Drug recommendation method based on artificial intelligence and related device

Abstract

The embodiment of the invention provides a drug recommendation method based on artificial intelligence and a related device. The method comprises the following steps: in a plurality of symptom categories, determining a target symptom category corresponding to a user; wherein the plurality of symptom categories respectively correspond to question-asking guidelines generated by a high-order large model for the symptom categories; the inquiry guide is used for indicating information which needs to be acquired from the user before the medicine is recommended to the user; calling an intelligent shopping guide model according to the generated dialogue content and a questioning guide corresponding to the target symptom category in a dialogue process with the user, and generating a questioning message for questioning the user; wherein the parameter scale of the intelligent shopping guide model is smaller than that of the high-order large model. According to the drug recommendation method based on artificial intelligence and the related device, occupation of computing power resources can be reduced on the basis of ensuring drug recommendation precision.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of computer technology, and in particular to an artificial intelligence-based drug recommendation method and related devices. Background Art

[0002] With the rapid development of internet healthcare, pharmaceutical sales have become a core business of e-commerce platforms. To improve the efficiency and experience of online drug purchases, e-commerce platforms offer intelligent shopping guides that engage in conversations with users and make drug recommendations based on these conversations. To enhance the accuracy of these recommendations, e-commerce platforms train these guides to identify user needs and recommend drugs based on information such as drug-related information, user self-reports of their condition, and historical recommendation records. However, invoking these guides to perform drug recommendations consumes significant computing resources. Summary of the Invention

[0003] In view of this, multiple embodiments of the present application are dedicated to providing an artificial intelligence-based drug recommendation method and related devices, which can reduce the occupation of computing resources while ensuring the accuracy of drug recommendations.

[0004] One embodiment of the present application provides an artificial intelligence-based drug recommendation method, comprising: determining a target symptom category corresponding to a user among multiple symptom categories; wherein the multiple symptom categories respectively correspond to follow-up guides generated by a high-order large model for the symptom category; the follow-up guide is used to indicate information that needs to be obtained from the user before recommending drugs to the user; in the process of conducting a conversation with the user, calling an intelligent shopping guide model according to the generated conversation content and the follow-up guide corresponding to the target symptom category to generate a question message to the user; wherein the parameter scale of the intelligent shopping guide model is smaller than the parameter scale of the high-order large model.

[0005] One embodiment of the present application provides an artificial intelligence-based drug recommendation method, including: obtaining drug application scope information and actual dialogue information corresponding to symptom categories; wherein the actual dialogue information is dialogue information generated by communication between a user and customer service; based on the drug application scope information and the actual dialogue information corresponding to the symptom category, calling a high-order large model to generate a follow-up guide corresponding to each of the symptom categories; wherein the follow-up guide is used to indicate information that needs to be obtained from the user before recommending drugs to the user; wherein the follow-up guide is used to provide to an intelligent shopping guide model that communicates with the user, and the parameter scale of the intelligent shopping guide model is smaller than the parameter scale of the high-order large model.

[0006] One embodiment of the present application provides an artificial intelligence-based drug recommendation device, comprising: a category determination module, used to determine a target symptom category corresponding to a user among multiple symptom categories; wherein the multiple symptom categories respectively correspond to follow-up guides generated by a high-order large model for the symptom category; the follow-up guide is used to indicate information that needs to be obtained from the user before recommending drugs to the user; a message generation module, used to call an intelligent shopping guide model according to the generated conversation content and the follow-up guide corresponding to the target symptom category during a conversation with the user, and generate a question message to ask the user; wherein the parameter scale of the intelligent shopping guide model is smaller than the parameter scale of the high-order large model.

[0007] One embodiment of the present application provides an artificial intelligence-based drug recommendation device, including: an information acquisition module, used to obtain the applicable scope information of the drug and the actual dialogue information corresponding to the symptom category; wherein the actual dialogue information is the dialogue information generated by the communication between the user and the customer service; a guide generation module, used to call a high-order large model to generate a follow-up guide corresponding to each of the symptom categories based on the applicable scope information of the drug and the actual dialogue information corresponding to the symptom category; wherein the follow-up guide is used to indicate the information that needs to be obtained from the user before recommending the drug to the user; wherein the follow-up guide is used to provide to an intelligent shopping guide model that communicates with the user, and the parameter scale of the intelligent shopping guide model is smaller than the parameter scale of the high-order large model.

[0008] One embodiment of the present application provides a computer device, which includes a memory and a processor, wherein the memory stores at least one computer program, and the at least one computer program is loaded and executed by the processor to implement the aforementioned artificial intelligence-based drug recommendation method.

[0009] One embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores at least one computer program, and when the at least one computer program is executed by a processor, it can implement the aforementioned artificial intelligence-based drug recommendation method.

[0010] One embodiment of the present application provides a computer program product, which is used to implement the aforementioned artificial intelligence-based drug recommendation method.

[0011] In multiple implementations provided in the present application, there is a follow-up guide generated by a high-order large model for symptom categories, and the follow-up guide is used to indicate the information that needs to be obtained from the user before recommending drugs to the user. In the process of having a conversation with the user, based on the conversation content and the follow-up guide for the target symptom category corresponding to the user, an intelligent shopping guide model with a smaller parameter scale than the high-order large model can be called to generate a question message to trigger the continuation of the conversation until sufficient information is obtained to recommend drugs. Among them, the follow-up guide is generated by the high-order large model based on language comprehension and knowledge integration capabilities, and is then used as the basis for drug recommendations by the intelligent shopping guide model. The intelligent shopping guide model does not need to have the ability to generate a follow-up guide, so the intelligent shopping guide model can have a smaller parameter scale. Calling the intelligent shopping guide model can reduce the occupation of computing resources while ensuring the accuracy of drug recommendations. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 A system diagram of the application of an artificial intelligence-based drug recommendation method provided for one embodiment of the present application.

[0013] Figure 2 A flowchart of an artificial intelligence-based drug recommendation method provided for one embodiment of the present application.

[0014] Figure 3 A flowchart of an artificial intelligence-based drug recommendation method provided for another embodiment of the present application.

[0015] Figure 4 A module diagram of an artificial intelligence-based drug recommendation device provided for one embodiment of the present application.

[0016] Figure 5 A module diagram of an artificial intelligence-based drug recommendation device provided for another embodiment of the present application.

[0017] Figure 6 A schematic diagram of a computer device provided for one embodiment of the present application. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments.

[0019] In the description of the embodiments of the present application, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the embodiments of the present application, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0020] The traditional online drug purchase process requires users to compose and input information such as symptom keywords, drug type, brand, and name into the e-commerce platform. The platform then sorts and displays results based on product sales, positive reviews, or the merchant's advertising strategy. This approach requires users to have some medical knowledge or understand drug nomenclature and efficacy; otherwise, it's difficult for the e-commerce platform to return the expected results in a single search.

[0021] To help users without relevant knowledge shop online, e-commerce platforms offer intelligent shopping guides, which use conversational methods to obtain the information needed for drug recommendations. These robots, based on question templates and pre-set rules, ask users specific questions (e.g., "What are your primary symptoms?", "How long have the symptoms lasted?"), guiding them to provide more information that aids in making accurate drug recommendations. This information, gleaned from the conversation, allows for adaptive drug recommendations.

[0022] In related technologies, deep learning models with larger parameter scales (e.g., large pre-trained models with tens of billions / hundreds of billions of parameters) are usually collected as intelligent shopping guides to undertake tasks such as natural language understanding, symptom recognition, and recommendation decisions. Since models with larger parameter scales have better semantic representation and language understanding capabilities, and can absorb knowledge from multiple fields and across disciplines, they can combine medicine, pharmacy, and everyday language to generate richer question-and-answer strategies. When applied to drug recommendations, they have the expected user demand mining capabilities and the ability to accurately recommend drugs. However, when models with larger parameter scales are called as intelligent shopping guides, there are problems such as the need to occupy more computing resources, high complexity in deployment and maintenance, and slow response speeds, making it difficult to meet the real-time and high concurrency requirements required for drug recommendation scenarios.

[0023] Therefore, it is necessary to provide a drug recommendation method based on artificial intelligence, which has a follow-up guide generated by a high-order large model for symptom categories, and the follow-up guide is used to indicate the information that needs to be obtained from the user before recommending drugs to the user. In the process of dialogue with the user, based on the dialogue content and the follow-up guide for the target symptom category corresponding to the user, an intelligent shopping guide model with a smaller parameter scale than the high-order large model can be called to generate a question message to trigger the continuation of the dialogue until sufficient information is obtained to recommend drugs. Among them, the follow-up guide is generated by the high-order large model based on language understanding and knowledge integration capabilities, and is then used as the basis for drug recommendations by the intelligent shopping guide model. The intelligent shopping guide model does not need to have the ability to generate a follow-up guide, so the intelligent shopping guide model can have a smaller parameter scale. Calling the intelligent shopping guide model can reduce the occupation of computing resources while ensuring the accuracy of drug recommendations.

[0024] See also Figure 1 . In multiple embodiments provided in the present application, the drug recommendation method based on artificial intelligence can be applied to a drug recommendation device based on artificial intelligence. The drug recommendation device based on artificial intelligence can be an electronic device with certain computing power and network access capabilities. The electronic device can be a desktop computer, a laptop computer, a tablet computer, a smart phone, or a server. The server can also be a distributed server, including multiple processors, memories, network communication modules, etc. that work together to achieve various functions. Alternatively, the server can also be a server cluster formed by several servers, which has higher computing and data processing capabilities. With the development of science and technology, the server can also be implemented using new forms of technical means, such as a new type of "server" based on quantum computing. Of course, in some embodiments, the operating status determination device can also be a program module running in an electronic device.

[0025] In this embodiment, the drug recommendation device based on artificial intelligence can communicate with the client in a dialogue. During the dialogue, the target symptom category corresponding to the user is determined from multiple symptom categories. Since there is a follow-up question guide generated by the high-order large model for the symptom category, and the follow-up question guide is used to indicate the information that needs to be obtained from the user before recommending drugs to the user. Therefore, based on the generated dialogue content and the follow-up question guide corresponding to the target symptom category, the intelligent shopping guide model with a parameter scale smaller than the high-order large model can be called to generate a question message to the user. In this way, multiple rounds of dialogue with the client can be achieved until sufficient information is obtained to serve as the basis for drug recommendations, and drug recommendations are made to the client accordingly.

[0026] In this embodiment, the AI-based drug recommendation device can also be used to call the high-level large model to generate follow-up questioning guidelines for each symptom category based on the drug's applicable scope information and actual conversation information corresponding to the symptom category, and then provide the follow-up questioning guidelines to the intelligent shopping guide model. Optionally, the intelligent shopping guide model and the high-level large model run on different electronic devices, with the electronic device running the intelligent shopping guide model used to communicate with the client, and the electronic device running the high-level large model used to communicate with the electronic device running the intelligent shopping guide model.

[0027] One embodiment of the present application provides an example application scenario of an artificial intelligence-based drug recommendation method. The artificial intelligence-based drug recommendation method is applied to an artificial intelligence-based drug recommendation device to reduce the use of computing resources while ensuring the accuracy of drug recommendations.

[0028] For example, the following medication is available on e-commerce platforms: XX brand hydrotalcite chewable tablets. The indications for XX brand hydrotalcite chewable tablets include: stomach pain, heartburn, acid reflux, and other symptoms caused by excessive gastric acid; indigestion and bloating; adjunctive treatment for acute gastritis and acute attacks of chronic gastritis; and prevention of gastric mucosal damage caused by nonsteroidal anti-inflammatory drugs (NSAIDs) or other medications.

[0029] During the conversation with the user, the user asked, "I have been having stomach pain recently. What medicine can I take to relieve it?"

[0030] Furthermore, based on the existing conversation content and the follow-up question guide corresponding to stomach pain, the intelligent shopping guide model is called to generate a question to the user: "What is your gender?"

[0031] The follow-up guide is generated by a high-order large model, which has a larger parameter scale than the intelligent shopping guide model. The follow-up guide for stomachache is expressed as:

[0032]

[0033] Furthermore, the smart shopping guide model can also continue to ask "How long have the symptoms lasted?" based on the follow-up question guide and in combination with the user's response to "What is your gender?"

[0034] After obtaining information such as gender, age, symptoms, duration, etc. that is sufficient for accurate drug recommendations through repeated questioning, based on the existing conversation content such as "So, what can I eat to relieve the symptoms? It must be effective quickly" and the user's high response speed in the conversation, it was analyzed that the user has a relatively positive willingness to communicate. In addition, it was detected that the conversation content contains the questions "Give me some recommendations quickly" and "Does this disease have a high cure rate?"

[0035] On this basis, according to the user's demand of "relieving symptoms", the doubts about drug safety "worrying that it cannot be used in combination with lactobacillus tablets", and the user's preference of "wanting to take medicine with less strong effects", XX brand aluminum carbonate magnesium chewable tablets suitable for users are selected from a large number of medicines for recommendation.

[0036] During this process, there is no need to call a high-order large model with a large parameter scale. Instead, a smart shopping guide model with a smaller parameter scale is obtained through distillation training of the high-order large model. Calling the smart shopping guide model to perform drug recommendations can save computing power.

[0037] See also Figure 2 The embodiment of the present application provides a drug recommendation method based on artificial intelligence. The drug recommendation method based on artificial intelligence can be applied to a drug recommendation device based on artificial intelligence. The drug recommendation method based on artificial intelligence can include the following steps.

[0038] Step S110: Determine the target symptom category corresponding to the user among multiple symptom categories; wherein the multiple symptom categories respectively correspond to follow-up guidelines generated by the high-order large model for the symptom categories; the follow-up guidelines are used to indicate the information that needs to be obtained from the user before recommending drugs to the user.

[0039] Step S120: During a conversation with the user, the intelligent shopping guide model is called according to the generated conversation content and the follow-up guide corresponding to the target symptom category to generate a question message to the user; wherein the parameter scale of the intelligent shopping guide model is smaller than the parameter scale of the high-order large model.

[0040] In this embodiment, symptom categories refer to a series of abnormal changes in the body's functions, metabolism, and morphology during the course of an illness, resulting in subjective abnormal sensations or objective pathological changes. Symptom categories include, but are not limited to, pain, dizziness, fever, jaundice, dyspnea, mucosal bleeding, abdominal mass, obesity, weight loss, polyuria, and oliguria.

[0041] In this embodiment, the target symptom category corresponding to the user can be determined based on the content of the conversation that has already occurred, or based on the user's search history / browsing history within a unit of time before the conversation. The target symptom category corresponding to the user (e.g., hair loss) is selected from multiple symptom categories.

[0042] In this embodiment, there is a follow-up question guide generated by a high-order large model for each symptom category. Among them, the high-order large model is a deep learning model that is pre-trained on large-scale medical and verbal materials and has a large parameter scale (e.g., hundreds of billions of parameters). It has the expected language understanding, knowledge reasoning and text generation capabilities, and can generate high-quality follow-up question guides based on symptom categories. Among them, the follow-up question guide is used to guide the intelligent shopping guide model to ask questions based on the generated conversation content, so as to obtain more and more comprehensive disease-related information that is convenient for accurate drug recommendations. Disease-related information includes but is not limited to: gender, age, allergy history, medical history, self-care level, symptom duration, severity, previous medication, and risk of combined medication.

[0043] In this embodiment, the follow-up guide includes a sequence of follow-up questions in various scenarios. Different scenarios correspond to conversations with different amounts of information, avoiding repeated follow-up questions about information already provided by the user in the conversation (such as gender). This helps improve the efficiency of drug recommendations and optimizes the user's experience of conversations with the shopping guide, making the question messages output by the intelligent shopping guide model closer to real-life responses. For example, the follow-up guide includes the following sequence of "gender-symptoms-duration" in scenarios where no information is known, the following sequence of "symptoms-duration" in scenarios where gender is known, and the following sequence of "gender-duration" in scenarios where symptoms are known.

[0044] In addition, the order of follow-up questions corresponding to different symptom categories in different / same scenarios may be the same, different, or partially the same, and the embodiments of the present application do not limit this. For example, for the symptom category of thirst and polyuria, when it is known that the patient has gestational diabetes, the order of follow-up questions specified in the follow-up guide may be "duration of symptoms-other symptoms". Since the gender of patients with gestational diabetes is definitely female, the follow-up guide does not ask about gender. For example, for the symptom category of baldness, when it is known that the duration of the symptom is one year and there are no other symptoms, the order of follow-up questions specified in the follow-up guide may be "gender-family history". Since the probability of male inheritance of the symptom of baldness is higher than that of female, it is necessary to ask about gender and family history. It can be seen that different symptoms may correspond to different follow-up guidelines, so as to conduct targeted and effective follow-up questions, which is conducive to accurate drug recommendations.

[0045] In this implementation, the intelligent shopping guide model is an AI model built using deep learning or lightweight machine learning techniques, with a smaller parameter size than high-level models. It is used to execute follow-up questions during conversations and provide corresponding drug recommendations. The intelligent shopping guide model combines the conversation context and the requirements of the follow-up questions to output specific questions and the final drug recommendations.

[0046] In this embodiment, during a conversation with a user, the intelligent shopping guide model can be called based on the generated conversation content (e.g., "I have been having headaches recently, especially on the left temple") and the follow-up guide corresponding to the target symptom category (e.g., headache). The intelligent shopping guide model can determine the key information required for the known drug recommendation based on the conversation content, and generate a question message based on the follow-up order corresponding to the key information in the follow-up guide. Among them, the question message (e.g., "How long did your headache last?", "Apart from the temple, do you feel any discomfort in other parts of your body?", "Have you ever taken painkillers? If so, how effective is it?", "Do you have high blood pressure or other chronic diseases?") is to obtain information from the user that is not included in the existing conversation content. The purpose of generating the question message is to accurately collect the supplementary information required for the recommendation.

[0047] In some embodiments, the question message may include multiple questions, allowing the user to respond to multiple questions in one round of conversation (e.g., "lasted for three days, only in the temple area, took acetaminophen, only temporarily relieved, no hypertension"), thereby improving information collection efficiency and thereby improving drug recommendation efficiency.

[0048] In some embodiments, a user may mistakenly input information during a conversation. Therefore, the user's response to the question message and the context can be combined to evaluate the user's corresponding symptom category (e.g., bloating). If the evaluation result is inconsistent with the target symptom category (e.g., stomach pain), the target symptom category is updated. In subsequent conversations, the smart shopping guide model is called according to the generated conversation content and the follow-up guide corresponding to the updated target symptom category (e.g., bloating) to generate a question message to the user to ensure that the information required for drug recommendation can be accurately obtained.

[0049] In some embodiments, when a potential medical risk (e.g., a history of drug allergies) is detected in a conversation, a prompt message is generated to remind the user to seek medical treatment offline (e.g., "It is recommended to seek medical treatment first or further examination").

[0050] In this embodiment, there is a follow-up guide generated by a high-order large model for symptom categories, and the follow-up guide is used to indicate the information that needs to be obtained from the user before recommending drugs to the user. In the process of talking with the user, based on the content of the conversation and the follow-up guide of the target symptom category corresponding to the user, an intelligent shopping guide model with a smaller parameter scale than the high-order large model can be called to generate a question message to trigger the continuation of the conversation until sufficient information is obtained to recommend drugs. Among them, the follow-up guide is generated by the high-order large model based on language comprehension and knowledge integration capabilities, and then used as the basis for drug recommendations of the intelligent shopping guide model. The intelligent shopping guide model does not need to have the ability to generate a follow-up guide, so the intelligent shopping guide model can have a smaller parameter scale. Calling the intelligent shopping guide model can reduce the occupation of computing resources while ensuring the accuracy of drug recommendations.

[0051] In some embodiments, the artificial intelligence-based drug recommendation device can analyze the communication intention based on the generated conversation content during the conversation with the user; when the communication intention meets the conditions for continued communication, the intelligent shopping guide model is called according to the generated conversation content and the follow-up guide corresponding to the target symptom category to generate a question message to the user.

[0052] In this embodiment, communication intention refers to the user's subjective inclination or motivation to continue interacting with the intelligent shopping guide system, answering questions, and obtaining recommendations during the current conversation. In some embodiments, communication intention can be categorized into multiple types, including: a user's active desire for in-depth consultation, a user's implicit hesitation to continue communication, and a user's negative willingness, indicating dislike for the conversation content. In some embodiments, communication intention is expressed numerically as the user's willingness to continue communication.

[0053] In some embodiments, analyzing communication intention based on the content of the generated conversation can be performed by classifying the intent of user responses in the conversation. Specifically, if the user response expresses a question, a detailed description of a symptom, or a desire for further understanding, the communication intention is identified as positive; if the user response contains perfunctory information (e.g., "Never mind"), a rejection (e.g., "No"), or a complaint (e.g., "That's not smart"), the communication intention is identified as negative.

[0054] Alternatively, the method of analyzing the willingness to communicate based on the content of the generated conversation can be: based on the analysis of the emotional tendency (e.g., positive, neutral, negative) and tone intensity (e.g., exaggerated sentences, interrogative sentences, imperative sentences) of the user's response, a higher positive emotion or questioning tone is identified as a signal for determining the willingness to communicate.

[0055] Alternatively, the method of analyzing the communication willingness based on the content of the generated conversation can be: determining the communication willingness based on the response time and / or response frequency of the user's reply, the positivity of the communication willingness is negatively correlated with the response time and positively correlated with the response frequency, that is, the shorter the response time or the higher the response frequency, the higher the communication willingness.

[0056] In this embodiment, when the communication intention is determined to be positive or the communication intention value is greater than a specified value, the communication intention is determined to meet the communication continuation condition. The communication continuation condition is used to define the intensity of the intention to continue communication.

[0057] In this embodiment, analyzing communication willingness can avoid excessive follow-up questions or awkward silences. When communication willingness is low, that is, when communication willingness does not meet the conditions for continued communication, the conversation can be terminated in a timely manner, switching to a simple recommendation mode or switching to manual customer service mode to avoid the user feeling interrupted or wasting time, and saving the number of model calls, thereby saving computing power. When communication willingness is high, that is, when communication willingness meets the conditions for continued communication, the intelligent shopping guide model can be called based on the generated conversation content and the follow-up question guide corresponding to the target symptom category to generate question messages to the user, achieving moderately in-depth questioning to obtain more information that is conducive to accurate drug recommendations.

[0058] In some embodiments, an artificial intelligence-based drug recommendation device can generate drug recommendation information based on the conversation content when it detects that the degree of information collection is as expected, or when it detects that there is information in the conversation content that the user requests drug recommendations, or when it analyzes that the user's acceptance of drug recommendations is as expected; wherein the degree of information collection is used to indicate the degree to which the conversation content conforms to the follow-up guideline corresponding to the target symptom category.

[0059] In this embodiment, in order to make timely drug recommendations, when the various information that should be collected as indicated by the follow-up guide is collected, it is determined that the degree of information collection meets expectations, and based on this, drug recommendation information is generated according to the content of the conversation, which can avoid inappropriate recommendations due to insufficient information, and can make more personalized and more appropriate drug recommendations, avoiding the problem of secondary communication due to insufficient previous communication. Specifically, the degree of information collection represents the matching ratio or completeness between the user information obtained in the current conversation (such as symptom details, medication history, allergy history, etc.) and the necessary items listed in the follow-up guide corresponding to the symptom category. The expectation refers to the limitation of the type of information that should be collected (such as gender, age, duration, symptoms, previous medications, etc.) / matching ratio (such as 80%).

[0060] In this embodiment, if a user requests a drug recommendation during a conversation, drug recommendation information is generated promptly based on the conversation content to respect the user's initiative. In some embodiments, if the information collected from the existing conversation content is insufficient, drug recommendation information can be generated based on the information already included in the conversation content. In some embodiments, if the information collected from the existing conversation content is insufficient, the intelligent shopping guide model is invoked according to the follow-up question guide to generate a question message to prompt the user to provide more information in order to provide accurate drug recommendations.

[0061] In this embodiment, the user's acceptance of drug recommendations can also be analyzed based on one or more information such as conversation content (e.g., containing positive sentiment words), historical conversation records (e.g., records of receiving drug recommendations), user portraits, behavioral characteristics (e.g., frequent proactive inquiries, fast reply speed, and rich reply content). Drug recommendation acceptance is used to measure the user's attitude towards the event of obtaining drug recommendations as soon as possible. When the drug recommendation acceptance meets the expected acceptance, drug recommendation information is generated based on the conversation content. When the drug recommendation acceptance does not meet the expected acceptance, forced recommendations can be avoided to avoid causing negative emotions in users. For users with a higher acceptance of drug recommendations, multiple rounds of in-depth conversations can be conducted. For users with a lower acceptance of drug recommendations, concise replies can be given or transferred to manual work to reasonably allocate shopping guide system resources.

[0062] In this embodiment, the drug recommendation information can be represented as text, interactive cards, links, controls, etc. The drug recommendation information includes information such as the drug name, dosage, usage, precautions, etc. For example, the drug recommendation information may be "Based on your current headache, you may consider taking ibuprofen sustained-release tablets, 200 mg each time, twice daily. If you experience stomach discomfort, you may also supplement with a gastric mucosal protective agent."

[0063] In some embodiments, the AI-based drug recommendation device is further configured with a timed return visit function, which is used to send return visit information to the user at a specified time according to the medication cycle corresponding to the drug recommendation information (e.g., how is the pain relief after taking the medication? Are there any adverse reactions?), so as to continuously optimize the smart shopping guide model based on the user's feedback content and feedback time to improve the recommendation accuracy of the smart shopping guide model. Among them, the specified time is related to the medication cycle, and the specified time corresponding to different drug recommendation information can be different.

[0064] In some embodiments, an artificial intelligence-based drug recommendation device can mine reference information used as a basis for drug recommendations from the generated conversation content during a conversation with a user, and call an intelligent shopping guide model based on the reference information to generate drug recommendation information; wherein the reference information includes at least one of the following: user needs, doubts related to drug safety, and user preferences.

[0065] In this embodiment, to make accurate drug recommendations, user needs can be analyzed using a multi-class intent classifier, a named entity recognition model, and a contextual association analysis model, which categorizes user input into user intents such as "symptom relief, preventive medication, and chronic maintenance." User needs, to a certain extent, indicate the user's expectations for drug recommendation information.

[0066] In this embodiment, the user's doubts about medication safety can also be determined based on the medication questions expressed by the user in the conversation (such as, "Are there any side effects? Can it be used together with XXX medicine?") and the negative emotional content contained in the analyzed conversation (such as, "This medicine has serious side effects!"). The doubts related to drug safety are used to characterize the user's concerns about medication safety and can be used to guide the shopping guide system to provide targeted responses.

[0067] In this embodiment, user preferences can also be analyzed based on historical purchase records, click-through rates, historical drug reviews, user portraits, and user self-descriptions (such as "I like xxx", "I don't want to eat XXX"). User preferences can be updated in real time with the recommendations in the dialogue rounds to ensure that the user preferences used as the basis for the final drug recommendation can accurately represent the user.

[0068] In this embodiment, the reference information including user needs, doubts related to drug safety, and user preferences refers to the data extracted and structured from each user input in multiple rounds of conversations with the user, which can be used to guide drug recommendation decisions.

[0069] See also Figure 3 The embodiment of the present application provides a drug recommendation method based on artificial intelligence. The drug recommendation method based on artificial intelligence can be applied to a drug recommendation device based on artificial intelligence. The drug recommendation method based on artificial intelligence can include the following steps.

[0070] Step S210: Obtain the applicable scope information of the drug and the actual dialogue information corresponding to the symptom category; wherein the actual dialogue information is the dialogue information generated by the communication between the user and the customer service.

[0071] Step S220: Based on the applicable scope information of the drug and the actual conversation information corresponding to the symptom category, call the high-order large model to generate a follow-up guide corresponding to each of the symptom categories; wherein the follow-up guide is used to indicate the information that needs to be obtained from the user before recommending the drug to the user; wherein the follow-up guide is used to be provided to the intelligent shopping guide model that communicates with the user, and the parameter scale of the intelligent shopping guide model is smaller than the parameter scale of the high-order large model.

[0072] In this embodiment, the scope of application information of a drug includes information such as indications, usage and dosage, contraindications, and population restrictions, and is used to indicate the disease scenarios in which the drug can be used safely and effectively. The scope of application information can be extracted from the drug package insert and / or the medical knowledge base.

[0073] In this embodiment, the actual conversation information corresponding to the symptom category records the conversation records generated in the real consultation scenario, including the symptom description raised by the user, follow-up questions, customer service answers, shopping guide words and multi-round conversation details, which can express the user's real language expression habits, concerns and needs.

[0074] For example, if the symptom category is excessive stomach acid, the shopping guide words include but are not limited to the following:

[0075]

[0076]

[0077] In this implementation, the high-level large model with a larger parameter scale possesses superior semantic understanding and knowledge integration capabilities compared to the smaller intelligent shopping guide model. Invoking the high-level large model based on scope information and actual conversation information can generate follow-up guidance for each symptom category. Invoking the intelligent shopping guide model based on the follow-up guidance eliminates the strategy generation task previously required by the intelligent shopping guide model. Invoking the intelligent shopping guide model within the shopping guide system allows for more efficient acquisition of the required shopping guide conversations and drug recommendation results.

[0078] In some embodiments, the artificial intelligence-based drug recommendation device can call a high-level large model to generate a thought chain based on the actual conversation information corresponding to the symptom category; wherein, the thought chain is used to provide to the intelligent shopping guide model for use as a basis for drug recommendation based on analysis of user needs, drug safety-related doubts, and at least one of user preferences.

[0079] In this implementation, a thought chain refers to the intermediate reasoning steps presented in readable text before the high-level large model generates the final reasoning result. These steps can represent the entire thought process from input (such as user symptom description and conversation context) to output (such as recommendation basis), making the reasoning path transparent and easy to check. Calling the high-level large model to generate a thought chain can leverage its stronger language processing capabilities to mine the causal relationships and medication logic in the corpus and represent them in the form of a thought chain. This, in turn, serves as the basis for distillation training of the intelligent shopping guide model. This can constrain the intelligent shopping guide model's reasoning process, making it rational and accurate, and reducing the intelligent shopping guide model's hallucination rate.

[0080] In some embodiments, the artificial intelligence-based drug recommendation device can call the high-order large model according to the actual dialogue information corresponding to the symptom category, so that the high-order large model generates a reference reasoning result based on the thinking chain; uses the reference reasoning result to perform distillation training on the smart shopping guide model; performs similarity evaluation on the output of the smart shopping guide model based on the reference reasoning result, and performs parameter iteration on the smart shopping guide model based on the similarity evaluation; wherein, the similarity evaluation is used to guide the output of the smart shopping guide model to approach the reference reasoning result of the high-order large model; the similarity evaluation is generated based on the thinking chain loss value and the model parameter loss value between the reference reasoning result and the output of the smart shopping guide model; the weight of the thinking chain loss value is less than the weight of the model parameter loss value.

[0081] In this embodiment, in order to obtain an intelligent shopping guide model with reasoning capabilities close to those of a high-order large model, the intelligent shopping guide model can be distilled and trained based on the reference reasoning results output by the high-order large model. In the distillation training, the high-order large model is the teacher model, and the intelligent shopping guide model is the student model. The training goal of the intelligent shopping guide model is to imitate the behavior of the teacher model. The intelligent shopping guide model trained in this way can reduce the parameter scale and reasoning cost while maintaining performance.

[0082] In this embodiment, the reference reasoning result refers to the information containing the complete reasoning trajectory and final decision output by the high-order large model based on the thinking chain under the premise of given input. By iterating the parameters of the intelligent shopping guide model, the output of the intelligent shopping guide model is close to the reasoning trajectory and final decision of the high-order large model.

[0083] In this embodiment, similarity evaluation is used to quantify the output gap between the intelligent shopping guide model and the high-order large model. Optionally, the similarity evaluation is specifically the weighted sum of the thinking chain loss value and the model parameter loss value. Among them, the thinking chain loss value is used to characterize the difference between the intermediate reasoning steps of the intelligent shopping guide model and the high-order large model, and the model parameter loss value is used to characterize the difference between the final output of the intelligent shopping guide model and the high-order large model. The thinking chain loss value and the model parameter loss value can be calculated based on cross entropy, mean square error, etc. Among them, the weight of the thinking chain loss value is less than the weight of the model parameter loss value, which is intended to guide the training results to first ensure the matching degree between the final product recommendation and user needs, which is conducive to obtaining an intelligent shopping guide model that is more suitable for application in the shopping guide system.

[0084] See also Figure 4. The embodiment of the present application also provides an artificial intelligence-based drug recommendation device, including: a category determination module, used to determine the target symptom category corresponding to the user among multiple symptom categories; wherein, the multiple symptom categories respectively correspond to follow-up guides generated by a high-order large model for the symptom category; the follow-up guide is used to indicate the information that needs to be obtained from the user before recommending drugs to the user; a message generation module, used to call the intelligent shopping guide model according to the generated conversation content and the follow-up guide corresponding to the target symptom category during the conversation with the user, and generate a question message to ask the user; wherein, the parameter scale of the intelligent shopping guide model is smaller than the parameter scale of the high-order large model.

[0085] Regarding the specific functions and effects achieved by the drug recommendation device based on artificial intelligence, you can refer to other embodiments of this application for comparative explanation and will not go into details here. The various modules in the drug recommendation device based on artificial intelligence can be implemented in whole or in part by software, hardware, and a combination thereof. The modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0086] See also Figure 5 . The embodiment of the present application also provides an artificial intelligence-based drug recommendation device, including: an information acquisition module, used to obtain the applicable scope information of the drug and the actual dialogue information corresponding to the symptom category; wherein, the actual dialogue information is the dialogue information generated by the communication between the user and the customer service; a guide generation module, used to call the high-order large model to generate a follow-up guide corresponding to each of the symptom categories based on the applicable scope information of the drug and the actual dialogue information corresponding to the symptom category; wherein, the follow-up guide is used to indicate the information that needs to be obtained from the user before recommending the drug to the user; wherein, the follow-up guide is used to provide to the intelligent shopping guide model that communicates with the user, and the parameter scale of the intelligent shopping guide model is smaller than the parameter scale of the high-order large model.

[0087] Regarding the specific functions and effects achieved by the drug recommendation device based on artificial intelligence, you can refer to other embodiments of this application for comparative explanation and will not go into details here. The various modules in the drug recommendation device based on artificial intelligence can be implemented in whole or in part by software, hardware, and a combination thereof. The modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0088] An embodiment of the present application also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the processor implements the aforementioned artificial intelligence-based drug recommendation method.

[0089] The present application also provides a computer program product comprising instructions, which, when executed by a processor, enables a computer to implement the aforementioned artificial intelligence-based drug recommendation method.

[0090] See also Figure 6 . The embodiment of this description can provide a computer device, which includes: a memory, and one or more processors communicatively connected to the memory; the memory stores instructions that can be executed by the one or more processors, and the instructions are executed by the one or more processors to enable the one or more processors to implement the aforementioned artificial intelligence-based drug recommendation method.

[0091] In some embodiments, the computer device may include a processor, a non-volatile storage medium, an internal memory, a communication interface, a display device, and an input device connected by a system bus. The non-volatile storage medium may store an operating system and related computer programs.

[0092] The user information or user account information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, etc.) involved in multiple implementation methods of this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0093] It should be understood that the specific examples herein are only intended to help those skilled in the art better understand the embodiments of the present application, rather than to limit the scope of the present invention.

[0094] It can be understood that in the various implementation methods of this application, the size of the serial number of each process does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the implementation method of this application.

[0095] It can be understood that the various embodiments described in this application can be implemented individually or in combination, and the embodiments of this application are not limited to this.

[0096] Unless otherwise indicated, all technical and scientific terms used in the embodiments of the present application have the same meaning as those commonly understood by those skilled in the art in the technical field of the present application. The terms used in this application are only for the purpose of describing specific embodiments and are not intended to limit the scope of this application. The term "and / or" used in this application includes any and all combinations of one or more related listed items. The singular forms "a", "above", and "the" used in the embodiments of the present application and the appended claims are also intended to include plural forms, unless the context clearly indicates otherwise.

[0097] It is understood that the processor in the embodiments of the present application can be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above-mentioned method embodiment can be completed by hardware integrated logic circuits in the processor or software instructions. The above-mentioned processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The various methods, steps, and logic block diagrams disclosed in the embodiments of the present application can be implemented or executed. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in the embodiments of the present application can be directly implemented and executed by a hardware decoding processor, or by a combination of hardware and software modules in the decoding processor. The software module can be located in a storage medium mature in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, etc. The storage medium is located in the memory, and the processor reads the information in the memory and, in conjunction with its hardware, completes the steps of the above-mentioned method.

[0098] It will be understood that the memory in the embodiments of the present application may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (programmable ROM, PROM), an erasable programmable read-only memory (erasable PROM, EPROM), an electrically erasable programmable read-only memory (EEPROM) or flash memory. The volatile memory may be a random access memory (RAM). It should be noted that the memory of the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0099] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0100] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, devices and units can refer to the corresponding processes in the aforementioned method implementation methods and will not be repeated here.

[0101] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0102] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of this embodiment.

[0103] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0104] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0105] The above description is merely a specific embodiment of the present application, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A drug recommendation method based on artificial intelligence, characterized in that: The method comprises: Determining a target symptom category corresponding to the user among multiple symptom categories; wherein each of the multiple symptom categories corresponds to a follow-up guide generated by the high-level large model for the symptom category; the follow-up guide is used to indicate information that needs to be obtained from the user before recommending a drug to the user; During a conversation with a user, the intelligent shopping guide model is called according to the generated conversation content and the follow-up guide corresponding to the target symptom category to generate a question message to the user; wherein the parameter scale of the intelligent shopping guide model is smaller than the parameter scale of the high-order large model.

2. The method according to claim 1, characterized in that During a conversation with a user, the steps of calling an intelligent shopping guide model based on the conversation content and the follow-up question guide corresponding to the target symptom category to generate a question message to the user include: In the process of communicating with users, analyze the communication intention based on the content of the conversation; When the communication willingness meets the conditions for continuing communication, the intelligent shopping guide model is called according to the generated dialogue content and the follow-up question guide corresponding to the target symptom category to generate a question message to the user.

3. The method according to claim 1, characterized in that The method further comprises: When it is detected that the degree of information collection is in line with expectations, or when it is detected that there is information in the conversation content that the user requests drug recommendations, or when it is analyzed that the user's acceptance of drug recommendations is in line with expectations, drug recommendation information is generated based on the conversation content; wherein, the degree of information collection is used to indicate the degree to which the conversation content complies with the follow-up guide corresponding to the target symptom category.

4. The method according to claim 3, characterized in that The steps for generating drug recommendation information based on the conversation content include: During a conversation with a user, reference information used as a basis for drug recommendations is mined from the generated conversation content, and an intelligent shopping guide model is called based on the reference information to generate drug recommendation information; wherein the reference information includes at least one of the following: user needs, doubts related to drug safety, and user preferences.

5. A drug recommendation method based on artificial intelligence, characterized in that: The method comprises: Obtaining drug application scope information and actual conversation information corresponding to symptom categories; wherein the actual conversation information is the conversation information generated by the communication between the user and the customer service; Based on the applicable scope information of the drug and the actual dialogue information corresponding to the symptom category, the high-order large model is called to generate a follow-up guide corresponding to each of the symptom categories; wherein the follow-up guide is used to indicate the information that needs to be obtained from the user before recommending the drug to the user; wherein the follow-up guide is used to be provided to the intelligent shopping guide model that communicates with the user, and the parameter scale of the intelligent shopping guide model is smaller than the parameter scale of the high-order large model.

6. The method according to claim 5, characterized in that The method further comprises: Based on the actual conversation information corresponding to the symptom category, a high-level large model is called to generate a thought chain; wherein, the thought chain is used to provide to the intelligent shopping guide model for use as a basis for recommending at least one drug based on analysis of user needs, doubts related to drug safety, and user preferences.

7. The method according to claim 6, characterized in that The method further comprises: Calling the high-level large model according to actual dialogue information corresponding to the symptom category, so that the high-level large model generates a reference reasoning result based on the thought chain; Performing distillation training on the intelligent shopping guide model using the reference inference results; Performing a similarity evaluation on the output of the intelligent shopping guide model based on the reference reasoning result, and iterating parameters of the intelligent shopping guide model based on the similarity evaluation; Among them, the similarity evaluation is used to guide the output of the intelligent shopping guide model to approach the reference reasoning result of the high-order large model; the similarity evaluation is generated based on the thinking chain loss value and model parameter loss value between the reference reasoning result and the output of the intelligent shopping guide model; the weight of the thinking chain loss value is less than the weight of the model parameter loss value.

8. A drug recommendation device based on artificial intelligence, characterized in that: The device comprises: A category determination module is configured to determine a target symptom category corresponding to a user from a plurality of symptom categories; wherein each of the plurality of symptom categories corresponds to a follow-up inquiry guide generated by the high-level large model for the symptom category; the follow-up inquiry guide is configured to indicate information that needs to be obtained from the user before recommending a drug to the user; The message generation module is used to call the intelligent shopping guide model according to the generated conversation content and the follow-up guide corresponding to the target symptom category during the conversation with the user, and generate a question message to ask the user; wherein the parameter scale of the intelligent shopping guide model is smaller than the parameter scale of the high-order large model.

9. A drug recommendation device based on artificial intelligence, characterized in that: The device comprises: An information acquisition module is used to obtain information on the scope of application of the drug and actual conversation information corresponding to the symptom category; wherein the actual conversation information is the conversation information generated by the communication between the user and the customer service; A guide generation module is used to call a high-order large model to generate a follow-up guide corresponding to each symptom category based on the scope of application information of the drug and the actual dialogue information corresponding to the symptom category; wherein the follow-up guide is used to indicate the information that needs to be obtained from the user before recommending the drug to the user; wherein the follow-up guide is used to be provided to the intelligent shopping guide model that dialogues with the user, and the parameter scale of the intelligent shopping guide model is smaller than the parameter scale of the high-order large model.

10. A computer device, characterized in that: The computer device includes a memory and a processor, wherein the memory stores at least one computer program, and the at least one computer program is loaded and executed by the processor to implement the artificial intelligence-based drug recommendation method as described in any one of claims 1 to 7.

11. A computer-readable storage medium, characterized in that The computer-readable storage medium stores at least one computer program, and when the at least one computer program is executed by the processor, it can implement the artificial intelligence-based drug recommendation method as described in any one of claims 1 to 7.

12. A computer program product, characterized in that The computer program product is used to implement the artificial intelligence-based drug recommendation method as described in any one of claims 1 to 7.

Images