Multi-agent based interaction method and multi-agent based interaction system

Abstract

The embodiment of the present application discloses a kind of based on multi-agent interaction method and based on multi-agent interaction system.The terminal of the embodiment of the present application sends inquiry information to server, and the health consultation agent of server end determines the processing mode of current consultation round according to inquiry information, if it is determined that current processing mode is end health consultation, drug recommendation agent obtains the medication suggestion scheme of user according to the inquiry content and inquiry information of preceding consultation round, and then information integration agent obtains the health consultation feedback information including the drug recommendation instruction corresponding to medication suggestion scheme according to medication suggestion scheme, and sends health consultation feedback information to terminal.Health consultation feedback information is displayed after terminal receives health consultation feedback information.The embodiment of the present application can recommend drug to user based on the mode of multi-agent interaction, so it can improve the drug selection efficiency of user when user can not judge required drug, improve the drug purchase experience of user.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer, in particular to an interaction method based on multi-agent and an interaction system based on multi-agent. BACKGROUND

[0002] With the continuous development of computer technology and Internet technology, users can purchase medicines through online channels or offline channels. When purchasing medicines through online channels, users usually need to judge which type of medicine to purchase according to their own symptoms, but if the user cannot clearly judge which type of medicine to purchase, they usually need to spend some time consulting a doctor about the medication plan before purchasing the appropriate medicine, which will reduce the user's shopping experience. SUMMARY

[0003] Therefore, the embodiments of the present application provide an interaction method based on multi-agent and an interaction system based on multi-agent to recommend medicines to users in a multi-agent interaction manner, thereby improving the user's medicine selection efficiency and shopping experience when the user cannot judge the required medicine.

[0004] In a first aspect, the embodiments of the present application provide an interaction method based on multi-agent, applicable to a server, comprising:

[0005] In response to receiving the current inquiry information of the target user in the current consultation round, the health consultation agent determines the processing mode of the current consultation round according to the current inquiry information, the processing mode including ending health consultation, outputting question information or determining a drug suitability result, the drug suitability result being used to represent whether the target user is suitable for medication;

[0006] In response to the processing mode including ending health consultation, the medicine recommendation agent obtains a medication recommendation scheme of the target user according to the inquiry content of the previous consultation round and the current inquiry information;

[0007] In response to obtaining the medication recommendation scheme, the information integration agent obtains health consultation feedback information according to the medication recommendation scheme, the health consultation feedback information including a medicine recommendation explanation corresponding to the medication recommendation scheme;

[0008] Sending the health consultation feedback information to a terminal.

[0009] In a second aspect, the embodiments of the present application provide an interaction method based on multi-agent, applicable to a terminal, comprising:

[0010] Sending the current inquiry information;

[0011] In response to receiving the health consultation feedback information of the current inquiry, the health consultation feedback information is displayed. The health consultation feedback information is obtained by the information integration agent based on the target user's medication recommendation plan. The medication recommendation plan is obtained by the drug recommendation agent based on the inquiry content of the previous consultation round and the current inquiry information. The inquiry content is determined according to the processing method of the current consultation round. The processing method is determined by the health consultation agent based on the current inquiry information. The processing method includes ending the health consultation, outputting question information, or determining the medication suitability result. The medication suitability result is used to characterize whether the target user is suitable for medication. The health consultation feedback information includes the drug recommendation description corresponding to the medication recommendation plan.

[0012] Thirdly, embodiments of the present invention provide a multi-agent-based interaction method, the method comprising:

[0013] Send the current query information;

[0014] In response to receiving the current inquiry information from the target user in the current consultation round, the health consultation agent determines the processing method for the current consultation round based on the current inquiry information. The processing method includes ending the health consultation, outputting the question information, or determining the medication suitability result. The medication suitability result is used to characterize whether the target user is suitable for medication.

[0015] In response to the processing method including ending the health consultation, the drug recommendation agent obtains the target user's medication recommendation plan based on the inquiry content of the previous consultation rounds and the current inquiry information;

[0016] In response to obtaining the medication recommendation plan, the information integration agent obtains health consultation feedback information based on the medication recommendation plan, and the health consultation feedback information includes drug recommendation instructions corresponding to the medication recommendation plan;

[0017] Send the health consultation feedback information to the terminal;

[0018] In response to receiving the current inquiry information, the health consultation feedback information is displayed.

[0019] Fourthly, embodiments of the present invention provide a multi-agent-based interactive system, comprising:

[0020] The server is configured to, in response to receiving current inquiry information from a target user in the current consultation round, determine the processing method for the current consultation round based on the current inquiry information. If the processing method includes ending the health consultation, a drug recommendation agent obtains a medication recommendation plan for the target user based on the inquiry content of previous consultation rounds and the current inquiry information. Upon obtaining the medication recommendation plan, an information integration agent obtains health consultation feedback information based on the medication recommendation plan and sends the health consultation feedback information to the terminal. The processing method includes ending the health consultation, outputting question information, or determining a medication suitability result. The medication suitability result is used to characterize whether the target user is suitable for medication. The health consultation feedback information includes a drug recommendation description corresponding to the medication recommendation plan.

[0021] The terminal is configured to send current inquiry information and, in response to receiving the current inquiry information, display health consultation feedback information.

[0022] Fifthly, embodiments of the present invention provide a multi-agent-based interactive device, suitable for servers, the device comprising:

[0023] The processing method determination unit is used to respond to receiving the current inquiry information of the target user in the current consultation round. The health consultation agent determines the processing method of the current consultation round based on the current inquiry information. The processing method includes ending the health consultation, outputting the question information, or determining the medication suitability result. The medication suitability result is used to characterize whether the target user is suitable for medication.

[0024] The solution determination unit is used to respond to the processing method including ending the health consultation, and the drug recommendation agent obtains the medication recommendation solution for the target user based on the inquiry content of the previous consultation rounds and the current inquiry information;

[0025] The feedback information determination unit is used to respond to the acquisition of the medication recommendation plan. The information integration agent obtains health consultation feedback information based on the medication recommendation plan. The health consultation feedback information includes drug recommendation instructions corresponding to the medication recommendation plan.

[0026] The information sending unit is used to send the health consultation feedback information to the terminal.

[0027] Sixthly, embodiments of the present invention provide a multi-agent-based interactive device, suitable for a terminal, the device comprising:

[0028] The information sending unit is used to send the current query information;

[0029] An information receiving unit is configured to respond to receiving health consultation feedback information of the current inquiry information, and display the health consultation feedback information. The health consultation feedback information is obtained by an information integration agent based on the target user's medication recommendation plan. The medication recommendation plan is obtained by a drug recommendation agent based on the inquiry content of previous consultation rounds and the current inquiry information. The inquiry content is determined according to the processing method of the current consultation round. The processing method is determined by the health consultation agent based on the current inquiry information. The processing method includes ending the health consultation, outputting question information, or determining the medication suitability result. The medication suitability result is used to characterize whether the target user is suitable for medication. The health consultation feedback information includes drug recommendation instructions corresponding to the medication recommendation plan.

[0030] In a seventh aspect, embodiments of the present invention provide an electronic device, including a memory and a processor, the memory being used to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the processor to implement the method as described in any one of the first to third aspects.

[0031] Eighthly, embodiments of the present invention provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method as described in any one of the first to third aspects.

[0032] In a ninth aspect, embodiments of the present invention provide a computer program product comprising a computer program / instructions that, when executed by a processor, implement the method as described in any one of the first to third aspects.

[0033] In this embodiment of the invention, after the terminal sends inquiry information to the server, the server-side health consultation agent determines the processing method for the current consultation round based on the inquiry information. If the current processing method is determined to be ending the health consultation, the drug recommendation agent obtains the user's medication recommendation plan based on the inquiry content and information from previous consultation rounds. Then, the information integration agent obtains health consultation feedback information, including drug recommendation instructions corresponding to the medication recommendation plan, based on the medication recommendation plan, and sends the health consultation feedback information to the terminal. After receiving the health consultation feedback information, the terminal displays it. This embodiment of the invention can recommend drugs to users based on a multi-agent interaction method, thus improving the efficiency of drug selection and enhancing the user's drug purchase experience when the user cannot determine the required drug. Attached Figure Description

[0034] The above and other objects, features and advantages of the present invention will become clearer from the following description of embodiments of the invention with reference to the accompanying drawings, in which:

[0035] Figure 1 This is a schematic diagram of the hardware architecture of a multi-agent interactive system according to an embodiment of the present invention;

[0036] Figure 2 This is a structural block diagram of the intelligent interactive system according to an embodiment of the present invention;

[0037] Figure 3 This is a flowchart of a multi-agent-based interaction method according to an embodiment of the present invention;

[0038] Figure 4 This is a flowchart of a multi-agent-based interaction method according to an embodiment of the present invention;

[0039] Figure 5 This is a flowchart of a multi-agent-based interaction method according to an embodiment of the present invention;

[0040] Figure 6 This is a schematic diagram illustrating the process of the intelligent interactive system sending messages to the terminal according to an embodiment of the present invention;

[0041] Figure 7 This is a flowchart of a multi-agent-based interaction method according to an embodiment of the present invention;

[0042] Figure 8 This is a flowchart of a multi-agent-based interaction method according to an embodiment of the present invention;

[0043] Figure 9 This is a flowchart of a multi-agent-based interaction method according to an embodiment of the present invention;

[0044] Figure 10 This is a flowchart of a multi-agent-based interaction method according to an embodiment of the present invention;

[0045] Figures 11-12 This is a schematic diagram of the interaction process of multiple agents according to an embodiment of the present invention;

[0046] Figure 13 This is a flowchart of a multi-agent-based interaction method according to an embodiment of the present invention;

[0047] Figure 14 This is a flowchart of a multi-agent-based interaction method according to an embodiment of the present invention;

[0048] Figure 15 This is a flowchart of the multi-agent-based interaction method of this invention on the server side;

[0049] Figure 16 This is a flowchart of the multi-agent-based interaction method of this invention on the terminal side;

[0050] Figure 17This is a schematic diagram of a multi-agent-based interactive device according to an embodiment of the present invention;

[0051] Figure 18 This is a schematic diagram of a multi-agent-based interactive device according to an embodiment of the present invention;

[0052] Figure 19 This is a schematic diagram of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0053] The present application is described below based on embodiments, but it is not limited to these embodiments. In the detailed description of the present application below, certain specific details are described in detail. Those skilled in the art can fully understand the present application without these details. To avoid obscuring the substance of the present application, well-known methods, processes, flows, elements, and circuits are not described in detail.

[0054] Furthermore, those skilled in the art should understand that the accompanying drawings provided herein are for illustrative purposes only and are not necessarily drawn to scale.

[0055] Unless the context explicitly requires it, words such as "including" or "contains" throughout the application should be interpreted as including rather than exclusive or exhaustive; that is, meaning "including but not limited to".

[0056] In the description of this application, it should be understood that the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance. Furthermore, in the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0057] The technical solution of this invention can be applied to the transaction and delivery services of instant e-commerce platforms, such as Taobao Flash Sale, Taoxianda, Ele.me food delivery, and retail. The technical solution of this invention can combine user input and provide users with non-diagnostic medical knowledge consultation and medication safety assistance services based on multi-agent interaction. However, the content output by the intelligent interaction system should not be used as the basis for disease diagnosis or treatment or any legally valid written proof, and cannot replace the judgment of a doctor.

[0058] Online drug purchases offer significant convenience to users. To facilitate targeted medication selection, some instant e-commerce platforms provide online doctor consultations. When buying medication online, if users are unsure which type of drug to purchase, they can consult a doctor for medication advice. However, this can be time-consuming due to a high volume of simultaneous inquiries, requiring users to wait for their consultation before purchasing the appropriate medication, thus negatively impacting the overall drug purchase experience. Conversely, some instant e-commerce platforms do not offer online doctor consultations. If users lack basic medical knowledge, they may struggle to purchase the correct medication. Therefore, existing online drug purchase methods can negatively impact the user experience.

[0059] To address the aforementioned issues, this invention proposes a multi-agent-based interaction method and a multi-agent-based interaction system. This method recommends medications to users through multi-agent interaction, thereby improving the efficiency of medication selection and enhancing the user's medication purchase experience when the user is unable to determine the required medication.

[0060] Figure 1 This is a schematic diagram of the hardware architecture of a multi-agent interactive system according to an embodiment of the present invention. Figure 1 As shown, the multi-agent interactive system of this invention includes a server 11 and a terminal 12 on the real-time e-commerce platform side.

[0061] Server 11 and terminal 12 can establish a communication connection through a network or other communication methods to achieve information and data exchange. It should be understood that, although... Figure 1 Only a certain number of servers 11 and terminals 12 are shown, but this does not mean that the number of each is limited. This system can contain multiple servers and multiple terminals.

[0062] Server 11 should be understood as a device that provides data processing, database, and communication facilities. For example, server 11 may refer to a single physical server with associated communication, data storage, and database facilities, or it may refer to a networked or aggregated collection of processors, associated networks, and storage devices that operate software and one or more database systems and application software supporting the services provided by the server. Server 11 may be a monolithic server or a distributed server spanning multiple computers or computer data centers, or it may be various types of cloud servers. In some embodiments, each server may include hardware, software, or embedded logical components for performing suitable functions supported or implemented by the server, or a combination of two or more such components.

[0063] Terminal 12 is a communication terminal capable of running computer programs. These communication terminals can be mobile phones, tablets, PDAs, wearable devices, vehicle-mounted terminals, etc. Terminal 12 has a communication module capable of wired or wireless communication. In some embodiments, terminal 12 includes at least one remote communication module, such as a communication circuit for WLAN, GPRS, or 2G / 3G / 4G / 5G remote communication. Terminal 12 also has a display device and an input device. The display device can be a liquid crystal display, an LED display, or a projection device. The input device can include, for example, a touchscreen, buttons, a pressure sensor, etc. Terminal 12 receives user commands through the input device and interacts with the user through the display device.

[0064] This invention provides an embodiment of an intelligent interactive system to assist users in completing the medication purchase process. The intelligent interactive system 20 is deployed on server 11 and includes multiple agents, each capable of autonomously exchanging information based on a pre-set interaction sequence.

[0065] Figure 2 This is a structural block diagram of the intelligent interactive system according to an embodiment of the present invention. Figure 2 As shown, the intelligent interaction system of this embodiment may include a health consultation intelligent agent 21, a drug recommendation intelligent agent 22, and an information integration intelligent agent 23. After receiving the user's current inquiry information in the current consultation round, the health consultation intelligent agent 21 determines the processing method for the current consultation round based on the current inquiry information. The processing method includes determining a health consultation judgment result, outputting question information, or determining a medication suitability result. The health consultation judgment result indicates whether the health consultation is terminated, and the medication suitability result indicates whether the target user is suitable for medication. When the processing method determined by the health consultation intelligent agent 21 includes determining a health consultation judgment result, and the health consultation judgment result is to terminate the health consultation, the drug recommendation intelligent agent 22 obtains the user's medication recommendation plan based on the inquiry content of the previous consultation round and the current inquiry information. After obtaining the user's medication recommendation plan, the information integration intelligent agent 23 obtains health consultation feedback information based on the user's medication recommendation plan. The health consultation feedback information includes drug recommendation instructions corresponding to the medication recommendation plan. Therefore, the server 11 can send health consultation feedback information to the terminal 12.

[0066] In an optional implementation of this invention, the intelligent interaction system 20 may further include an intent recognition agent (not shown in the figure) and an inquiry agent (not shown in the figure). The intent recognition agent can determine whether the target user is initiating an inquiry for the first time. If it is initiating an inquiry for the first time, the user can choose to interact with the inquiry agent; if it is not initiating an inquiry for the first time, the user can choose to interact with the health consultation agent 21. The inquiry agent can determine the user's symptom overview information and initial medication recommendation information based on the current inquiry information. The health consultation agent 21 can determine the processing method based on the current inquiry information and the content of previous rounds of inquiries. The inquiry content may include the user's symptom overview information and initial medication recommendation information.

[0067] In an optional implementation of this invention, the intelligent interaction system 20 may further include a qualification review agent (not shown in the figure). The qualification review agent can obtain the medication recommendation plan output by the drug recommendation agent 22 and determine the drug sales approval information corresponding to the medication recommendation plan. If the drug sales approval information indicates that the user is qualified to purchase, the qualification review agent can provide the medication recommendation plan output by the drug recommendation agent 22 to the information integration agent 23; if the drug sales approval information indicates that the user is not qualified to purchase, the qualification review agent can provide feedback to the drug recommendation agent 22. The drug recommendation agent 22 can obtain the user's medication recommendation plan again based on the inquiry content of previous consultation rounds and the current inquiry information.

[0068] In an optional implementation of this invention, the intelligent interaction system 20 may further include a human health consultation agent (not shown in the figure). The health consultation agent 21 may choose to interact with the human health consultation agent when the processing method includes determining a health consultation judgment result and determining a medication suitability result, and the health consultation judgment result indicates that the health consultation will not end, while the medication suitability result indicates that the user is not suitable for medication. The human health consultation agent can obtain human consultation prompts based on the current inquiry information. Therefore, the server 11 can send human consultation prompts to the terminal 12.

[0069] In an optional implementation of this invention, if the processing method includes determining the health consultation judgment result and determining the medication suitability result, and the health consultation judgment result indicates that the health consultation is not terminated, and the medication suitability result indicates that the target user is suitable for medication, the health consultation agent 21 can determine that the current consultation round has ended.

[0070] In an optional implementation of this invention, if the processing method is to output question information, the health consultation intelligent agent 21 can determine the current question information for the current round. Therefore, the server 11 can send the current question information to the terminal 12.

[0071] In one optional implementation of this invention, the intelligent interaction system 20 can call a predetermined toolkit to create and update the current question information based on the information type of the current question information.

[0072] Therefore, in this embodiment of the invention, after receiving the health consultation feedback information for the current round, the terminal 12 can display the health consultation feedback information.

[0073] In an optional implementation of this invention, after receiving the current question information for the current round, the terminal 12 may display the current question information.

[0074] In one optional implementation of this invention, after receiving a prompt for manual consultation, the terminal 12 may display the prompt.

[0075] In an optional implementation of this invention, the health consultation feedback information may include a page redirection control. Terminal 12 may, in response to the page redirection control being triggered, display a drug purchase page corresponding to the health consultation feedback information, the drug purchase page including an order submission control. Terminal 12 may also, in response to the order submission control being triggered, send an order submission request based on the drug selection information.

[0076] In an optional implementation of this invention, terminal 12 may, in response to the inclusion of prescription drugs in the drug purchase information, send a prescription issuance request based on the target user's symptom overview information and the drug identifier of the prescription drug. Terminal 12 may also, in response to receiving a target prescription, send an order submission request based on the drug purchase information and the target prescription, where the target prescription is the prescription corresponding to the prescription drug.

[0077] The following describes the method through examples. Figure 3 This is a flowchart of a multi-agent-based interaction method according to an embodiment of the present invention. Figure 3 As shown, the method in this embodiment includes the following steps:

[0078] Step S301: Send the current query information.

[0079] In this embodiment, the instant e-commerce platform can provide access to the intelligent interactive system to the terminal through client applications, mini-programs, or other means, such as jump controls. Taking a client application as an example, after the user triggers the jump control of the intelligent interactive system displayed in the client application, the terminal can jump to the interactive page of the intelligent interactive system and input query information through the interactive page. Therefore, in this step, the terminal can determine the query information input by the user in the interactive page of the intelligent interactive system as the current query information and send the current query information to the server.

[0080] Step S302: In response to receiving the current inquiry information from the target user in the current consultation round, the health consultation agent determines the processing method for the current consultation round based on the current inquiry information.

[0081] In this embodiment, the intelligent interaction system can realize the interaction process between at least some agents in the intelligent interaction system based on the query information of the target user in each consultation round. After receiving the current query information of the target user, the server can transmit the current query information to the intelligent interaction system through various existing communication methods, such as local inter-process communication, local transmission control protocol (TCP), remote procedure call framework (RPC), etc.

[0082] Therefore, in this step, the health consultation agent can determine the processing method for the current consultation round based on the current inquiry information received from the target user in the current consultation round.

[0083] In this embodiment, the health consultation agent can be implemented based on a large language model. The large language model used to implement the health consultation agent can be fine-tuned based on a first training sample set. The first training sample set can be determined based on doctor-patient dialogue information. The first training sample set can include health inquiry texts input by different users in each round of the online health consultation process, contextual information of the health inquiry texts, and corresponding health question texts and / or health consultation judgment tags and / or medication judgment tags. Specifically, if the doctor-patient dialogue information includes the health question text corresponding to the health inquiry text, the processing method can be determined to include outputting the question information; otherwise, it is determined not to include outputting the question information. If the doctor recommends medication based on the user's health inquiry text in the doctor-patient dialogue information, the health consultation judgment tag can be determined to indicate the end of the health consultation. If the doctor gives a medical advice that does not require medication based on the user's health inquiry text in the doctor-patient dialogue information, such as suggesting further examination, the health consultation judgment tag can be determined to indicate that the health consultation is not ended, and the medication judgment tag indicates that medication is not suitable. During the fine-tuning of the large language model used to implement the health consultation agent, the input of the large language model can be determined based on each health inquiry text, or each health inquiry text and the corresponding context information. The training objective of the large language model used to implement the health consultation agent can be determined based on the corresponding health question text / or health consultation judgment label and / or medication judgment label, until the large language model used to implement the health consultation agent reaches the corresponding training termination condition, such as loss function convergence, accuracy reaching a preset threshold, etc. This embodiment does not impose any restrictions on this.

[0084] The health consultation agent can determine the prompt words of the large language model based on the current inquiry information of the target user in the current consultation round, and determine the processing method for the current consultation round based on the large language model.

[0085] The processing method includes at least one of the following: determining the health consultation assessment result, outputting question information, and determining the medication suitability result. The health consultation assessment result indicates whether the health consultation has ended, and the medication suitability result indicates whether the target user is suitable for medication.

[0086] Optionally, to ensure that the health consultation agent can fully understand the target user's chief complaint (i.e., the target user's main symptoms, signs, and symptom duration), this embodiment can also score the large language model used to implement the health consultation agent, and fine-tune the large language model based on the score. The scoring rules corresponding to the health consultation agent can be set according to actual needs, and this embodiment does not impose any restrictions on this. In one optional implementation, the scoring rules corresponding to the health consultation agent can be referred to in the following table:

[0087]

[0088] Figure 4 This is a flowchart of a multi-agent-based interaction method according to an embodiment of the present invention. Figure 4 As shown, in an optional implementation of this embodiment, step S302 may include the following steps:

[0089] Step S401: In response to receiving the current query information, the intent recognition agent determines the corresponding intent recognition result based on the current query information.

[0090] Before determining the processing method for the current consultation round based on the current inquiry information, the intelligent agent of the health consultation system can first judge the intent of the target user to determine whether the target user is initiating an inquiry for the first time.

[0091] In this embodiment, the intent recognition agent can be implemented based on an intent recognition model, such as a deep learning-based recurrent neural network, transformer, long short-term memory artificial neural network, or other types of machine learning models. The model used to implement the intent recognition agent can be trained on a second training sample set, which may include multiple health inquiry texts and corresponding intent tags for each health inquiry text. During the training process of the model used to implement the intent recognition agent, the input to the model can be determined based on each health inquiry text, and the training objective can be determined based on the corresponding intent tags, until the model used to implement the intent recognition agent reaches the corresponding training termination condition.

[0092] Therefore, in this step, the intent recognition agent can determine the input of the corresponding model based on the current inquiry information of the target user in the current consultation round and the inquiry content in the previous consultation round, and determine the intent recognition result of the current inquiry information based on the corresponding model.

[0093] In this embodiment, the target user initiating an inquiry for the first time indicates that the target user's current inquiry message demonstrates a clear intent for the first time in this dialogue. More specifically, it indicates that the target user's current inquiry message demonstrates a clear intent for health consultation for the first time in this dialogue. For example, if the target user's current inquiry message is "I'm feeling a little unwell," the intent recognition agent can determine that the current inquiry message has a clear intent. Furthermore, it can combine the intent recognition results of inquiries from previous consultation rounds to determine whether this clear intent is the first time the target user has expressed it in this dialogue.

[0094] Step S402: Determine whether the target user is initiating the inquiry for the first time.

[0095] If it is determined that the target user is initiating the inquiry for the first time, step S403 can be executed; otherwise, step S404 can be executed.

[0096] Step S403: The query agent determines the target user's symptom overview information and initial medication recommendation information based on the current query information.

[0097] If the intelligence agent determines that the target user is initiating an inquiry for the first time, it can pass the current inquiry information to the inquiry intelligence agent. Therefore, in this step, the inquiry intelligence agent can combine the current inquiry information with the target user's historical health consultation information, historical medical records, historical medication purchase records, etc., to determine the target user's symptom overview and initial medication recommendations.

[0098] The query agent only needs to perform a preliminary screening of drugs based on the user's symptoms. Therefore, in this embodiment, the initial medication recommendation information for the target user may include information on various drugs that match the target user's symptoms.

[0099] In this embodiment, the query agent can also be implemented based on a large language model. The large language model used to implement the query agent can be fine-tuned based on a third training sample set. This third training sample set may include multiple health query texts, summary information corresponding to each health query text, and a set of candidate drugs corresponding to the summary information. The set of candidate drugs can be determined based on the summary information and a medical knowledge graph. During the fine-tuning of the large language model used to implement the query agent, the input to the large language model can be determined based on each health query text, and the training objective of the large language model can be determined based on the corresponding summary information and the set of candidate drugs, until the large language model used to implement the query agent reaches the corresponding training termination condition.

[0100] Therefore, in this step, the query agent can determine the prompt words of the large language model based on the current query information of the target user in the current consultation round, and determine the target user's symptom overview and initial medication advice information.

[0101] Step S404: The health consultation AI determines the processing method based on the current inquiry information and content.

[0102] If the intent-based intelligent agent determines that the target user is not initiating an inquiry for the first time, it can pass the current inquiry information to the health consultation intelligent agent. Simultaneously, the inquiry intelligent agent can also pass the content of previous consultation rounds, including the target user's symptom overview and initial medication recommendations, to the health consultation intelligent agent. Therefore, in this step, the health consultation intelligent agent can determine the prompt words for the large language model based on the content of previous consultation rounds and the current inquiry information, and determine the processing method for the current consultation round based on the large language model.

[0103] In step S303, in response to the processing method including determining the health consultation judgment result, and the health consultation judgment result being the end of health consultation, the drug recommendation agent obtains the target user's medication recommendation plan based on the inquiry content of the previous consultation rounds and the current inquiry information.

[0104] In this embodiment, if the health consultation agent determines that the processing method for the current consultation round includes determining the health consultation judgment result, and the health consultation judgment result is to end the health consultation, it means that based on the information provided by the target user, the target user's physical condition can be clearly inferred, and there is no need to conduct further health consultation with the target user. Furthermore, suitable medication can be recommended to the target user. Therefore, the health consultation agent can transmit the content of the previous round's inquiries and the current inquiry information to the drug recommendation agent.

[0105] In this step, the drug recommendation agent can obtain medication recommendations for the target user based on the content of previous consultation rounds and the current consultation information. These recommendations may include the name of the recommended drug and the recommended method of administration.

[0106] When determining recommended medications in a medication recommendation plan, the medication recommendation agent can assess the rationality of each medication in the initial medication recommendation information to ensure medication safety for the target user. Rationality assessments may include, but are not limited to: contraindications for special populations, the target user's historical allergy information, and whether there are any drug incompatibilities. Specifically, contraindications for special populations include the prohibition of certain types of medications for specific groups such as pregnant women, breastfeeding women, children, the elderly, and patients with liver or kidney disease; historical allergy information for the target user refers to any allergic reactions to drug components; and drug incompatibilities refer to interactions between drugs that could affect efficacy or cause toxic reactions.

[0107] In this embodiment, the drug recommendation agent can be implemented based on a large language model. The large language model used to implement the drug recommendation agent can be fine-tuned based on a fourth training sample set. This fourth training sample set may include multiple health inquiry texts, contextual information of each health inquiry text including summary information and a candidate drug set, and the user's actual drug purchase information. During the fine-tuning of the large language model used to implement the drug recommendation agent, the input to the large language model can be determined based on each health inquiry text and its contextual information. The training objective of the large language model can be determined based on the corresponding actual drug purchase information, until the large language model used to implement the drug recommendation agent reaches the corresponding training termination condition.

[0108] Therefore, in this step, the drug recommendation agent can determine the prompt words of the large language model based on the current inquiry information of the target user in the current consultation round and the inquiry content of the previous consultation round, and determine the medication recommendation plan for the target user based on the large language model.

[0109] Optionally, to further ensure medication safety for target users, this embodiment can also score the large language model used to implement the drug recommendation agent, and fine-tune the large language model based on the score. The scoring rules corresponding to the drug recommendation agent can be set according to actual needs, and this embodiment does not impose any restrictions on this. In one optional implementation, the scoring rules corresponding to the drug recommendation agent can be referred to in the following table:

[0110]

[0111] In step S304, in response to obtaining the medication recommendation plan, the information integration agent obtains health consultation feedback information based on the medication recommendation plan.

[0112] After determining the recommended medication plan for the target user, the drug recommendation agent can transmit the inquiry content and medication recommendations from previous consultation rounds to the information integration agent. Therefore, in this step, the information integration agent can obtain the target user's health consultation feedback information based on the inquiry content and medication recommendations from previous consultation rounds.

[0113] In this embodiment, the health consultation feedback information may include drug recommendations corresponding to the medication suggestion plan, helping the target user to better understand and accept the medication suggestion plan. Optionally, the health consultation feedback information may also include an overview of the target user's condition, helping the user to better understand their own physical condition.

[0114] In this embodiment, the information integration agent can be implemented based on a large language model. The large language model used to implement the information integration agent can be fine-tuned based on a fifth training sample set. This fifth training sample set may include multiple health inquiry texts, contextual information for each health inquiry text (including the user's actual medication purchase information), and corresponding health consultation summary information. During the fine-tuning of the large language model used to implement the information integration agent, the input to the large language model can be determined based on each health inquiry text and its contextual information. The training objective of the large language model can be determined based on the corresponding health consultation summary information, until the large language model reaches the corresponding training termination condition.

[0115] Therefore, in this step, the information integration agent can determine the prompt words of the large language model based on the current inquiry information of the target user in the current consultation round and the inquiry content of the previous consultation round, and determine the health consultation feedback information of the target user based on the large language model.

[0116] Optionally, to make the health consultation feedback information output by the information integration agent more suitable for medication recommendations, this embodiment can also score the large language model used to implement the information integration agent, and fine-tune the large language model based on the score. The scoring rules corresponding to the information integration agent can be set according to actual needs, and this embodiment does not impose any restrictions on this. In one optional implementation, the scoring rules corresponding to the information integration agent can be referred to in the following table:

[0117]

[0118] Figure 5 This is a flowchart of a multi-agent-based interaction method according to an embodiment of the present invention. Figure 5 As shown, in an optional implementation of this embodiment, step S304 may further include the following steps:

[0119] In step S501, in response to obtaining the medication recommendation plan, the qualification review agent determines the drug sales information corresponding to the medication recommendation plan.

[0120] In one optional implementation of this embodiment, to further improve the rationality of drug recommendations and reduce the difficulty for target users to purchase drugs, the drug recommendation agent, after determining the medication recommendation plan for the target user, can transmit the inquiry content and medication recommendation plan from previous consultation rounds to the qualification review agent. Therefore, in this step, the qualification review agent can determine the corresponding drug sales approval information based on the medication recommendation plan output by the drug recommendation agent.

[0121] In this embodiment, drug sales access information is used to indicate whether the target user is qualified to purchase the recommended drug. Qualification may include whether the recommended drug is sold in the target user's region, whether the recommended drug is sold on an instant e-commerce platform, and whether the recommended drug is a prohibited drug for online sale.

[0122] Optionally, the eligibility verification agent can be based on a pre-set drug list and matching rules. For example, if the drug list is a list of drugs available for sale on an instant e-commerce platform, the matching rule can be: if the recommended drug matches the drug list, it can be determined that the drug's eligibility to sell indicates that the target user is qualified to purchase the recommended drug; if the drug list contains drugs prohibited from online sale, the matching rule can be: if the recommended drug matches the drug list, it can be determined that the drug's eligibility to sell indicates that the target user is not qualified to purchase the recommended drug.

[0123] In step S502, in response to the drug sales information indicating that the target user is qualified to purchase, the information integration agent obtains health consultation feedback information.

[0124] If the eligibility verification agent determines that the drug's marketability information corresponding to the medication recommendation indicates that the target user is eligible to purchase it, it can transmit the inquiry content and medication recommendation from previous consultation rounds to the information integration agent. Therefore, in this step, the information integration agent can obtain health consultation feedback information.

[0125] Step S305: Send health consultation feedback information to the terminal.

[0126] After obtaining health consultation feedback information, the information integration intelligent agent can send the health consultation feedback information to the terminal through the server.

[0127] In this embodiment, the information integration agent can invoke a predetermined toolkit to create and update health consultation feedback information based on the information type, thereby improving the flexibility of information interaction. The information type of health consultation feedback information may include at least one of streaming, non-streaming, and anomalous information. The predetermined toolkit may be IMSDK (Instant Messaging Software Development Kit).

[0128] Optionally, if the information type is streaming, during the streaming response process, when the large language model of the information integration agent returns streaming output, the stack SDK can trigger the onMessage(content) event handler function based on the streaming output and pass the content as a callback parameter to the information integration agent. After receiving the content, the information integration agent first calls the IMSDK's createMessage method to create an incomplete streaming message and sends the incomplete streaming message to the terminal. Subsequently, the information integration agent can call the IMSDK's updateMessageContent method to incrementally append the content to the incomplete message body and synchronize it to the terminal. After the streaming response ends, the stack SDK will trigger the onAgentFinish event and pass the structured result to the information integration agent. After receiving the onAgentFinish event, the information integration agent can call the IMSDK's updateExt method to insert the structured result into the extended area of ​​the streaming message according to protocol matching rules (such as the mapping between drug recommendation instructions and card templates) and send it to the terminal. If the message type is non-streaming, the Stack SDK can return the complete message (result) as a callback parameter to the message integration agent. Upon receiving the complete message, the agent can call the IMSDK's `createMessage` method to create a message and send the complete message to the terminal. If the message type is error, it indicates that the message integration agent's inference has failed. In this case, the Stack SDK will trigger the `onError` event and pass the error code to the agent. Upon receiving the error code, the agent will generate a user-readable error message through protocol matching, call the IMSDK's `SendErrorMessage` method to create a message, and send the message to the terminal.

[0129] Figure 6 This is a schematic diagram illustrating the process of the intelligent interactive system sending messages to the terminal according to an embodiment of the present invention. Figure 6As shown, if the message type is streaming, and in the streaming response, after the stack SDK (not shown in the figure) triggers the onMessage(content) event handler function and returns the content callback parameter to the intelligent interaction system 61, the intelligent interaction system 61 can first call the createMessage method of IMSDK62 to create an incomplete streaming message, and then call the updateMessageContent method of IMSDK62 to append the content to the incomplete message body and synchronize the message to the terminal 63. If the message type is streaming and the streaming response ends, after the stack SDK triggers the onAgentFinish event and passes the structured result to the intelligent interaction system 61, the intelligent interaction system 61 can call the updateExt method of IMSDK to insert the structured data into the extended area of ​​the streaming message and send it to the terminal. If the message type is non-streaming, after the stack SDK passes the result callback function to the intelligent interaction system 61, the intelligent interaction system 61 can call the createMessage method to create the message and send the complete message to the terminal 63. If the message type is error, the stack SDK triggers the onError event and passes the error code to the intelligent interaction system 61. The intelligent interaction system 61 can generate user-readable error information through protocol matching, and call the IMSDK's SendErrorMessage method to create a message, and then send the message to the terminal.

[0130] Step S306: In response to receiving the health consultation feedback information for the current inquiry, display the health consultation feedback information.

[0131] In this step, after receiving the health consultation feedback information of the current inquiry, the terminal can display the health consultation feedback information on the interactive page of the intelligent interactive system to recommend medicines to the target user.

[0132] Figure 7 This is a flowchart of a multi-agent-based interaction method according to an embodiment of the present invention. Figure 7 As shown, in one optional implementation, the method of this embodiment may further include the following steps:

[0133] In step S701, in response to the drug sales information indicating that the target user is not qualified to purchase, the drug recommendation agent obtains a medication suggestion plan again.

[0134] In one optional implementation of this embodiment, if the eligibility verification agent determines that the drug sales information indicates the target user is not eligible to purchase the drug, it can feed back the verification result to the drug recommendation agent. The drug recommendation agent can then obtain a medication recommendation plan for the target user based on the inquiry content from previous consultation rounds and the current inquiry information.

[0135] It is easy to understand that in this embodiment, step S701 can be executed after step S501.

[0136] Figure 8 This is a flowchart of a multi-agent-based interaction method according to an embodiment of the present invention. Figure 8 As shown, in one optional implementation, the method of this embodiment may further include the following steps:

[0137] Step S801, in response to the processing method including determining the health consultation judgment result and determining the medication suitability result, wherein the health consultation judgment result indicates that the health consultation is not terminated, and the medication suitability result indicates that the target user is not suitable for medication, and the artificial health consultation agent obtains artificial consultation prompt information based on the current inquiry information.

[0138] In one optional implementation of this embodiment, if the processing method output by the health consultation agent includes determining a health consultation judgment result and determining a medication suitability result, and the health consultation judgment result indicates that the health consultation is not terminated, and the medication suitability result indicates that the target user is not suitable for medication, meaning that the target user's condition is difficult to alleviate with medication, or that the target user's condition is highly complex, then the inquiry content from previous consultation rounds and the current inquiry information can be transmitted to the artificial health consultation agent. Therefore, in this step, the artificial health consultation agent can obtain artificial consultation prompts based on the inquiry content from previous consultation rounds and the current inquiry information.

[0139] In this embodiment, the artificial health consultation agent can also be implemented based on a large language model. The large language model used to implement the artificial health consultation agent can be fine-tuned based on a sixth training sample set. The sixth training sample set may include multiple health inquiry texts, contextual information of each health inquiry text including user summary information and candidate drug sets, and corresponding human consultation guidance information. During the fine-tuning of the large language model used to implement the artificial health consultation agent, the input to the large language model can be determined based on each health inquiry text and its contextual information, and the training objective of the large language model can be determined based on the corresponding human consultation guidance information, until the large language model used to implement the artificial health consultation agent reaches the corresponding training termination condition.

[0140] Therefore, in this step, the AI ​​health consultation agent can determine the prompt words of the large language model based on the current inquiry information in the current consultation round, and determine the human consultation prompt information. Optionally, the human consultation prompt information may include a link to the human consultation interaction page.

[0141] It is easy to understand that in this embodiment, step S801 can be executed after step S302.

[0142] Step S802: Send a prompt message for manual consultation to the terminal.

[0143] After receiving the human consultation prompt, the AI ​​agent for health consultation can send the prompt to the terminal via the server. The method for sending the human consultation prompt in this step is similar to that for sending health consultation feedback information, and will not be elaborated further here.

[0144] Step S803: In response to receiving the manual consultation prompt message, display the manual consultation prompt message.

[0145] In this step, after receiving the prompt for human consultation, the terminal can display the prompt on the interactive page of the intelligent interaction system to encourage the user to select human health consultation.

[0146] Figure 9 This is a flowchart of a multi-agent-based interaction method according to an embodiment of the present invention. Figure 9 As shown, in one optional implementation, the method of this embodiment may further include the following steps:

[0147] Step S901: In response to the processing method of determining the health consultation judgment result and determining the medication suitability result, where the health consultation judgment result indicates that the health consultation is not terminated and the medication suitability result indicates that the target user is suitable for medication, the health consultation agent determines that the current consultation round is terminated.

[0148] In one optional implementation of this embodiment, if the processing method output by the health consultation agent includes determining the health consultation judgment result and determining the medication suitability result, and the health consultation judgment result indicates that the health consultation is not over, and the medication suitability result indicates that the target user is suitable for medication, it means that the health consultation agent has not fully understood the target user's symptoms, and therefore it can be determined that the current consultation round has ended.

[0149] It is easy to understand that in this embodiment, step S901 can be executed after step S302.

[0150] Figure 10 This is a flowchart of a multi-agent-based interaction method according to an embodiment of the present invention. Figure 10As shown, in one optional implementation, the method of this embodiment may further include the following steps:

[0151] In step S1001, in response to the processing method of outputting question information, the health consultation agent determines the current question information for the current consultation round.

[0152] In one optional implementation of this embodiment, if the processing method output by the health consultation agent includes outputting question information, the health consultation agent can determine the current question information of the current consultation round based on the current question information and the question content of the previous consultation round, so as to understand the target user's condition through the current question information.

[0153] It is easy to understand that in this embodiment, step S1001 can be executed after step S302.

[0154] Step S1002: Send the current query information to the terminal.

[0155] After obtaining the current question information, the health consultation agent can send the question information to the terminal through the server. The method for sending the current question information in this step is similar to the method for sending health consultation feedback information, and will not be repeated here.

[0156] Step S1003: In response to receiving the current query information, display the current query information.

[0157] In this step, after receiving the current query information, the terminal can display the current query information on the interactive page of the intelligent interaction system to prompt the user to input the corresponding query information based on the current query information.

[0158] Figures 11-12 This is a schematic diagram illustrating the interaction process of multiple agents according to an embodiment of the present invention. It is easy to understand. Figure 11 This is a schematic diagram of the first half of the interaction process. Figure 12 This is a schematic diagram of the latter half of the interaction process, that is... Figure 12 for Figure 11 The continuation part. For example... Figures 11-12As shown, after receiving the current inquiry information from the target user, the intelligent interaction system can use the intent recognition agent 1101 as input, taking the previous inquiry content (i.e., the inquiry content from previous consultation rounds) and the current inquiry information as input, and output the intent recognition result of the current inquiry information, i.e., whether the target user initiated the inquiry for the first time (i.e., the first question). If yes, the intent recognition agent 1101 can pass the previous inquiry content and the current inquiry information to the inquiry agent 1102; if not, the intent recognition agent 1101 and the inquiry agent 1102 can pass the previous inquiry content to the health consultation agent 1103. The inquiry agent 1102 can use the previous inquiry content and the current inquiry information as input, and output the target user's symptom overview information and initial medication recommendation information, and then pass the target user's current inquiry information, symptom overview information, and initial medication recommendation information as part of the previous inquiry content to the health consultation agent 1103. The health consultation agent 1103 takes the previous inquiry content as input and outputs the question (i.e., the current inquiry information), whether the consultation is ended, and whether medication is suitable (i.e., whether the target user is suitable for medication). If the consultation is ended, the health consultation agent 1103 passes the target user's current inquiry information, symptom overview information, and initial medication advice information as part of the previous inquiry content to the drug recommendation agent 1104. The drug recommendation agent 1104 takes the previous inquiry content as input and outputs a medication recommendation plan, which is then passed as part of the previous inquiry content to the eligibility review agent 1105. The eligibility review agent 1105 determines the drug's marketability information based on the medication recommendation plan in the previous inquiry content. If the drug's marketability information indicates that the target user is eligible to purchase the drug, the eligibility review agent 1105 passes the medication recommendation plan as part of the previous inquiry content to the information integration agent 1106; if the drug's marketability information indicates that the target user is not eligible to purchase the drug, the review result is fed back to the drug recommendation agent 1104. The drug recommendation agent 1104 will again take the previous inquiry content as input and output a medication recommendation plan. The information integration agent 1106 can take the previous inquiry content as input and output health consultation feedback information, while simultaneously determining the end of the current consultation round. If it is determined not to end, and the target user is not suitable for medication recommendation, the health consultation agent 1103 will pass the target user's current inquiry information, symptom overview information, and initial medication recommendation information as part of the previous inquiry content to the human health consultation agent 1107. The human health consultation agent 1107 will take the previous inquiry content as input and output human consultation prompt information, while simultaneously determining the end of the current consultation round. If it is determined not to end, and the target user is suitable for medication recommendation, the health consultation agent 1103 can determine the end of the current consultation round.

[0159] Figure 13This is a flowchart of a multi-agent-based interaction method according to an embodiment of the present invention. Figure 13 As shown, in one optional implementation, the method of this embodiment may further include the following steps:

[0160] Step S1301: In response to the page redirection control being triggered, the drug purchase page corresponding to the health consultation feedback information is displayed.

[0161] In an optional implementation of this embodiment, the health consultation feedback information may further include a page redirection control. Therefore, in this step, if the terminal detects that the page redirection control has been triggered, for example, if the target user clicks the page redirection control, it can redirect to and display the drug purchase page corresponding to the health consultation feedback information.

[0162] The drug purchase page may include at least one recommended drug from the medication suggestion plan and an order submission control. The setup of the drug purchase page can be customized according to actual needs; this embodiment does not impose any restrictions.

[0163] It is easy to understand that step S1301 can be performed after step S306.

[0164] Step S1302: In response to the order submission control being triggered, an order submission request is sent based on the drug selection information.

[0165] In this step, if the order submission control is triggered, the terminal can determine the drug purchase information based on the selected recommended drugs and send an order submission request to the server.

[0166] Figure 14 This is a flowchart of a multi-agent-based interaction method according to an embodiment of the present invention. Figure 14 As shown, in one optional implementation, step S1302 may include the following steps:

[0167] Step S1401: In response to the order submission control being triggered, determine whether the drug selection information includes prescription drugs.

[0168] In one optional implementation of this embodiment, the medication recommendation scheme may also include prescription drugs. Therefore, after the order submission control is triggered, the terminal can determine whether the drug selection information includes prescription drugs.

[0169] In step S1402, in response to the fact that the drug purchase information includes prescription drugs, a prescription issuance request is sent based on the target user's symptom overview information and the drug identification of the prescription drug.

[0170] In this step, if the drug selection information includes prescription drugs, the terminal can send a prescription request based on the target user's symptom summary information and the prescription drug's identification, so that the target user can be qualified to purchase prescription drugs.

[0171] Step S1403: In response to receiving the target prescription, an order submission request is sent based on the drug selection information and the target prescription.

[0172] After the server issues a prescription for a drug through a human or intelligent agent (i.e., the target prescription), it can send the target prescription back to the terminal. Therefore, upon receiving the target prescription, the terminal can send an order submission request based on the drug selection information and the target prescription.

[0173] In this embodiment of the invention, the process of intention recognition, health consultation, and drug recommendation can be automatically performed based on the target user's inquiry through multi-agent interaction, which reduces the time cost of online health consultation for users and helps them quickly and accurately select the right drugs and submit drug purchase orders, thus effectively improving the efficiency of drug selection for users.

[0174] In this embodiment of the invention, after the terminal sends inquiry information to the server, the server-side health consultation agent determines the processing method for the current consultation round based on the inquiry information. If the current processing method is determined to be ending the health consultation, the drug recommendation agent obtains the user's medication recommendation plan based on the inquiry content and information from previous consultation rounds. Then, the information integration agent obtains health consultation feedback information, including corresponding drug recommendation instructions, based on the medication recommendation plan, and sends this feedback information to the terminal. Upon receiving the health consultation feedback information, the terminal displays it. This embodiment of the invention can recommend drugs to users based on multi-agent interaction, thus improving the efficiency of drug selection and enhancing the user's drug purchasing experience when the user cannot determine the required drug.

[0175] Figure 15 This is a flowchart of the multi-agent-based interaction method of this invention on the server side. For example... Figure 15 As shown, the method in this embodiment includes the following steps on the server side:

[0176] Step S1501: In response to receiving the current inquiry information from the target user in the current consultation round, the health consultation agent determines the processing method for the current consultation round based on the current inquiry information.

[0177] In this embodiment, the implementation of step S1501 is the same as that of step S302, and will not be described again here.

[0178] In step S1502, in response to the processing method including determining the health consultation judgment result, and the health consultation judgment result being the end of health consultation, the drug recommendation agent obtains the target user's medication recommendation plan based on the inquiry content of the previous consultation rounds and the current inquiry information.

[0179] In this embodiment, the implementation method of step S1502 is the same as that of step S303, and will not be described again here.

[0180] In step S1503, in response to obtaining the medication recommendation plan, the information integration agent obtains health consultation feedback information based on the medication recommendation plan.

[0181] In this embodiment, the implementation method of step S1503 is the same as that of step S304, and will not be described again here.

[0182] Step S1504: Send health consultation feedback information to the terminal.

[0183] In this embodiment, the implementation method of step S1504 is the same as that of step S305, and will not be described again here.

[0184] In this embodiment of the invention, after the terminal sends inquiry information to the server, the server-side health consultation agent determines the processing method for the current consultation round based on the inquiry information. If the current processing method is determined to be ending the health consultation, the drug recommendation agent obtains the user's medication recommendation plan based on the inquiry content and information from previous consultation rounds. Then, the information integration agent obtains health consultation feedback information, including drug recommendation instructions corresponding to the medication recommendation plan, based on the medication recommendation plan, and sends the health consultation feedback information to the terminal. After receiving the health consultation feedback information, the terminal displays it. This embodiment of the invention can recommend drugs to users based on a multi-agent interaction method, thus improving the efficiency of drug selection and enhancing the user's drug purchase experience when the user cannot determine the required drug.

[0185] Figure 16 This is a flowchart of the multi-agent-based interaction method of this invention on the terminal side. For example... Figure 16 As shown, the method in this embodiment includes the following steps on the terminal side:

[0186] Step S1601: Send the current query information.

[0187] In this embodiment, the implementation method of step S1601 is the same as that of step S301, and will not be described again here.

[0188] Step S1602: In response to receiving the health consultation feedback information for the current inquiry, display the health consultation feedback information.

[0189] In this embodiment, the implementation method of step S1602 is the same as that of step S306, and will not be described again here.

[0190] In this embodiment of the invention, after the terminal sends inquiry information to the server, the server-side health consultation agent determines the processing method for the current consultation round based on the inquiry information. If the current processing method is determined to be ending the health consultation, the drug recommendation agent obtains the user's medication recommendation plan based on the inquiry content and information from previous consultation rounds. Then, the information integration agent obtains health consultation feedback information, including drug recommendation instructions corresponding to the medication recommendation plan, based on the medication recommendation plan, and sends the health consultation feedback information to the terminal. After receiving the health consultation feedback information, the terminal displays it. This embodiment of the invention can recommend drugs to users based on a multi-agent interaction method, thus improving the efficiency of drug selection and enhancing the user's drug purchase experience when the user cannot determine the required drug.

[0191] Figure 17 This is a schematic diagram of a multi-agent-based interactive device according to an embodiment of the present invention, applicable to servers. For example... Figure 17 As shown, the multi-agent interactive device in this embodiment includes a processing method determination unit 1701, a scheme determination unit 1702, a feedback information determination unit 1703, and an information sending unit 1704.

[0192] The processing method determination unit 1701 is used to respond to receiving the current inquiry information of the target user in the current consultation round, and the health consultation agent determines the processing method of the current consultation round based on the current inquiry information. The processing method includes at least one of determining a health consultation judgment result, outputting question information, and determining a medication suitability result. The health consultation judgment result is used to indicate whether the health consultation is ended, and the medication suitability result is used to indicate whether the target user is suitable for medication. The scheme determination unit 1702 is used to respond to the processing method including determining a health consultation judgment result, and the health consultation judgment result is ending the health consultation, and the drug recommendation agent obtains the medication recommendation scheme for the target user based on the inquiry content of the previous consultation round and the current inquiry information. The feedback information determination unit 1703 is used to respond to obtaining the medication recommendation scheme, and the information integration agent obtains health consultation feedback information based on the medication recommendation scheme. The health consultation feedback information includes the drug recommendation description corresponding to the medication recommendation scheme. The information sending unit 1704 is used to send the health consultation feedback information to the terminal.

[0193] In this embodiment of the invention, after the terminal sends inquiry information to the server, the server-side health consultation agent determines the processing method for the current consultation round based on the inquiry information. If the current processing method is determined to be ending the health consultation, the drug recommendation agent obtains the user's medication recommendation plan based on the inquiry content and information from previous consultation rounds. Then, the information integration agent obtains health consultation feedback information, including drug recommendation instructions corresponding to the medication recommendation plan, based on the medication recommendation plan, and sends the health consultation feedback information to the terminal. After receiving the health consultation feedback information, the terminal displays it. This embodiment of the invention can recommend drugs to users based on a multi-agent interaction method, thus improving the efficiency of drug selection and enhancing the user's drug purchase experience when the user cannot determine the required drug.

[0194] Figure 18 This is a schematic diagram of a multi-agent-based interactive device according to an embodiment of the present invention, applicable to a terminal. For example... Figure 18 As shown, the multi-agent interactive device in this embodiment includes an information sending unit 1801 and an information receiving unit 1802.

[0195] The information sending unit 1801 is used to send current inquiry information; the information receiving unit 1802 is used to respond to the health consultation feedback information received from the current inquiry information and display the health consultation feedback information. The health consultation feedback information is obtained by the information integration agent based on the target user's medication recommendation plan. The medication recommendation plan obtains the health consultation feedback information by the drug recommendation agent based on the inquiry content of the previous consultation round and the current inquiry information. The processing method of the medication recommendation plan in response to the current consultation round includes determining the health consultation judgment result, and the health consultation judgment result is a determination to end the health consultation. The processing method is determined by the health consultation agent based on the current inquiry information. The processing method includes at least one of determining the health consultation judgment result, outputting question information, and determining the medication suitability result. The health consultation judgment result is used to indicate whether the health consultation is ended, and the medication suitability result is used to indicate whether the target user is suitable for medication. The health consultation feedback information includes the drug recommendation description corresponding to the medication recommendation plan.

[0196] In this embodiment of the invention, after the terminal sends inquiry information to the server, the server-side health consultation agent determines the processing method for the current consultation round based on the inquiry information. If the current processing method is determined to be ending the health consultation, the drug recommendation agent obtains the user's medication recommendation plan based on the inquiry content and information from previous consultation rounds. Then, the information integration agent obtains health consultation feedback information, including drug recommendation instructions corresponding to the medication recommendation plan, based on the medication recommendation plan, and sends the health consultation feedback information to the terminal. After receiving the health consultation feedback information, the terminal displays it. This embodiment of the invention can recommend drugs to users based on a multi-agent interaction method, thus improving the efficiency of drug selection and enhancing the user's drug purchase experience when the user cannot determine the required drug.

[0197] Figure 19 This is a schematic diagram of an electronic device according to an embodiment of the present invention. In this embodiment, the electronic device includes a server, a terminal, etc. Figure 19 As shown, the electronic device includes at least one processor 1901; a memory 1902 communicatively connected to at least one processor 1901; and a communication component 1903 communicatively connected to a scanning device, wherein the communication component 1903 receives and transmits data under the control of the processor 1901; wherein the memory 1902 stores instructions executable by at least one processor 1901, which are executed by at least one processor 1901 to implement the above-described multi-agent-based interaction method.

[0198] Specifically, the electronic device includes: one or more processors 1901 and a memory 1902. Figure 19 Taking a processor 1901 as an example, the processor 1901 and memory 1902 can be connected via a bus or other means. Figure 19 Taking a bus connection as an example, memory 1902, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer-executable programs, and modules. Processor 1901 executes various functional applications and data processing of the device by running the non-volatile software programs, instructions, and modules stored in memory 1902, thereby realizing the aforementioned multi-agent-based interaction method.

[0199] Memory 1902 may include a program storage area and a data storage area, wherein the program storage area may store the operating system and applications required for at least one function; the data storage area may store an option list, etc. Furthermore, memory 1902 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 embodiments, memory 1902 may optionally include memory remotely located relative to processor 1901, and these remote memories can be connected to external devices 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.

[0200] One or more modules are stored in memory 1902 and, when executed by one or more processors 1901, execute the multi-agent-based interaction method in any of the above method embodiments.

[0201] The above-mentioned products can perform the methods provided in the embodiments of this application, and have the corresponding functional modules and beneficial effects of performing the methods. For technical details not described in detail in this embodiment, please refer to the methods provided in the embodiments of this application.

[0202] In this embodiment of the invention, after the terminal sends inquiry information to the server, the server-side health consultation agent determines the processing method for the current consultation round based on the inquiry information. If the current processing method is determined to be ending the health consultation, the drug recommendation agent obtains the user's medication recommendation plan based on the inquiry content and information from previous consultation rounds. Then, the information integration agent obtains health consultation feedback information, including drug recommendation instructions corresponding to the medication recommendation plan, based on the medication recommendation plan, and sends the health consultation feedback information to the terminal. After receiving the health consultation feedback information, the terminal displays it. This embodiment of the invention can recommend drugs to users based on a multi-agent interaction method, thus improving the efficiency of drug selection and enhancing the user's drug purchase experience when the user cannot determine the required drug.

[0203] Another embodiment of the present invention relates to a non-volatile storage medium for storing a computer-readable program for use by a computer to execute some or all of the above-described method embodiments.

[0204] That is, those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. This program is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods described in the embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0205] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A multi-agent-based interaction method, applicable to servers, characterized in that, The method includes: In response to receiving the current inquiry information from the target user in the current consultation round, the health consultation agent determines the processing method for the current consultation round based on the current inquiry information. The processing method includes at least one of determining the health consultation judgment result, outputting the question information, and determining the medication suitability result. The health consultation judgment result is used to indicate whether the health consultation is to be ended, and the medication suitability result is used to indicate whether the target user is suitable for medication. In response to the processing method including determining the health consultation judgment result, and the health consultation judgment result being the end of health consultation, the drug recommendation agent obtains the medication recommendation plan for the target user based on the inquiry content of the previous consultation rounds and the current inquiry information; In response to obtaining the medication recommendation plan, the information integration agent obtains health consultation feedback information based on the medication recommendation plan, and the health consultation feedback information includes drug recommendation instructions corresponding to the medication recommendation plan; Send the health consultation feedback information to the terminal; Wherein, the process of the health consultation agent determining the current consultation round based on the current inquiry information received from the target user in the current consultation round includes: In response to receiving the current query information, the intent recognition agent determines the corresponding intent recognition result based on the current query information. The intent recognition result is used to characterize whether the target user has initiated a query for the first time. In response to the intent recognition result indicating that the target user has initiated an inquiry for the first time, the inquiry agent determines the target user's symptom overview information and initial medication recommendation information based on the current inquiry information; In response to the fact that the target user has initiated an inquiry before, the health consultation agent determines the processing method based on the current inquiry information and the inquiry content, which includes the symptom overview information and the initial medication recommendation information.

2. The method according to claim 1, characterized in that, In response to obtaining the medication recommendation plan, the information integration agent obtains health consultation feedback information based on the medication recommendation plan, including: In response to obtaining the medication recommendation plan, the qualification verification agent determines the drug sales information corresponding to the medication recommendation plan based on the medication recommendation plan. The drug sales information is used to indicate whether the target user is qualified to purchase the recommended drug. The recommended drug is the drug in the medication recommendation plan. In response to the drug sales approval information indicating that the target user is qualified to purchase, the information integration agent obtains the health consultation feedback information.

3. The method according to claim 2, characterized in that, The method further includes: In response to the drug sales information indicating that the target user is not eligible to purchase the drug, the drug recommendation agent obtains the medication suggestion plan again.

4. The method according to claim 1, characterized in that, The method further includes: In response to the processing method, which includes determining the health consultation judgment result and determining the medication suitability result, wherein the health consultation judgment result indicates that the health consultation is not terminated and the medication suitability result indicates that the target user is not suitable for medication, the artificial health consultation agent obtains artificial consultation prompt information based on the current inquiry information; Send the human consultation prompt message to the terminal.

5. The method according to claim 1, characterized in that, The method further includes: In response to the processing method of determining the health consultation judgment result and determining the medication suitability result, wherein the health consultation judgment result indicates that the health consultation is not terminated and the medication suitability result indicates that the target user is suitable for medication, the health consultation agent determines that the current consultation round is terminated.

6. The method according to claim 1, characterized in that, The method further includes: In response to the processing method of outputting question information, the health consultation agent determines the current question information for the current consultation round; The current question information is sent to the terminal.

7. The method according to claim 6, characterized in that, Sending the current query information to the terminal includes: Based on the information type of the current question information, a predefined toolkit is invoked to create and update the current question information, wherein the information type includes at least one of streaming, non-streaming, and exception.

8. A multi-agent-based interaction method, applicable to terminals, characterized in that, The method includes: Send the current query information; In response to receiving the current inquiry information, the health consultation feedback information is displayed. This feedback information is obtained by an information integration agent based on the target user's medication recommendation plan. The medication recommendation plan obtains the health consultation feedback information by a drug recommendation agent based on the inquiry content of previous consultation rounds and the current inquiry information. The processing method for the current consultation round includes determining a health consultation judgment result, whereby the health consultation judgment result is a determination to end the health consultation. This processing method is determined by the health consultation agent based on the current inquiry information and includes at least one of determining a health consultation judgment result, outputting question information, and determining a medication suitability result. The health consultation judgment result indicates whether the health consultation has ended, and the medication suitability result indicates whether the target user is suitable for medication. The health consultation feedback information includes a drug recommendation description corresponding to the medication recommendation plan. The processing method is determined in the following way: In response to receiving the current query information, the intent recognition agent determines the corresponding intent recognition result based on the current query information. The intent recognition result is used to characterize whether the target user has initiated a query for the first time. In response to the intent recognition result indicating that the target user has initiated an inquiry for the first time, the inquiry agent determines the target user's symptom overview information and initial medication recommendation information based on the current inquiry information; In response to the fact that the target user has initiated an inquiry before, the health consultation agent determines the processing method based on the current inquiry information and the inquiry content, which includes the symptom overview information and the initial medication recommendation information.

9. The method according to claim 8, characterized in that, The method further includes: In response to receiving the current query information, the current question information is displayed, and the current question information is determined by the health consultation agent in response to the processing method being to output question information.

10. The method according to claim 8, characterized in that, The method further includes: In response to receiving a human consultation prompt, the human consultation prompt is displayed. The human consultation prompt is provided by the human health consultation agent in response to the processing method, which includes determining the health consultation judgment result and determining the medication suitability result. The health consultation judgment result indicates that the health consultation will not end, and the medication suitability result indicates that the target user is not suitable for medication.

11. The method according to claim 8, characterized in that, The health consultation feedback information includes a page redirection control; The method further includes: In response to the page redirection control being triggered, the drug purchase page corresponding to the health consultation feedback information is displayed, and the drug purchase page includes an order submission control; In response to the order submission control being triggered, an order submission request is sent based on the drug purchase information.

12. The method according to claim 11, characterized in that, The response to the order submission control being triggered, sending an order submission request based on the drug selection information includes: In response to the order submission control being triggered, determine whether the drug selection information includes prescription drugs; In response to the fact that the drug purchase information includes prescription drugs, a prescription request is sent based on the target user's symptom overview information and the drug identification of the prescription drug; In response to receiving a target prescription, an order submission request is sent based on the drug selection information and the target prescription, wherein the target prescription is the prescription corresponding to the prescription drug.

13. A multi-agent-based interaction method, characterized in that, The method includes: Send the current query information; In response to receiving the current inquiry information from the target user in the current consultation round, the health consultation agent determines the processing method for the current consultation round based on the current inquiry information. The processing method includes at least one of determining the health consultation judgment result, outputting the question information, and determining the medication suitability result. The health consultation judgment result is used to indicate whether the health consultation is to be ended, and the medication suitability result is used to indicate whether the target user is suitable for medication. In response to the processing method including determining the health consultation judgment result, and the health consultation judgment result being the end of health consultation, the drug recommendation agent obtains the medication recommendation plan for the target user based on the inquiry content of the previous consultation rounds and the current inquiry information; In response to obtaining the medication recommendation plan, the information integration agent obtains health consultation feedback information based on the medication recommendation plan, and the health consultation feedback information includes drug recommendation instructions corresponding to the medication recommendation plan; Send the health consultation feedback information to the terminal; In response to receiving the current inquiry information, the health consultation feedback information is displayed; Wherein, the process of the health consultation agent determining the current consultation round based on the current inquiry information received from the target user in the current consultation round includes: In response to receiving the current query information, the intent recognition agent determines the corresponding intent recognition result based on the current query information. The intent recognition result is used to characterize whether the target user has initiated a query for the first time. In response to the intent recognition result indicating that the target user has initiated an inquiry for the first time, the inquiry agent determines the target user's symptom overview information and initial medication recommendation information based on the current inquiry information; In response to the fact that the target user has initiated an inquiry before, the health consultation agent determines the processing method based on the current inquiry information and the inquiry content, which includes the symptom overview information and the initial medication recommendation information.

14. A multi-agent-based interactive system, characterized in that, include: The server is configured to, in response to receiving current inquiry information from a target user in the current consultation round, determine the processing method for the current consultation round based on the current inquiry information, wherein the processing method includes determining a health consultation judgment result, and the health consultation judgment result is to end the health consultation; a drug recommendation agent obtains a medication recommendation plan for the target user based on the inquiry content of previous consultation rounds and the current inquiry information; in response to obtaining the medication recommendation plan, an information integration agent obtains health consultation feedback information based on the medication recommendation plan and sends the health consultation feedback information to the terminal; the processing method includes at least one of determining a health consultation judgment result, outputting question information, and determining a medication suitability result; the health consultation judgment result is used to characterize whether the health consultation is ended; the medication suitability result is used to characterize whether the target user is suitable for medication; the health consultation feedback information includes a drug recommendation description corresponding to the medication recommendation plan. as well as The terminal is configured to send current inquiry information and, in response to receiving the current inquiry information, display health consultation feedback information. Wherein, the process of the health consultation agent determining the current consultation round based on the current inquiry information received from the target user in the current consultation round includes: In response to receiving the current query information, the intent recognition agent determines the corresponding intent recognition result based on the current query information. The intent recognition result is used to characterize whether the target user has initiated a query for the first time. In response to the intent recognition result indicating that the target user has initiated an inquiry for the first time, the inquiry agent determines the target user's symptom overview information and initial medication recommendation information based on the current inquiry information; In response to the fact that the target user has initiated an inquiry before, the health consultation agent determines the processing method based on the current inquiry information and the inquiry content, which includes the symptom overview information and the initial medication recommendation information.

15. A multi-agent-based interactive device, suitable for servers, characterized in that, The device includes: The processing method determination unit is used to respond to receiving the current inquiry information of the target user in the current consultation round. The health consultation agent determines the processing method of the current consultation round based on the current inquiry information. The processing method includes at least one of determining the health consultation judgment result, outputting the question information, and determining the medication suitability result. The health consultation judgment result is used to characterize whether the health consultation ends, and the medication suitability result is used to characterize whether the target user is suitable for medication. The scheme determination unit is used to respond to the processing method including determining the health consultation judgment result, wherein the health consultation judgment result is to end the health consultation, and the drug recommendation agent obtains the medication recommendation scheme for the target user based on the inquiry content of the previous consultation rounds and the current inquiry information; The feedback information determination unit is used to respond to the acquisition of the medication recommendation plan. The information integration agent obtains health consultation feedback information based on the medication recommendation plan. The health consultation feedback information includes drug recommendation instructions corresponding to the medication recommendation plan. The information sending unit is used to send the health consultation feedback information to the terminal; The processing method determination unit is further used for: In response to receiving the current query information, the intent recognition agent determines the corresponding intent recognition result based on the current query information. The intent recognition result is used to characterize whether the target user has initiated a query for the first time. In response to the intent recognition result indicating that the target user has initiated an inquiry for the first time, the inquiry agent determines the target user's symptom overview information and initial medication recommendation information based on the current inquiry information; In response to the fact that the target user has initiated an inquiry before, the health consultation agent determines the processing method based on the current inquiry information and the inquiry content, which includes the symptom overview information and the initial medication recommendation information.

16. A multi-agent-based interactive device, suitable for a terminal, characterized in that, The device includes: The information sending unit is used to send the current query information; An information receiving unit is configured to display health consultation feedback information in response to receiving the current inquiry information. The health consultation feedback information is obtained by an information integration agent based on the target user's medication recommendation plan. The medication recommendation plan obtains the health consultation feedback information by a drug recommendation agent based on the inquiry content of previous consultation rounds and the current inquiry information. The processing method of the medication recommendation plan in response to the current consultation round includes determining a health consultation judgment result, whereby the health consultation judgment result is a determination to end the health consultation. The processing method is determined by the health consultation agent based on the current inquiry information and includes at least one of determining a health consultation judgment result, outputting question information, and determining a medication suitability result. The health consultation judgment result is used to characterize whether the health consultation has ended, and the medication suitability result is used to characterize whether the target user is suitable for medication. The health consultation feedback information includes a drug recommendation description corresponding to the medication recommendation plan. The processing method is determined in the following way: In response to receiving the current query information, the intent recognition agent determines the corresponding intent recognition result based on the current query information. The intent recognition result is used to characterize whether the target user has initiated a query for the first time. In response to the intent recognition result indicating that the target user has initiated an inquiry for the first time, the inquiry agent determines the target user's symptom overview information and initial medication recommendation information based on the current inquiry information; In response to the fact that the target user has initiated an inquiry before, the health consultation agent determines the processing method based on the current inquiry information and the inquiry content, which includes the symptom overview information and the initial medication recommendation information.

17. An electronic device comprising a memory and a processor, characterized in that, The memory is used to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the processor to implement the method as described in any one of claims 1-13.

18. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method as described in any one of claims 1-13.

19. A computer program product, characterized in that, The computer program product includes a computer program / instruction that, when executed by a processor, implements the method as described in any one of claims 1-13.

Images