Intelligent inquiry interaction method and device, storage medium and computer equipment

Abstract

According to the intelligent inquiry interaction method and device, the storage medium and the computer equipment provided by the invention, when the user needs to carry out inquiry, the inquiry prompt word corresponding to the inquiry case of the user is firstly determined, and then the target answer corresponding to the inquiry prompt word is returned by utilizing the target intelligent inquiry model; the answer range of the model is narrowed and guided through the inquiry prompt words, so that the accuracy of answers is improved, and the requirement of manual inquiry is reduced. When it is monitored that the user triggers an artificial inquiry event, similarity matching is conducted on all pre-stored expert label words based on the inquiry prompt words, a matching result is obtained, the matching result is optimized through an inquiry scoring system, a target inquiry expert corresponding to the inquiry prompt words is determined based on the optimized matching result, and then artificial inquiry is conducted. And the suitability of expert matching is improved. Through the method, the matching precision of the large language model and the experts can be improved through the inquiry keywords so as to meet the personalized requirements of the user, and then the inquiry quality and efficiency are improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to an intelligent medical consultation interaction method, device, storage medium and computer equipment. Background Art

[0002] In today's wave of digital transformation, artificial intelligence technology, especially the application of large language models, is gradually penetrating into all walks of life, profoundly changing traditional service models and user experience. Among them, the medical and health field, as the focus of people's livelihood, has an urgent need for efficient and accurate services. Based on this, the intelligent consultation interaction method based on open source large language models came into being.

[0003] However, the current large language model does not segment the application field, which means that its operation requires huge computing power support, and once faced with complex and changing segmented field questions, its limitations are fully exposed, so it is necessary to introduce manual consultation through manual or simple rule matching of consultation experts. Therefore, both the large language model and expert matching in this method have certain limitations, making it difficult to fully meet the personalized needs of users, resulting in low quality and efficiency of consultation. Summary of the invention

[0004] The purpose of this application is to solve at least one of the above-mentioned technical defects, especially the technical defects that the large language model and expert matching in the prior art have certain limitations and cannot fully meet the personalized needs of users, resulting in low quality and efficiency of medical consultation.

[0005] The present application provides an intelligent medical consultation interaction method, the method comprising:

[0006] Obtaining the medical consultation case input by the user, and determining the medical consultation prompt words corresponding to the medical consultation case;

[0007] Using the target intelligent medical inquiry model, returning a target answer corresponding to the medical inquiry prompt word, and returning a supplementary answer corresponding to the supplementary text after receiving the supplementary text input by the user;

[0008] When the user triggers a manual consultation event, similarity matching is performed on each expert tag word in the expert platform based on the consultation prompt word to obtain a matching result corresponding to each expert tag word;

[0009] The consultation scoring system is used to optimize the matching results of each expert tag word, so as to determine the target consultation expert corresponding to the consultation prompt word based on the optimized matching results, and establish a consultation channel between the target consultation expert and the user until the consultation is completed.

[0010] Optionally, determining the medical consultation prompt words corresponding to the medical consultation case includes:

[0011] Using the TF-IDF algorithm to mine information on the medical consultation cases, and extracting medical consultation keywords in the medical consultation cases according to the mining results;

[0012] Generate a medical consultation prompt word corresponding to the medical consultation case based on the medical consultation keyword.

[0013] Optionally, the using the target intelligent medical inquiry model to return a target answer corresponding to the medical inquiry prompt word includes:

[0014] The medical inquiry prompt words are input into the target intelligent medical inquiry model so that the target intelligent medical inquiry model extracts the text features of the medical inquiry prompt words, searches for answers in a language database based on the text features, and outputs a target answer corresponding to the medical inquiry prompt words based on the search results.

[0015] Optionally, returning a supplementary answer corresponding to the supplementary text after receiving the supplementary text input by the user includes:

[0016] When receiving the supplementary text input by the user, extracting the supplementary keywords in the supplementary text, and updating based on the supplementary keywords and the medical consultation prompt words;

[0017] The updated medical inquiry prompt words are input into the target intelligent medical inquiry model to obtain a supplementary answer corresponding to the supplementary text output by the target intelligent medical inquiry model.

[0018] Optionally, performing similarity matching on each expert tag word in the expert platform based on the medical consultation prompt word to obtain a matching result corresponding to each expert tag word includes:

[0019] Performing semantic analysis on the medical consultation prompt word, and determining the medical consultation field label corresponding to the medical consultation prompt word according to the analysis result;

[0020] The medical inquiry field label is matched with each expert label word in the expert platform for similarity, and a matching result corresponding to each expert label word is obtained.

[0021] Optionally, the use of the medical inquiry scoring system to optimize the matching results of each expert tag word includes:

[0022] For each expert tag word, obtain multiple scoring types in the medical consultation scoring system and the scoring score of the expert tag word under each scoring type;

[0023] An adjustment coefficient for each scoring type is determined, so as to adjust the weight of each scoring score according to each adjustment coefficient to obtain an optimization coefficient, and the expert tag word matching result is optimized based on the optimization coefficient.

[0024] Optionally, the determining a target consultation expert corresponding to the consultation prompt word based on the optimized matching result includes:

[0025] Determine the consulting experts corresponding to each expert tag word and the consulting status of each consulting expert, and select consulting experts whose consulting status is idle as candidate consulting experts;

[0026] The candidate consultation experts are sorted according to the optimized matching results to obtain a sorting result, and the candidate consultation expert with the highest similarity in the sorting result is used as the target consultation expert corresponding to the consultation prompt word.

[0027] The present application also provides an intelligent medical consultation interactive device, comprising:

[0028] A data acquisition module is used to acquire the medical consultation case input by the user and determine the medical consultation prompt words corresponding to the medical consultation case;

[0029] A model interaction module, used to return a target answer corresponding to the medical inquiry prompt word using a target intelligent medical inquiry model, and to return a supplementary answer corresponding to the supplementary text after receiving the supplementary text input by the user;

[0030] A label matching module is used to perform similarity matching on each expert label word in the expert platform based on the consultation prompt word when monitoring the user triggering the manual consultation event, and obtain a matching result corresponding to each expert label word;

[0031] The expert interaction module is used to optimize the matching results of each expert tag word using the consultation scoring system, so as to determine the target consultation expert corresponding to the consultation prompt word based on the optimized matching results, and establish a consultation channel between the target consultation expert and the user until the consultation is completed.

[0032] The present application also provides a storage medium, in which computer-readable instructions are stored. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the intelligent medical consultation interaction method as described in any one of claims 1 to 7.

[0033] The present application also provides a computer device, comprising: one or more processors, and a memory;

[0034] The memory stores computer-readable instructions, and when the computer-readable instructions are executed by the one or more processors, the steps of the intelligent medical consultation interaction method as described in any one of claims 1 to 7 are performed.

[0035] It can be seen from the above technical solutions that the embodiments of the present application have the following advantages:

[0036] The intelligent medical consultation interaction method, device, storage medium and computer equipment provided by the present application can first obtain the medical consultation case input by the user and determine the medical consultation prompt word corresponding to the medical consultation case when the user needs to conduct a medical consultation, and then use the target intelligent medical consultation model to return the target answer corresponding to the medical consultation prompt word, and return the supplementary answer corresponding to the supplementary text after receiving the supplementary text input by the user. Here, the medical consultation prompt word can limit and guide the answer range of the subsequent model, reduce the computing power and time required for model reasoning, improve the accuracy of the answer output, and reduce the need for manual medical consultation. In addition, when the user triggers a manual medical consultation event, the similarity matching of each expert label word in the expert platform can be performed based on the medical consultation prompt word to obtain the matching result corresponding to each expert label word, thereby improving the adaptability of expert matching. Then, the medical consultation scoring system can be used to optimize the matching results of each expert label word, so as to determine the target medical consultation expert corresponding to the medical consultation prompt word based on the optimized matching result, so as to further improve the adaptability of expert matching, and then establish a medical consultation channel between the target medical consultation expert and the user until the end of the medical consultation. Through this method, the present application can improve the answer accuracy of the large language model and the adaptation accuracy of expert matching through consultation keywords to meet the user's personalized needs, thereby improving the quality and efficiency of the consultation. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0038] Figure 1 A flowchart of an intelligent medical consultation interaction method provided in an embodiment of the present application;

[0039] Figure 2 A schematic diagram of the structure of an intelligent medical consultation interactive device provided in an embodiment of the present application;

[0040] Figure 3 A schematic diagram of the internal flow of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0041] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0042] The current large language model does not segment the application field, which means that its operation requires huge computing power support, and once faced with complex and changing segmented field questions, its limitations are fully exposed, so it is necessary to introduce manual consultation through manual or simple rule matching of consultation experts. Therefore, both the large language model and expert matching in this method have certain limitations, making it difficult to fully meet the personalized needs of users, resulting in low consultation quality and efficiency.

[0043] Based on this, this application proposes the following technical solutions, see below for details:

[0044] In one embodiment, Figure 1 As shown, Figure 1 A flowchart of an intelligent medical consultation interaction method provided in an embodiment of the present application; the present application provides an intelligent medical consultation interaction method, which specifically includes the following:

[0045] S110: Obtain the medical consultation case input by the user, and determine the medical consultation prompt words corresponding to the medical consultation case.

[0046] In this embodiment, when the user needs to conduct an online consultation, the user can input relevant consultation information into the computer device to form a consultation case, so that the computer device can generate corresponding consultation prompt words based on the consultation case, so that the consultation prompt words can be used to narrow and guide the answer scope of the subsequent model, thereby reducing the computing power and time required for model reasoning.

[0047] Specifically, when the computer device detects a consultation request sent by a user, it can obtain a corresponding blank consultation form and return it to the user, so that the user can enter relevant consultation information in the blank consultation form, such as individual medical history description, lesion site description, immune background description, etc., thereby forming a consultation case with complete data. The computer device can then automatically generate corresponding consultation prompt words based on the consultation case, so as to more quickly identify key symptoms and causes through the consultation prompt words, reduce the interference of redundant information, and thereby avoid unnecessary reasoning extensions in the subsequent model during the answering process, thereby ensuring the accuracy of the answer output.

[0048] It should be noted that the blank medical consultation form here consists of multiple interactive items, each of which can be divided into text description items and drop-down selection items, so that users can input relevant medical consultation information in a standardized and accurate manner, thereby improving the efficiency and accuracy of generating medical consultation prompts. In addition, the interactive items in the blank medical consultation form can also be divided into mandatory items and optional items, so that the basic information of the medical consultation case can be obtained through mandatory items, and the detailed characteristics of the disease can be improved through optional items to narrow the scope of answer reasoning.

[0049] For example, the interactive items in the blank form of the medical consultation of the present application may include individual symptom description, disease characteristic description, disease state description, etc. Among them, individual symptom description and disease state description are text description items, but individual symptom description is a required item, such as descriptions of Escherichia coli-piglet yellow diarrhea, porcine epidemic diarrhea or porcine transmissible gastroenteritis; the disease state description can be divided into motor / neurological signs, reproductive disorder signs and other related items, among which motor / neurological signs and reproductive disorder signs are required items, and other related items are optional items. The disease characteristic description is a drop-down selection item, which may include multiple characteristic descriptions such as incidence, mortality rate, course of disease, season of onset, and stage of onset. Multiple selection items are set under each characteristic description, such as the choice of incidence rate may include greater than or equal to 10% and less than 10%, the choice of mortality rate may include greater than or equal to 30% and less than 30%, and the choice of course of disease may include subclinical (within 7 days) and chronic (greater than 7 days), etc. These drop-down selection items are all required items.

[0050] S120: using the target intelligent medical inquiry model to return a target answer corresponding to the medical inquiry prompt word, and returning a supplementary answer corresponding to the supplementary text after receiving the supplementary text input by the user.

[0051] In this step, after determining the medical inquiry keywords corresponding to the medical inquiry case through step S110, the computer device can first determine the target intelligent medical inquiry model used for answer reasoning, and then input the medical inquiry keywords into the target medical inquiry model so that the target medical inquiry model outputs the target answer corresponding to the medical inquiry keywords. If the user still has questions about the target answer, the user can continue to input supplementary text for the medical inquiry case in the computer device so that the target medical inquiry model can continue to reason and output the supplementary answer corresponding to the supplementary text, thereby realizing intelligent medical inquiry interaction between the user and the target intelligent medical inquiry model, thereby improving the accuracy of the model answer output and reducing the need for manual medical consultation.

[0052] Among them, the target intelligent model refers to a model that infers the answer to the input medical inquiry keywords and obtains the target answer. When the target intelligent model is training the model, different types of medical inquiry keywords can be used as training samples, and each training sample is labeled with a sample label, that is, the corresponding target answer. When all training samples are labeled, the training samples with sample labels can be input into the preset pre-training model for forward propagation to train the model, and the preset loss function can be used to tune the model parameters during the back propagation process of the model. When the model meets certain training conditions or parameter convergence conditions, such as the number of iterations reaches the set value, the training is considered to be completed. At this time, the trained model can be used as the final target intelligent model. Therefore, after the computer device inputs the medical inquiry keywords into the target intelligent model, the target intelligent model can automatically infer the target answer to the medical inquiry keywords and output it.

[0053] It is understandable that when the computer device uses the target intelligent model to conduct a consultation interaction, it can first output the target answer corresponding to the consultation keyword through the target intelligent model to provide the customer with a preliminary diagnosis result. If the user still has questions or is dissatisfied with the target answer, he can also enter more supplementary text based on the current consultation case in the computer device, such as a more detailed description of the condition or new symptom changes. In this way, the computer device can perform deeper reasoning through the target intelligent model and output a supplementary answer corresponding to the supplementary text. In this way, a continuous and interactive intelligent consultation process can be formed between the user and the target intelligent consultation model. Each user's supplementary input will guide the model to conduct further reasoning and diagnosis, gradually refine and adjust the answer, thereby improving the accuracy of the answer output by the model and reducing dependence on manual consultation.

[0054] S130: When a user triggers a manual consultation event, similarity matching is performed on each expert tag word in the expert platform based on the consultation prompt word to obtain a matching result corresponding to each expert tag word.

[0055] In this step, after the consultation interaction with the target intelligent model is completed through step S120, the user can also initiate a request to enter manual consultation. When the computer device monitors the user triggering the manual consultation event, it can perform similarity matching on the expert label words corresponding to each expert in the expert platform based on the consultation prompt words, and obtain the matching result corresponding to each expert label word, so as to improve the adaptability between the experts connected to the manual consultation and the consultation cases, thereby improving the quality and efficiency of the consultation interaction.

[0056] Specifically, the computer device can perform similarity matching on each expert tag word in the access expert platform based on the previously generated consultation prompt words. There are multiple experts in the expert platform here, and each expert has a corresponding expert tag word, which is marked with key information such as the corresponding expert's professional field, disease type, diagnosis and treatment experience, etc. By comparing the similarity between the consultation prompt words and the expert tag words, the computer device can generate a matching result for each expert, and then recommend experts whose professional fields are highly consistent with the consultation cases to users.

[0057] Furthermore, during the user's consultation interaction with the target intelligent model, the computer device can monitor the user's behavior. When the user clicks the consultation end button, a pop-up window prompt for the need for manual consultation can be initiated in the consultation interaction interface. If the user chooses not to require manual consultation, the intelligent consultation interaction process is deemed to be completed, and the computer device exits the consultation interaction interface; if the user chooses to require manual consultation, a manual consultation event is triggered, and the corresponding expert is matched to the consultation case entered by the user to enter manual consultation. Therefore, after the user triggers the manual consultation event, the present application can use the consultation prompt word to match the expert, thereby greatly shortening the time it takes for the user to find the expert and maximizing the consistency between the expert and the consultation case.

[0058] S140: Optimize the matching results of each expert tag word using the consultation scoring system, determine the target consultation expert corresponding to the consultation prompt word based on the optimized matching results, and establish a consultation channel between the target consultation expert and the user until the consultation is completed.

[0059] In this step, after obtaining the matching results between the consultation keywords and each expert label word through S130, the computer device can also use the consultation scoring system to optimize the matching results of each expert label word, and determine the target consultation expert corresponding to the consultation prompt word based on the optimized matching results, thereby further improving the adaptability of the expert matching, and then establish a consultation channel between the target consultation expert and the user to meet the user's personalized needs until the consultation is completed and the consultation channel is disconnected.

[0060] Among them, the consultation scoring system is a scoring mechanism used to evaluate and optimize expert matching results, aiming to ensure that users can establish a consultation channel with the most suitable experts; the system can comprehensively evaluate experts through multiple scoring dimensions, such as user scores, administrator scores, number of consultation cases and response speed, so computer equipment can optimize the matching degree between experts and users based on these scoring results, ensuring the efficiency of consultation interaction and satisfaction of user experience.

[0061] It is understandable that the computer device can use the scoring system to comprehensively evaluate the matching results of the expert label words and the consultation prompt words, optimize the matching results, and then measure the compatibility between the expert and the consultation case through the optimized matching results. Once the target consultation expert corresponding to the consultation prompt word is determined, the computer device can automatically establish a consultation channel between the user and the target consultation expert, so that the user can directly communicate with the expert one-on-one, fully meeting the user's personalized needs; when the consultation is over, the computer device will automatically disconnect the consultation channel, thereby completing the entire intelligent consultation interaction process.

[0062] In the above embodiment, when the user needs to conduct a consultation, the consultation case input by the user can be obtained first, and the consultation prompt word corresponding to the consultation case can be determined, and then the target intelligent consultation model can be used to return the target answer corresponding to the consultation prompt word, and after receiving the supplementary text input by the user, the supplementary answer corresponding to the supplementary text can be returned. Here, the consultation prompt word can be used to narrow and guide the answer range of the subsequent model, reduce the computing power and time required for model reasoning, improve the accuracy of the answer output, and reduce the need for manual consultation. In addition, when the user triggers a manual consultation event, the similarity matching of each expert label word in the expert platform can be performed based on the consultation prompt word to obtain the matching result corresponding to each expert label word, thereby improving the adaptability of expert matching. Then, the consultation scoring system can be used to optimize the matching results of each expert label word, so as to determine the target consultation expert corresponding to the consultation prompt word based on the optimized matching result, so as to further improve the adaptability of expert matching, and then establish a consultation channel between the target consultation expert and the user until the consultation is completed. Through this method, the present application can improve the answer accuracy of the large language model and the adaptation accuracy of expert matching through consultation keywords to meet the user's personalized needs, thereby improving the quality and efficiency of the consultation.

[0063] In one embodiment, determining the consultation prompt words corresponding to the consultation case in step S110 may include:

[0064] S111: Use the TF-IDF algorithm to mine information on the medical consultation cases, and extract medical consultation keywords in the medical consultation cases based on the mining results.

[0065] S112: Generate a medical consultation prompt word corresponding to the medical consultation case based on the medical consultation keywords.

[0066] In this embodiment, when the computer device determines the medical consultation prompt words corresponding to the medical consultation case, it can use the TF-IDF algorithm to perform information mining on the medical consultation case, and extract the medical consultation keywords in the medical consultation case based on the mining results, and then generate the medical consultation prompt words corresponding to the medical consultation case based on the medical consultation keywords, so that the medical consultation prompt words can be used to narrow and guide the answer scope of the subsequent model, reduce the computing power and time required for model reasoning, and improve the accuracy of the answer output.

[0067] Among them, the TF-IDF algorithm refers to a statistical method used to evaluate the importance and relevance of a certain word in a document set, and is commonly used in information retrieval and text mining; it can measure the weight of a word in a document, while reducing the influence of words that appear commonly in all documents (such as "de", "shi"), so as to highlight the discriminative keywords. The TF-IDF algorithm mainly consists of two parts: term frequency (TF) and inverse document frequency (IDF). Here, the term frequency can be used to measure the frequency of a certain word appearing in a document, and the inverse document frequency can measure the universality of a certain word appearing in all documents.

[0068] Specifically, when the computer device extracts keywords from the medical consultation cases using the TF-IDF algorithm, it can first convert the text data in the medical consultation cases into a TF-IDF feature matrix based on the TfidfVectorizer tool in the Scikit-Learn library, and then extract the medical consultation keywords from it. In this process, first, the term frequency TF can be calculated, and the specific formula is as follows:

[0069]

[0070] In the formula, represents the number of times the word t appears in the medical consultation case d; represents the total number of words in the medical consultation case d.

[0071] Then, calculate the inverse document frequency TDF, and the specific formula is as follows:

[0072]

[0073] In the formula, N represents the total number of medical consultation cases in the database; represents the number of medical consultation cases containing the word t.

[0074] Finally, combine the term frequency TF and the inverse document frequency TDF into TF-IDF, and the specific formula is as follows:

[0075]

[0076] It can be understood that the medical consultation keywords in this application are extracted from the medical consultation cases input by the user through the TF-IDF algorithm. As the main content of the text data in the medical consultation cases, it can help the computer device identify and organize information, and represent the core concepts of the medical consultation cases; while the medical consultation prompt words are constructed based on the medical consultation keywords, and usually include context, questions or goals, which are used to help narrow the scope of the model's answers, reduce the divergence of the model, and guide the model to generate more accurate answers, mainly playing the role of restricting the model output.

[0077] For example, in a medical case input by a user, the computer device uses the TF-IDF algorithm to extract the medical keyword "severe diarrhea in weaned piglets" from the medical case input by the user. At this time, the medical prompt word constructed by the computer device based on the keywords of the file is "detailed explanation of the causes and treatment methods of diarrhea in piglets". In this way, the computer device can guide the model to generate more accurate answers.

[0078] In one embodiment, in step S120, using the target intelligent medical inquiry model to return a target answer corresponding to the medical inquiry prompt word may include:

[0079] S121: Input the medical inquiry prompt words into the target intelligent medical inquiry model so that the target intelligent medical inquiry model extracts the text features of the medical inquiry prompt words, searches for answers in the language database based on the text features, and outputs the target answers corresponding to the medical inquiry prompt words based on the search results.

[0080] In this embodiment, when the computer uses the target intelligent medical consultation model to return the target answer corresponding to the medical consultation prompt word, the medical consultation prompt word can be input into the target intelligent medical consultation model, so that the target intelligent medical consultation model can extract the text features of the medical consultation prompt word, and perform answer search in the language database based on the text features, and output the target answer corresponding to the medical consultation prompt word based on the search results, thereby realizing a high-precision automated medical consultation interaction process.

[0081] Among them, the language database is the core resource for building the target intelligent consultation model. It stores text information, terminology, knowledge and rules related to the corresponding field through the collection, processing, annotation, structuring, training and integration of large-scale text data. It can support the target intelligent consultation model to understand, process and retrieve the target prompt words, and generate accurate target answers and output them, thereby realizing an efficient and accurate automated consultation process.

[0082] Furthermore, after the target intelligent medical consultation model is obtained through training, the computer device can store the trained target intelligent medical consultation model so that when the answer to the medical consultation prompt word needs to be inferred and retrieved later, the pre-stored target intelligent medical consultation model can be directly called for answer inference retrieval. In addition, the target intelligent medical consultation model in this application can select a large language model (LLM) as a pre-training model for improvement and training, or other models with similar functions can be selected for improvement and training, without specific limitation here.

[0083] In one embodiment, after receiving the supplementary text input by the user, returning a supplementary answer corresponding to the supplementary text in step S120 includes:

[0084] S122: When the supplementary text input by the user is received, the supplementary keywords in the supplementary text are extracted, and the medical consultation prompt words are updated based on the supplementary keywords.

[0085] S123: Input the updated medical consultation prompt words into the target intelligent medical consultation model to obtain a supplementary answer corresponding to the supplementary text output by the target intelligent medical consultation model.

[0086] In this embodiment, after receiving the supplementary text input by the user, the computer device can extract the supplementary keywords in the supplementary text and update the medical consultation prompt words based on the supplementary keywords to construct a more complete context information. The computer device can then input the updated medical consultation prompt words into the target intelligent medical consultation model and obtain the supplementary answer corresponding to the supplementary text output by the target intelligent medical consultation model, thereby realizing high-precision medical consultation interaction between the user and the target intelligent medical consultation model.

[0087] It is understandable that the supplementary keywords extracted by the computer device from the supplementary text can provide new core content for the medical consultation case, and then update the medical consultation prompt words in real time, providing richer contextual information and details for the target intelligent medical consultation model, so that the target intelligent medical consultation model can perform in-depth analysis and reasoning to generate supplementary answers that are highly relevant to the supplementary text.

[0088] In one embodiment, in step S130, similarity matching is performed on each expert tag word in the expert platform based on the medical consultation prompt word to obtain a matching result corresponding to each expert tag word, which may include:

[0089] S131: semantically analyze the medical consultation prompt words, and determine the medical consultation field label corresponding to the medical consultation prompt words according to the analysis result.

[0090] S132: Perform similarity matching between the medical inquiry field label and each expert label word in the expert platform to obtain a matching result corresponding to each expert label word.

[0091] In this embodiment, after constructing the medical consultation prompt words of the medical consultation case, the computer device can first perform semantic analysis on the medical consultation prompt words, and determine the medical consultation field label corresponding to the medical consultation prompt words based on the analysis results, and then perform similarity matching between the medical consultation field label and each expert label word in the expert platform to obtain the matching result corresponding to each expert label word, thereby improving the adaptability of expert matching.

[0092] It is understandable that through semantic analysis of the consultation prompt words, the computer device can identify the core concepts, symptoms, disease names and related medical terms in the analysis results, and then convert them into structured information to form consultation field labels, such as porcine epidemic diarrhea, porcine transmissible gastroenteritis, etc. Through the consultation field labels, the computer device can quickly identify the user's consultation needs and lay the foundation for subsequent expert matching.

[0093] Specifically, the expert platform stores expert label words corresponding to multiple experts. These expert label words are marked with key information such as the corresponding expert's professional field, disease type, diagnosis and treatment experience, etc., which provide a basis for matching for the computer device. Therefore, the computer device can perform similarity matching between the consultation field label of the consultation case and each expert label word in the expert platform, so as to obtain the matching result corresponding to each expert label word. When performing similarity matching, the computer device can use a variety of matching algorithms, such as cosine similarity or more complex deep learning methods, in order to quantify the correlation between the two and obtain matching results.

[0094] In one embodiment, optimizing the matching results of each expert tag word by using the medical inquiry scoring system in step S140 may include:

[0095] S141: For each expert tag word, obtain multiple scoring types in the medical consultation scoring system and the scoring score of the expert tag word under each scoring type.

[0096] S142: Determine an adjustment coefficient for each scoring type, adjust the weight of each scoring score according to each adjustment coefficient, obtain an optimization coefficient, and optimize the matching result of the expert tag word based on the optimization coefficient.

[0097] In this embodiment, when optimizing the matching results of each expert label word, the computer device may first determine multiple scoring types in the medical consultation scoring system and the scoring scores of each expert label word under each scoring type. Then, for each expert label word, the computer device may determine the adjustment coefficient of each scoring type, and then perform weight adjustment on each scoring score corresponding to the expert label word according to each adjustment coefficient to obtain an optimization coefficient, and then optimize the matching results of the expert label word based on the optimization coefficient, thereby further improving the adaptability of expert matching.

[0098] It is understandable that the consultation scoring system stores the scores of each expert in multiple scoring types, so that the experts can be comprehensively evaluated from multiple dimensions through various scoring scores, and then the optimization coefficient is generated to optimize the matching results corresponding to each expert tag word. The scoring types here can include user scoring, administrator scoring, number of consultation cases and response speed.

[0099] To elaborate, user scoring refers to the scoring given by users to experts after completing the consultation, which reflects the user's satisfaction with the expert's service and is an important indicator for measuring the expert's performance in actual consultation; administrator scoring refers to the evaluation of experts by managers, which can review the expert's service quality from a more professional perspective; the number of consultation cases refers to the total number of consultations handled by experts on the platform, which is used to measure the expert's consultation experience level; response speed refers to the expert's response time to the user's consultation request, which is an important dimension for evaluating the expert's service capabilities and work efficiency.

[0100] It should be noted that the adjustment coefficients of each scoring type in this application can be flexibly adjusted according to the consultation scenario and user needs, and there is no restriction here. For example, in an emergency disease scenario, the adjustment coefficient of the response speed can be increased accordingly; in a complex disease scenario, the adjustment coefficient of the administrator score or the number of consultation cases can be increased accordingly; if the user values ​​the service attitude, the adjustment coefficient of the user score can be increased accordingly; if the user values ​​the professional background, the adjustment coefficient of the administrator score or the number of consultation cases can be increased accordingly.

[0101] In one embodiment, determining the target consultation expert corresponding to the consultation prompt word based on the optimized matching result in step S140 may include:

[0102] S143: Determine the consulting experts corresponding to each expert tag word and the consulting status of each consulting expert, and select consulting experts whose consulting status is idle as candidate consulting experts.

[0103] S144: Sort the candidate consultation experts according to the optimized matching results to obtain a sorting result, and use the candidate consultation expert with the highest similarity in the sorting result as the target consultation expert corresponding to the consultation prompt word.

[0104] In this embodiment, when the computer device determines the target consultation expert corresponding to the consultation prompt word, it can first determine the consultation experts corresponding to each expert label word and the consultation status of each consultation expert, and screen out the consultation experts whose consultation status is idle as candidate consultation experts, and then sort the candidate consultation experts according to the optimized matching result to obtain the sorting result, and use the candidate consultation expert with the highest similarity in the sorting result as the target consultation expert corresponding to the consultation prompt word, thereby improving the efficiency of the intelligent consultation interaction process.

[0105] It is understandable that the expert platform is provided with a consultation status corresponding to each expert, and the consultation status may include idle, busy or offline. Idle means that the expert currently has no consultation requests to be processed and is in a state where consultation interaction can be entered immediately; busy means that the expert is processing consultation requests and is connected to other users, and is in a state where new consultation interaction cannot be entered; and offline means that the expert is not currently logged into the expert platform or is in a state where no consultation can be received. Therefore, the computer device can filter out consultation experts with idle consultation status as candidate consultation experts, and select the target consultation expert from the candidate consultation experts to enter the consultation interaction, thereby improving the efficiency of the intelligent consultation interaction process.

[0106] Specifically, after generating the ranking results of each alternative consultation expert based on the optimized matching results, the computer device can automatically select the alternative consultation expert with the highest similarity in the ranking results as the target consultation expert corresponding to the consultation case input by the user, thereby ensuring that the target consultation expert matched to the user can be highly consistent with the user's needs in the professional field, thereby improving the quality of intelligent consultation interaction.

[0107] The following is a description of the intelligent medical consultation interaction device provided in an embodiment of the present application. The intelligent medical consultation interaction device described below and the intelligent medical consultation interaction method described above can be referenced to each other.

[0108] In one embodiment, Figure 2 As shown, Figure 2 A schematic diagram of the structure of an intelligent medical consultation interaction device provided in an embodiment of the present application; the present application also provides an intelligent medical consultation interaction device, including a data acquisition module 210, a model interaction module 220, a label matching module 230 and an expert interaction module 240, which specifically include the following:

[0109] The data acquisition module 210 is used to acquire the medical consultation case input by the user and determine the medical consultation prompt words corresponding to the medical consultation case.

[0110] The model interaction module 220 is used to use the target intelligent medical inquiry model to return the target answer corresponding to the medical inquiry prompt word, and to return the supplementary answer corresponding to the supplementary text after receiving the supplementary text input by the user.

[0111] The label matching module 230 is used to perform similarity matching on each expert label word in the expert platform based on the consultation prompt word when monitoring the user triggering the manual consultation event, and obtain the matching result corresponding to each expert label word.

[0112] The expert interaction module 240 is used to optimize the matching results of each expert tag word using the consultation scoring system, so as to determine the target consultation expert corresponding to the consultation prompt word based on the optimized matching results, and establish a consultation channel between the target consultation expert and the user until the consultation is completed.

[0113] In the above embodiment, when the user needs to conduct a consultation, the consultation case input by the user can be obtained first, and the consultation prompt word corresponding to the consultation case can be determined, and then the target intelligent consultation model can be used to return the target answer corresponding to the consultation prompt word, and after receiving the supplementary text input by the user, the supplementary answer corresponding to the supplementary text can be returned. Here, the consultation prompt word can be used to narrow and guide the answer range of the subsequent model, reduce the computing power and time required for model reasoning, improve the accuracy of the answer output, and reduce the need for manual consultation. In addition, when the user triggers a manual consultation event, the similarity matching of each expert label word in the expert platform can be performed based on the consultation prompt word to obtain the matching result corresponding to each expert label word, thereby improving the adaptability of expert matching. Then, the consultation scoring system can be used to optimize the matching results of each expert label word, so as to determine the target consultation expert corresponding to the consultation prompt word based on the optimized matching result, so as to further improve the adaptability of expert matching, and then establish a consultation channel between the target consultation expert and the user until the consultation is completed. Through this method, the present application can improve the answer accuracy of the large language model and the adaptation accuracy of expert matching through consultation keywords to meet the user's personalized needs, thereby improving the quality and efficiency of the consultation.

[0114] In one embodiment, the data acquisition module 210 may include:

[0115] The keyword extraction submodule is used to mine information of medical consultation cases using the TF-IDF algorithm, and to extract medical consultation keywords in the medical consultation cases based on the mining results.

[0116] The prompt word generation submodule is used to generate the consultation prompt words corresponding to the consultation case based on the consultation keywords.

[0117] In one embodiment, the model interaction module 220 may include:

[0118] The answer retrieval submodule is used to input the medical consultation prompt words into the target intelligent medical consultation model so that the target intelligent medical consultation model can extract the text features of the medical consultation prompt words, perform answer retrieval in the language database based on the text features, and output the target answer corresponding to the medical consultation prompt words based on the retrieval results.

[0119] In one embodiment, the model interaction module 220 further includes:

[0120] The prompt word updating submodule is used to extract the supplementary keywords in the supplementary text when receiving the supplementary text input by the user, and update the medical consultation prompt words based on the supplementary keywords.

[0121] The answer supplement submodule is used to input the updated medical consultation prompt words into the target intelligent medical consultation model to obtain the supplementary answers corresponding to the supplementary text output by the target intelligent medical consultation model.

[0122] In one embodiment, the tag matching module 230 may include:

[0123] The label determination submodule is used to perform semantic analysis on the medical consultation prompt words and determine the medical consultation field label corresponding to the medical consultation prompt words according to the analysis results.

[0124] The label matching submodule is used to perform similarity matching between the labels of the consultation field and each expert label word in the expert platform to obtain the matching results corresponding to each expert label word.

[0125] In one embodiment, optimizing the matching results of each expert tag word by using the medical inquiry scoring system in step S140 may include:

[0126] The submodule is used to obtain, for each expert label word, multiple scoring types in the medical consultation scoring system, and the scoring score of the expert label word under each scoring type.

[0127] The submodule is used to determine the adjustment coefficient of each scoring type, to adjust the weight of each scoring score according to each adjustment coefficient, to obtain the optimization coefficient, and to optimize the matching result of the expert tag word based on the optimization coefficient.

[0128] In one embodiment, the expert interaction module 240 may include:

[0129] The expert screening submodule is used to determine the consulting experts corresponding to each expert tag word and the consulting status of each consulting expert, and screen out consulting experts whose consulting status is idle as candidate consulting experts.

[0130] The expert consultation submodule is used to sort the candidate consultation experts according to the optimized matching results to obtain the sorting results, and use the candidate consultation expert with the highest similarity in the sorting results as the target consultation expert corresponding to the consultation prompt word.

[0131] In one embodiment, the present application also provides a storage medium, which stores computer-readable instructions. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the intelligent medical consultation interaction method as described in any of the above embodiments.

[0132] In one embodiment, the present application also provides a computer device having computer-readable instructions stored therein. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the intelligent medical consultation interaction method as described in any one of the above embodiments.

[0133] Indicatively, Figure 3 As shown, Figure 3 This is a schematic diagram of the internal structure of a computer device provided in an embodiment of the present application. The computer device 300 may be provided as a server. Figure 3 The computer device 300 includes a processing component 302, which further includes one or more processors, and a memory resource represented by a memory 301, for storing instructions executable by the processing component 302, such as an application. The application stored in the memory 301 may include one or more modules, each corresponding to a set of instructions. In addition, the processing component 302 is configured to execute instructions to perform the intelligent consultation interaction method of any of the above embodiments.

[0134] The computer device 300 may further include a power supply component 303 configured to perform power management of the computer device 300, a wired or wireless network interface 304 configured to connect the computer device 300 to a network, and an input / output (I / O) interface 305. The computer device 300 may operate based on an operating system stored in the memory 301, such as Windows Server TM, Mac OS X TM, Unix TM, Linux TM, Free BSD TM, or the like.

[0135] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0136] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0137] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The various embodiments can be combined as needed, and the same or similar parts can refer to each other.

[0138] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An intelligent medical consultation interactive method, characterized in that: The method comprises: Obtaining the medical consultation case input by the user, and determining the medical consultation prompt words corresponding to the medical consultation case; Using the target intelligent medical inquiry model, returning a target answer corresponding to the medical inquiry prompt word, and returning a supplementary answer corresponding to the supplementary text after receiving the supplementary text input by the user; When the user triggers a manual consultation event, similarity matching is performed on each expert tag word in the expert platform based on the consultation prompt word to obtain a matching result corresponding to each expert tag word; The consultation scoring system is used to optimize the matching results of each expert tag word, so as to determine the target consultation expert corresponding to the consultation prompt word based on the optimized matching results, and establish a consultation channel between the target consultation expert and the user until the consultation is completed.

2. The intelligent medical consultation interaction method according to claim 1, characterized in that: The step of determining the medical consultation prompt words corresponding to the medical consultation case includes: Using the TF-IDF algorithm to mine information on the medical consultation cases, and extracting medical consultation keywords in the medical consultation cases according to the mining results; Generate a medical consultation prompt word corresponding to the medical consultation case based on the medical consultation keyword.

3. The intelligent medical consultation interactive method according to claim 1, characterized in that: The target intelligent medical inquiry model is used to return a target answer corresponding to the medical inquiry prompt word, including: The medical inquiry prompt words are input into the target intelligent medical inquiry model so that the target intelligent medical inquiry model extracts the text features of the medical inquiry prompt words, searches for answers in a language database based on the text features, and outputs a target answer corresponding to the medical inquiry prompt words based on the search results.

4. The intelligent medical consultation interactive method according to claim 1, characterized in that: The step of returning a supplementary answer corresponding to the supplementary text after receiving the supplementary text input by the user comprises: When receiving the supplementary text input by the user, extracting supplementary keywords in the supplementary text, and updating the medical consultation prompt words based on the supplementary keywords; The updated medical inquiry prompt words are input into the target intelligent medical inquiry model to obtain a supplementary answer corresponding to the supplementary text output by the target intelligent medical inquiry model.

5. The intelligent medical consultation interaction method according to claim 1, characterized in that: The similarity matching of each expert label word in the expert platform based on the medical consultation prompt word is performed to obtain a matching result corresponding to each expert label word, including: Performing semantic analysis on the medical consultation prompt word, and determining the medical consultation field label corresponding to the medical consultation prompt word according to the analysis result; The medical inquiry field label is matched with each expert label word in the expert platform for similarity, and a matching result corresponding to each expert label word is obtained.

6. The intelligent medical consultation interaction method according to claim 1, characterized in that: The use of the questionnaire scoring system to optimize the matching results of each expert tag word includes: For each expert tag word, obtain multiple scoring types in the medical consultation scoring system and the scoring score of the expert tag word under each scoring type; An adjustment coefficient for each scoring type is determined, so as to adjust the weight of each scoring score according to each adjustment coefficient to obtain an optimization coefficient, and the matching result of the expert tag word is optimized based on the optimization coefficient.

7. The intelligent medical consultation interactive method according to claim 1, characterized in that: The step of determining the target consultation expert corresponding to the consultation prompt word based on the optimized matching result includes: Determine the consulting experts corresponding to each expert tag word and the consulting status of each consulting expert, and select consulting experts whose consulting status is idle as candidate consulting experts; The candidate consultation experts are sorted according to the optimized matching results to obtain a sorting result, and the candidate consultation expert with the highest similarity in the sorting result is used as the target consultation expert corresponding to the consultation prompt word.

8. An intelligent medical consultation interactive device, characterized in that: include: A data acquisition module is used to acquire the medical consultation case input by the user and determine the medical consultation prompt words corresponding to the medical consultation case; A model interaction module, used to return a target answer corresponding to the medical inquiry prompt word using a target intelligent medical inquiry model, and to return a supplementary answer corresponding to the supplementary text after receiving the supplementary text input by the user; A label matching module is used to perform similarity matching on each expert label word in the expert platform based on the consultation prompt word when monitoring the user triggering the manual consultation event, and obtain a matching result corresponding to each expert label word; The expert interaction module is used to optimize the matching results of each expert tag word using the consultation scoring system, so as to determine the target consultation expert corresponding to the consultation prompt word based on the optimized matching results, and establish a consultation channel between the target consultation expert and the user until the consultation is completed.

9. A storage medium, characterized in that: The storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the intelligent medical consultation interaction method as described in any one of claims 1 to 7.

10. A computer device, characterized in that: include: one or more processors, and memory; The memory stores computer-readable instructions, and when the computer-readable instructions are executed by the one or more processors, the steps of the intelligent medical consultation interaction method as described in any one of claims 1 to 7 are performed.

Images