and the user reply data through the medical knowledge agent; determining a target department from the candidate department list according to the department confidence through a dynamic scheduling agent, and performing triage feedback processing according to the target department through the user interaction agent.

[0006] In a second aspect, embodiments of the present application provide a multi-agent collaborative triage device, comprising an inquiry response module, a symptom analysis module, a follow-up processing module, a confidence determination module and a triage feedback module, wherein: The inquiry response module is configured to acquire user inquiry data through the user interaction agent, perform entity extraction processing and intent analysis processing on the user inquiry data, and obtain symptom information and user intent. The symptom analysis module is configured to, when the user intent is a triage request, generate a candidate department list according to the symptom information through a triage decision agent, and determine key discriminant symptoms of each candidate department in the candidate department list through a medical knowledge agent. The follow-up processing module is configured to perform symptom follow-up processing according to the key discriminant symptoms through the user interaction agent, to acquire user reply data. The confidence determination module is configured to determine department confidence of each candidate department according to the user inquiry data and the user reply data through the medical knowledge agent. The guidance feedback module is configured to determine a target department from the candidate department list according to the department confidence through a dynamic scheduling agent, and perform guidance feedback processing according to the target department through the user interaction agent.

[0007] In a third aspect, an embodiment of the present application provides a multi-agent collaborative guidance device, comprising a memory and one or more processors. The memory is configured to store one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the multi-agent collaborative guidance method according to the first aspect.

[0008] In a fourth aspect, an embodiment of the present application provides a storage medium storing computer executable instructions, which, when executed by a computer processor, are used to perform the multi-agent collaborative guidance method according to the first aspect.

[0009] The user inquiry data is acquired through the user interaction agent, the user inquiry data is subjected to entity extraction processing and intent analysis processing to obtain symptom information and user intent, when the user intent is a triage request, a candidate department list is generated according to the symptom information through a triage decision agent, key discriminant symptoms of each candidate department in the candidate department list are determined through a medical knowledge agent, symptom follow-up processing is performed according to the key discriminant symptoms through the user interaction agent to acquire user reply data, department confidence of each candidate department is determined according to the user inquiry data and the user reply data through the medical knowledge agent, a target department is determined from the candidate department list according to the department confidence through a dynamic scheduling agent, and guidance feedback processing is performed according to the target department through the user interaction agent, so that the individualized guidance demands of the user can be accurately responded to, and the user experience is improved. BRIEF DESCRIPTION OF DRAWINGS

[0010] Figure 1 is a flowchart of a multi-agent collaborative guidance method provided by an embodiment of the present application; Figure 2 is a flowchart of another multi-agent collaborative guidance method provided by an embodiment of the present application; Figure 3 is a structural schematic diagram of a multi-agent collaborative guidance device provided by an embodiment of the present application; Figure 4 is a structural schematic diagram of a multi-agent collaborative guidance device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0011] In order to make the objects, technical solutions and advantages of the present application clearer, the following further describes specific embodiments of the present application with reference to the accompanying drawings. It can be understood that the specific embodiments described herein are only used to explain the present application, but not to limit the present application. In addition, it should be noted that, for the convenience of description, only parts related to the present application are shown in the drawings, but not all contents. Before discussing the example embodiments in more detail, it should be mentioned that some example embodiments are described as processes or methods depicted by flowcharts. Although the flowcharts describe each operation (or step) as a sequential process, many of the operations can be implemented in parallel, concurrently or simultaneously. In addition, the order of the operations can be rearranged. The above process can be terminated when its operations are completed, but can also have additional steps not included in the drawings. The above process can correspond to a method, function, procedure, subroutine, subprogram, etc.

[0012] Figure 1 A flowchart of a multi-agent collaborative guidance method provided by an embodiment of the present application is given, and the multi-agent collaborative guidance method provided by the embodiment of the present application can be executed by a multi-agent collaborative guidance device. The multi-agent collaborative guidance device can be realized by hardware and / or software, and integrated in a multi-agent collaborative guidance device.

[0013] The following describes a multi-agent collaborative guidance method executed by a multi-agent collaborative guidance device as an example. Referring to Figure 1 , the multi-agent collaborative guidance method comprises: S110: obtaining user inquiry data through a user interaction agent, performing entity extraction processing and intent analysis processing on the user inquiry data, and obtaining symptom information and user intent.

[0014] The application provides a plurality of agent intelligent agents to cooperate to complete a guide diagnosis process for a user. The guide diagnosis device of the multi-agent cooperation can be connected with a real-time hospital data interface to access a hospital information system (HIS) and obtain dynamic data such as department load and doctor scheduling. The agent provided by the application can include one or a combination of a plurality of user interaction agent, triage decision agent, medical knowledge agent, sentiment computing agent, dynamic scheduling agent and data security agent. The user interaction agent can process voice and / or text input and provide natural language understanding (NLU) and dialogue management (DM). The user interaction agent can be based on a natural language processing (NLP) model (for example, a natural language processing model based on BERT and medical entity recognition). The triage decision agent can perform complex symptom reasoning based on the mapping relationship between symptoms and departments. The triage decision agent can be built based on a medical knowledge graph (which can be built based on a medical knowledge graph Neo4j, containing the association relationship between symptoms, diseases and departments) and a Bayesian network. The medical knowledge agent can be used for medical knowledge retrieval and verification, supporting Neo4j knowledge graph and retrieval-augmented generation (RAG) retrieval. The sentiment computing agent can realize emotion recognition and personification feedback, which can be built based on a convolutional neural network (CNN) and a long short-term memory (LSTM) emotion model, and can be optimized in combination with a speech synthesis technology. The dynamic scheduling agent can realize task allocation and resource optimization of each agent, a contract net protocol and specific crowd reinforcement learning. The data security agent can realize patient privacy protection and compliance checking. Each agent provided by the application can be trained based on federated learning and data desensitization algorithm.

[0015] Exemplarily, the user inquiry data is obtained through the user interaction agent. The user inquiry data can be text and / or voice, that is, the user can provide the user inquiry data by inputting text and / or voice. The user inquiry data can reflect the user's current symptoms and can also reflect the user's medical history.

[0016] After obtaining the user inquiry data through the user interaction agent, entity extraction processing and intent analysis processing are performed on the user inquiry data to obtain symptom information and user intent. The entity extraction processing can extract the text related to symptoms and diseases from the user inquiry data, and generate the symptom information according to the text related to symptoms and diseases. The intent analysis processing can analyze the user inquiry data to determine whether the user wants to request triage or general query, and generate the user intent reflecting the request for triage or general query.

[0017] For example, a user can input their medical consultation data via voice or text. For instance, the user consultation data could be the voice of "I have abdominal pain." The user interaction agent can use a voice recognition model configured for the medical scenario to perform voice recognition to obtain the user consultation data. Then, entity extraction is performed on the user consultation data to extract symptom information (such as "abdominal pain"), and intent analysis is performed on the user consultation data to determine whether the user's intent is a triage request or a general question and answer.

[0018] In one possible embodiment, the multi-agent collaborative triage method provided in this application, after performing entity extraction and intent analysis on user consultation data to obtain symptom information and user intent, further includes: if the user intent is a question-and-answer request, generating question-and-answer feedback information based on the user consultation data through a medical knowledge agent; and performing question-and-answer feedback processing based on the question-and-answer feedback information through a user interaction agent.

[0019] For example, when the user's intent is a question-and-answer request, a medical knowledge agent generates question-and-answer feedback information based on the user's consultation data. For instance, the medical knowledge agent inputs the user's consultation data and analyzes and processes it to obtain the feedback information. After obtaining the feedback information, a user interaction agent can then process the feedback.

[0020] For example, when a user's consultation data is "What fruits can I eat when I have a cold?", the user interaction agent submits the consultation data to the medical knowledge agent for processing. The medical knowledge agent uses Neo4j knowledge graph and RAG retrieval technology to find "fruits rich in vitamin C, such as oranges and grapefruits, are suitable to eat when you have a cold," and verifies the relevant knowledge. The corresponding generated question-and-answer feedback information could be "Eating more fruits rich in vitamin C, such as oranges and grapefruits, when you have a cold can help boost immunity and promote recovery." This application, by generating question-and-answer feedback information based on the user's consultation data through the medical knowledge agent when the user's intent is a question-and-answer request, and then processing the question-and-answer feedback information through the user interaction agent, efficiently and accurately responds to the user's medical question-and-answer needs, thereby improving the user experience.

[0021] S120: When the user's intent is a triage request, the triage decision agent generates a list of candidate departments based on symptom information, and the medical knowledge agent determines the key diagnostic symptoms for each candidate department in the list.

[0022] For example, when the user's intent is a triage request, the triage decision agent generates a candidate department list based on the relationship between symptoms, diseases, and departments, according to the symptom information. The candidate department list records one or more candidate departments that can be used to conduct consultations on the entities corresponding to the symptom information. For example, when the user enters the consultation data "abdominal pain", a candidate department list including "gastroenterology, urology, and gynecology (gynecology is filtered out if the user is male)" is generated.

[0023] In one embodiment, after generating a list of candidate departments, a medical knowledge agent can be used to determine the key diagnostic symptoms associated with each candidate department in the list. In one embodiment, different departments correspond to different key diagnostic symptoms. For example, the key diagnostic symptom corresponding to "gastroenterology" may be "accompanied by diarrhea or vomiting", the key diagnostic symptom corresponding to "urology" may be "accompanied by hematuria or difficulty urinating", and the key diagnostic symptom corresponding to "gynecology" may be "menstrual cycle-related pain".

[0024] S130: The user interaction agent performs follow-up questioning based on key diagnostic symptoms to obtain user response data.

[0025] For example, after identifying key diagnostic symptoms, a user interaction agent can perform follow-up questioning based on these symptoms and obtain user response data based on the follow-up questioning. For instance, the agent can sequentially ask the user if they experience symptoms such as "accompanied by diarrhea or vomiting," "accompanied by hematuria or difficulty urinating," and "menstrual cycle-related pain." The user can respond to these follow-up questions (through text and / or voice) and the agent can obtain the user response data.

[0026] Optionally, when asking follow-up questions about key diagnostic symptoms, the user interface can display inquiry prompts related to the key diagnostic symptoms, as well as a response button (e.g., displaying "Yes" / "No", or showing "Key Diagnostic Symptoms" / "No Key Diagnostic Symptoms"). Users can provide their response data through the response button.

[0027] S140: Determine the confidence level of each candidate department based on user consultation data and user response data through a medical knowledge agent.

[0028] For example, a medical knowledge agent determines the confidence level of each candidate department based on user consultation data and user response data. For instance, the medical knowledge agent analyzes and processes the user consultation data and user response data using a trained confidence analysis model to determine the confidence level of each candidate department. The more relevant the user consultation data and user response data are to the symptoms and diseases that the candidate department can handle, the higher the corresponding department confidence level.

[0029] Optionally, a medical knowledge agent can perform knowledge graph queries and multi-hop inference based on user consultation data and user response data, and use a Bayesian network to calculate the department confidence score for each candidate department. The formula for calculating the department confidence score using a Bayesian network can be expressed as:

[0030] Where S represents user consultation data, H represents user response data, and Dept represents candidate departments.

[0031] S150: The dynamic scheduling agent determines the target department from the candidate department list based on the department confidence level, and the user interaction agent performs triage feedback processing based on the target department.

[0032] For example, after determining the confidence level of each candidate department, a dynamic scheduling agent selects one or more candidate departments as target departments from the candidate department list according to the department confidence level in descending order. A user interaction agent then processes the referral feedback based on the target departments to recommend suitable departments to the user. For instance, the user interaction agent uses a large language model to analyze user consultation data, user response data, and target departments, outputting referral feedback text (medical script text) recommending target departments based on the consultation data and user response data. This text is then presented to the user via text and / or voice. For example, it might display to the user, "Based on your symptoms of abdominal pain and diarrhea, gastroenterology is the most suitable department for you. Urology can also handle some abdominal discomfort problems. Which would you prefer?" The user can then choose the appropriate department based on this information.

[0033] The above describes a system that acquires user consultation data through a user interaction agent, performs entity extraction and intent analysis on the data to obtain symptom information and user intent. When the user intent is a triage request, a triage decision agent generates a candidate department list based on the symptom information, a medical knowledge agent determines the key diagnostic symptoms for each candidate department, a user interaction agent performs follow-up questioning based on the key diagnostic symptoms to obtain user response data, a medical knowledge agent determines the department confidence level for each candidate department based on the consultation and response data, a dynamic scheduling agent determines the target department from the candidate department list based on the department confidence level, and a user interaction agent provides triage feedback based on the target department. This system can accurately respond to users' personalized triage needs and improve user experience.

[0034] Based on the above embodiments, Figure 2 A flowchart of another multi-agent collaborative triage method provided in this application embodiment is given, which is a concretization of the above-described multi-agent collaborative triage method. (Reference)Figure 2 This multi-agent collaborative triage method includes: S210: Obtain user consultation data through user interaction agent, perform entity extraction and intent analysis on user consultation data to obtain symptom information and user intent.

[0035] S220: When the user's intent is a triage request, the triage decision agent generates a list of candidate departments based on symptom information, and the medical knowledge agent determines the key diagnostic symptoms for each candidate department in the list.

[0036] S230: The user interaction agent performs follow-up questioning based on key diagnostic symptoms to obtain user response data.

[0037] S240: Determine the confidence level of each candidate department based on user consultation data and user response data through a medical knowledge agent.

[0038] S250: Determine the agent decision delay and department load for each candidate department through dynamic scheduling agent.

[0039] S260: By dynamically scheduling the agent, the confidence level, agent decision delay, and departmental load are weighted based on confidence level weight, delay weight, and load weight to obtain the departmental score of each candidate department. The target department is then determined from the candidate department list based on the departmental score.

[0040] For example, the agent decision delay and department load of each candidate department are determined by a dynamic scheduling agent. The agent decision delay can be understood as the delay or time taken by the corresponding agent to determine the candidate department. The department load can be used to know the load of the corresponding department. The load can be determined by the department waiting time (patient queuing time). For example, the reciprocal of the department waiting time can be used as the load. The longer the department waiting time, the more serious the department load.

[0041] In one embodiment, the agent dynamically schedules the department confidence, agent decision delay, and department load based on pre-configured confidence weight, delay weight, and load weight to obtain the department score for each candidate department, and then determines the target department from the candidate department list based on the department score.

[0042] Optionally, the department confidence level, proxy decision delay, and department load are weighted. Specifically, the reciprocal of the department confidence level and proxy decision delay, along with the difference between a preset base (e.g., 1) and the department load, are weighted. For example, the department score can be determined using the following formula:

[0043] in, The confidence weight (e.g., 0.6). For the delay weight (e.g., 0.3), The load weight (e.g., 0.1). For departmental confidence level, Delay in agency decision-making To assess departmental load, optionally, before weighting departmental confidence, agent decision delay, and departmental load, these factors can be normalized, and then weighted. This application weights departmental confidence, agent decision delay, and departmental load based on confidence weight, delay weight, and load weight to obtain a departmental score for each candidate department. Based on these scores, a target department that better meets user needs is selected from the candidate department list, enabling precise response to users' personalized triage requests and improving user experience.

[0044] In one embodiment, the multi-agent collaborative triage method provided in this application is characterized in that, before weighting the department confidence, agent decision delay and department load by dynamically scheduling agents based on confidence weight, delay weight and load weight, it further includes: determining the load weight of each candidate department and updating the load weight according to symptom information.

[0045] In one embodiment, a corresponding delay weight can be pre-set for different symptom information. The more urgent the situation corresponding to the symptom information, the greater the corresponding delay weight. For example, the delay weight reflecting symptoms such as high fever and bleeding is greater than the delay weight reflecting symptoms such as colds and insomnia.

[0046] For example, the load weight of each candidate department is determined and updated according to the symptom information. Subsequently, the dynamic scheduling agent can perform weighted processing on the department confidence, agent decision delay and department load based on the confidence weight, delay weight and updated load weight, so that emergency symptoms can be more quickly assigned to candidate departments with less load and reduce the user's waiting time.

[0047] S270: Process patient guidance feedback based on the target department through a user interaction agent.

[0048] In one possible embodiment, before the multi-agent collaborative triage method provided in this application performs triage feedback processing based on the target department through the user interaction agent, it can also perform sentiment analysis processing on the user's consultation data through the sentiment computing agent to obtain the user's sentiment information.

[0049] For example, after obtaining user consultation data through a user interaction agent, the user consultation data can be submitted to an emotion computing agent, which will then perform emotion analysis on the user consultation data based on a trained emotion analysis model (e.g., emotion analysis based on text and / or audio spectrum features) to obtain user emotion information.

[0050] Based on this, when processing patient guidance feedback through a user interaction agent based on the target department, the feedback speed can be determined by the user interaction agent based on the user's emotional information. For example, the more anxious the user's emotional information reflects, the higher the corresponding feedback speed. User emotional information can be divided into different levels of anxiety, and corresponding feedback speeds can be set for different levels of anxiety.

[0051] After determining the feedback speed, the user interaction agent can process the triage feedback according to the feedback speed and the target department. For example, when the user is anxious, the user interaction agent can present the triage feedback text recommending the target department to the user at a faster presentation speed (such as voice playback speed), fully considering the user's real-time emotions and improving the user experience.

[0052] In one possible embodiment, the multi-agent collaborative triage method provided in this application, after performing sentiment analysis on user consultation data through an emotion computing agent to obtain user emotional information, can further refine the triage feedback processing based on the target department through a user interaction agent. This can be achieved by the user interaction agent determining the feedback emotional information based on the user's emotional information and then processing the triage feedback based on the feedback emotional information and the target department. For example, when the user's emotional information reflects a more anxious mood, the user interaction agent plays the triage feedback text recommending the target department in a more anxious tone and at a faster pace; when the user's emotional information reflects a calmer mood, the user interaction agent plays the triage feedback text recommending the target department in a more gentle tone and at a slower pace. By providing different expectations and speaking speeds to the target department based on different user emotional information, the method fully considers the user's real-time emotions and improves the user experience.

[0053] In one embodiment, this application can guide users through a preset digital avatar (e.g., a virtual character, virtual animal, etc.). The user interaction agent can generate guidance feedback text based on user consultation data, user response data, target department, agent decision delay of each candidate department, and department load, and drive the preset digital avatar (e.g., driven by Unity3D) to present the guidance feedback text. For example, the user interaction agent can drive the preset digital avatar to read aloud, "Based on your symptoms, gastroenterology is the preferred department, but the waiting time is currently high (approximately 2 hours). Urology can also handle related issues, and the waiting time is shorter (approximately 30 minutes). Which would you prefer?"

[0054] In one possible embodiment, the multi-agent collaborative triage method provided in this application, when processing triage feedback based on feedback emotional information and target department, may be as follows: generating triage feedback text and feedback expression information based on the target department and feedback emotional information, and driving a preset digital image to process triage feedback based on the triage feedback text and feedback expression information.

[0055] For example, after performing sentiment analysis on user consultation data through an emotion computing agent to obtain user emotional information, the user interaction agent generates triage feedback text and feedback emoticons based on the target department and the feedback emotional information.

[0056] For example, the user interaction agent inputs feedback emotional information, user consultation data, user response data, and target department into a trained large language model. The large language model is then required to output corresponding emotional feedback text and facial expressions (e.g., happy, anxious, angry, confused, calm) based on the user consultation data, user response data, and target department. The user interaction agent can set preset digital avatar expressions based on the feedback facial expressions and drive the preset digital avatar to read the feedback text. For example, when a user is in pain, the preset digital avatar's expression is furrowed, and the interface displays the department recommendation reason and workload status, such as "Neurology: Match rate 85%, waiting time 1.5 hours." By using different feedback text and facial expressions for different user emotional information, the preset digital avatar can be driven to provide feedback to the target department, fully considering the user's real-time emotions and improving the user experience.

[0057] The above describes a system that acquires user consultation data through a user interaction agent, performs entity extraction and intent analysis on this data to obtain symptom information and user intent. When the user intent is a triage request, a triage decision agent generates a candidate department list based on the symptom information, a medical knowledge agent determines the key diagnostic symptoms for each candidate department, a user interaction agent performs follow-up questioning based on the key diagnostic symptoms to obtain user response data, a medical knowledge agent determines the department confidence level for each candidate department based on the user consultation and response data, a dynamic scheduling agent selects the target department from the candidate department list based on the department confidence level, and a user interaction agent provides triage feedback based on the target department. This system can accurately respond to users' personalized triage requests and improve user experience. Furthermore, by weighting department confidence level, agent decision delay, and department load based on confidence level weight, latency weight, and load weight, a department score is obtained for each candidate department. Based on the department score, a target department that better meets the user's needs is selected from the candidate department list, which can accurately respond to users' personalized triage requests and improve user experience.

[0058] Figure 3 A schematic diagram of a multi-agent collaborative triage device provided in an embodiment of this application is given. (Reference) Figure 3 The multi-agent collaborative triage device includes a triage response module 31, a symptom analysis module 32, a follow-up questioning module 33, a confidence determination module 34, and a triage feedback module 35.

[0059] The system comprises the following modules: a consultation response module 31, which acquires user consultation data through a user interaction agent, performs entity extraction and intent analysis on the user consultation data to obtain symptom information and user intent; a symptom analysis module 32, which, when the user intent is a triage request, generates a candidate department list based on symptom information through a triage decision agent, and determines the key diagnostic symptoms for each candidate department through a medical knowledge agent; a follow-up questioning module 33, which performs follow-up questioning based on key diagnostic symptoms through a user interaction agent to obtain user response data; a confidence determination module 34, which determines the department confidence level of each candidate department through a medical knowledge agent based on user consultation data and user response data; and a triage feedback module 35, which determines the target department from the candidate department list based on department confidence level through a dynamic scheduling agent, and performs triage feedback processing based on the target department through a user interaction agent.

[0060] The above describes a system that acquires user consultation data through a user interaction agent, performs entity extraction and intent analysis on the data to obtain symptom information and user intent. When the user intent is a triage request, a triage decision agent generates a candidate department list based on the symptom information, a medical knowledge agent determines the key diagnostic symptoms for each candidate department, a user interaction agent performs follow-up questioning based on the key diagnostic symptoms to obtain user response data, a medical knowledge agent determines the department confidence level for each candidate department based on the consultation and response data, a dynamic scheduling agent determines the target department from the candidate department list based on the department confidence level, and a user interaction agent provides triage feedback based on the target department. This system can accurately respond to users' personalized triage needs and improve user experience.

[0061] In one possible embodiment, the multi-agent collaborative triage device further includes a department analysis module, which is used to determine the agent decision delay and department load of each candidate department through dynamic agent scheduling. Accordingly, the triage feedback module 35 determines the target department from the candidate department list based on the department confidence level through a dynamic scheduling agent, including: By dynamically scheduling agents, based on confidence weight, delay weight, and load weight, the confidence of departments, agent decision delay, and department load are weighted to obtain the department score of each candidate department. The target department is then determined from the candidate department list based on the department score.

[0062] In one possible embodiment, the multi-agent collaborative triage device further includes a weight update module, which is used for: Determine the workload weight of each candidate department and update the workload weight based on symptom information; Accordingly, the triage feedback module 35 dynamically schedules agents to weight the department's confidence level, agent decision delay, and departmental load based on confidence level weight, delay weight, and load weight, including: By dynamically scheduling agents, the confidence level, agent decision delay, and departmental load are weighted based on confidence weight, delay weight, and updated load weight.

[0063] In one possible embodiment, the multi-agent collaborative triage device further includes an emotion analysis module, which is used for: By using an emotion computing agent to perform emotion analysis on user consultation data, user emotional information can be obtained. Accordingly, the triage feedback module 35 processes triage feedback based on the target department through a user interaction agent, including: The user interaction agent determines the feedback speed based on the user's emotional information, and then processes the triage feedback according to the feedback speed and the target department.

[0064] In one possible embodiment, the triage feedback module 35 performs triage feedback processing based on the target department through a user interaction agent, including: The user interaction agent determines the feedback emotional information based on the user's emotional information, and then processes the triage feedback based on the feedback emotional information and the target department.

[0065] In one possible embodiment, the triage feedback module 35 performs triage feedback processing based on feedback emotional information and the target department, including: Based on the target department and feedback emotional information, generate triage feedback text and feedback emoticons, and drive the preset digital image to process triage feedback based on the triage feedback text and feedback emoticons.

[0066] In one possible embodiment, the multi-agent collaborative triage device further includes a question-and-answer processing module, which is used for: When the user's intent is a question-and-answer request, a medical knowledge agent generates question-and-answer feedback information based on the user's consultation data. The user interaction agent processes the question and answer feedback based on the information provided.

[0067] It is worth noting that in the above-mentioned multi-agent collaborative triage device embodiments, the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy differentiation and are not used to limit the protection scope of the embodiments of this application.

[0068] This application also provides a multi-agent collaborative triage device, which can integrate the multi-agent collaborative triage device provided in this application. Figure 4 This is a schematic diagram of the structure of a multi-agent collaborative triage device provided in an embodiment of this application. (Reference) Figure 4 The multi-agent collaborative triage device includes: an input device 43, an output device 44, a memory 42, and one or more processors 41; the memory 42 is used to store one or more programs; when one or more programs are executed by one or more processors 41, the one or more processors 41 implement the multi-agent collaborative triage method as provided in the above embodiments. The input device 43, output device 44, memory 42, and processors 41 can be connected via a bus or other means. Figure 4 Taking the example of a connection between China and Israel via a bus.

[0069] The memory 42, as a computing device readable storage medium, can be used to store software programs, computer-executable programs, and modules, such as program instructions / modules corresponding to the multi-agent collaborative triage method provided in any embodiment of this application (e.g., the inquiry response module 31, symptom analysis module 32, follow-up questioning processing module 33, confidence determination module 34, and triage feedback module 35 in the multi-agent collaborative triage device). The memory 42 may primarily include a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a function; the data storage area may store data created based on the use of the device, etc. Furthermore, the memory 42 may include high-speed random access memory and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some instances, the memory 42 may further include memory remotely located relative to the processor 41, and these remote memories can be connected to the device via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0070] Input device 43 can be used to receive input digital or character information, and to generate key signal inputs related to user settings and function control of the device. Output device 44 may include display devices such as a display screen.

[0071] The processor 41 executes various functional applications and data processing of the device by running software programs, instructions and modules stored in the memory 42, thereby realizing the above-mentioned multi-agent collaborative triage method.

[0072] The multi-agent collaborative triage device, equipment, and computer provided above can be used to execute the multi-agent collaborative triage method provided in any of the above embodiments, and have corresponding functions and beneficial effects.

[0073] This application embodiment also provides a storage medium for storing computer-executable instructions. When executed by a computer processor, the computer-executable instructions are used to execute the multi-agent collaborative triage method provided in the above embodiment. The multi-agent collaborative triage method includes: obtaining user consultation data through a user interaction agent; performing entity extraction and intent analysis processing on the user consultation data to obtain symptom information and user intent; when the user intent is a triage request, generating a candidate department list based on the symptom information through a triage decision agent, and determining the key diagnostic symptoms of each candidate department in the candidate department list through a medical knowledge agent; performing symptom follow-up processing based on the key diagnostic symptoms through a user interaction agent to obtain user response data; determining the department confidence level of each candidate department through a medical knowledge agent based on the user consultation data and user response data; determining the target department from the candidate department list through a dynamic scheduling agent based on the department confidence level, and performing triage feedback processing based on the target department through a user interaction agent.

[0074] Storage medium – any type of memory device or storage device. The term “storage medium” is intended to include: mounting media, such as CD-ROMs, floppy disks, or magnetic tape devices; computer system memory or random access memory, such as DRAM, DDR RAM, SRAM, EDO RAM, Rambus RAM, etc.; non-volatile memory, such as flash memory, magnetic media (e.g., hard disks or optical storage); registers or other similar types of memory elements, etc. Storage media may also include other types of memory or combinations thereof. Furthermore, storage media may reside in a first computer system in which a program is executed, or may reside in a different second computer system connected to the first computer system via a network (such as the Internet). The second computer system can provide program instructions to the first computer for execution. The term “storage medium” can include two or more storage media that may reside in different locations (e.g., in different computer systems connected via a network). Storage media may store program instructions (e.g., specifically implemented as a computer program) executable by one or more processors.

[0075] Of course, the computer-executable instructions stored in the storage medium provided in the embodiments of this application are not limited to the multi-agent collaborative triage method provided above, but can also perform related operations in the multi-agent collaborative triage method provided in any embodiment of this application.

[0076] The multi-agent collaborative triage device, equipment, and storage medium provided in the above embodiments can execute the multi-agent collaborative triage method provided in any embodiment of this application. For technical details not described in detail in the above embodiments, please refer to the multi-agent collaborative triage method provided in any embodiment of this application.

[0077] The above description is merely a preferred embodiment and the technical principles employed in this application. This application is not limited to the specific embodiments provided herein, and various obvious changes, readjustments, and substitutions that can be made by those skilled in the art will not depart from the scope of protection of this application. Therefore, although this application has been described in detail through the above embodiments, this application is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of this application, the scope of which is determined by the scope of the claims.

Claims

1. A method for guiding a patient by multi-agent cooperation, characterized by, The method comprises the following steps: obtaining user consultation data through a user interaction agent, performing entity extraction processing and intent analysis processing on the user consultation data to obtain symptom information and user intent; in the case that the user intent is a triage request, generating a candidate department list according to the symptom information through a triage decision agent, and determining the key discriminant symptoms of each candidate department in the candidate department list through a medical knowledge agent; performing symptom follow-up processing according to the key discriminant symptoms through the user interaction agent to obtain user reply data; determining the department confidence of each candidate department according to the user consultation data and the user reply data through the medical knowledge agent; determining a target department from the candidate department list according to the department confidence through a dynamic scheduling agent, and performing guidance feedback processing according to the target department through the user interaction agent.

2. The method of claim 1, wherein, Before the step of determining a target department from the candidate department list according to the department confidence through a dynamic scheduling agent, the method further comprises the following steps: determining the agent decision delay and department load of each candidate department through a dynamic scheduling agent; Accordingly, the step of determining a target department from the candidate department list according to the department confidence through a dynamic scheduling agent comprises the following steps: performing weighted processing on the department confidence, the agent decision delay and the department load based on the confidence weight, the delay weight and the load weight through a dynamic scheduling agent to obtain the department score of each candidate department, and determining a target department from the candidate department list according to the department score.

3. The method of claim 2, wherein, Before the step of performing weighted processing on the department confidence, the agent decision delay and the department load based on the confidence weight, the delay weight and the load weight, the method further comprises the following steps: determining the load weight of each candidate department, and updating the load weight according to the symptom information; Accordingly, the step of performing weighted processing on the department confidence, the agent decision delay and the department load based on the confidence weight, the delay weight and the load weight comprises the following steps: performing weighted processing on the department confidence, the agent decision delay and the department load based on the confidence weight, the delay weight and the updated load weight through a dynamic scheduling agent.

4. The multi-agent cooperative guidance method according to claim 1, wherein, Before the step of performing guidance feedback processing according to the target department through the user interaction agent, the method comprises the following steps: performing sentiment analysis processing on the user consultation data through a sentiment computing agent to obtain user sentiment information; Accordingly, the step of performing guidance feedback processing according to the target department through the user interaction agent comprises the following steps: determining a feedback speed according to the user sentiment information through the user interaction agent, and performing guidance feedback processing according to the feedback speed and the target department.

5. The multi-agent cooperative guidance method according to claim 1, wherein, Before the step of performing guidance feedback processing according to the target department through the user interaction agent, the method comprises the following steps: performing sentiment analysis processing on the user consultation data through a sentiment computing agent to obtain user sentiment information; Correspondingly, the guiding feedback processing according to the target department through the user interaction agent comprises: The feedback emotion information is determined according to the user emotion information through the user interaction agent, and the guiding feedback processing is performed according to the feedback emotion information and the target department.

6. The multi-agent cooperative guidance method according to claim 5, wherein, The guiding feedback processing according to the feedback emotion information and the target department comprises: The guiding feedback text and the feedback expression information are generated according to the target department and the feedback emotion information, and the preset digital image is driven to perform the guiding feedback processing according to the guiding feedback text and the feedback expression information.

7. The multi-agent cooperative guidance method according to claim 1, wherein, After the entity extraction processing and the intent analysis processing on the user inquiry data are performed to obtain the symptom information and the user intent, the method further comprises: In the case that the user intent is a question and answer request, the question and answer feedback information is generated according to the user inquiry data through a medical knowledge agent; The question and answer feedback processing is performed according to the question and answer feedback information through the user interaction agent.

8. A multi-agent cooperative guiding device, characterized by, The method comprises an inquiry response module, a symptom analysis module, a follow-up question processing module, a confidence determination module, and a guiding feedback module, wherein: The inquiry response module is configured to acquire user inquiry data through a user interaction agent, perform entity extraction processing and intent analysis processing on the user inquiry data, and obtain symptom information and a user intent; The symptom analysis module is configured to, in the case that the user intent is a triage request, generate a candidate department list according to the symptom information through a triage decision agent, and determine key discriminant symptoms of each candidate department in the candidate department list through a medical knowledge agent; The follow-up question processing module is configured to perform symptom follow-up question processing according to the key discriminant symptoms through the user interaction agent to acquire user reply data; The confidence determination module is configured to determine department confidence degrees of each candidate department according to the user inquiry data and the user reply data through the medical knowledge agent; The guiding feedback module is configured to determine a target department from the candidate department list according to the department confidence degrees through a dynamic scheduling agent, and perform guiding feedback processing according to the target department through the user interaction agent.

9. A multi-agent cooperative guiding device, characterized by, Comprise: a memory and one or more processors; The memory is configured to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the multi-agent cooperative guiding method according to any one of claims 1-7.

10. A storage medium storing computer-executable instructions, wherein: The computer executable instructions, when executed by a computer processor, are configured to perform the multi-agent cooperative guiding method according to any one of claims 1-7. The computer executable instructions, when executed by a computer processor, are configured to perform the multi-agent cooperative guiding method according to any one of claims 1-7.

Images