Automatic inquiry data updating system and method, computer equipment and storage medium

Abstract

The invention discloses an inquiry data automatic updating system, an inquiry data automatic updating method, computer equipment and a storage medium. In one scheme realized by the invention, a data acquisition module is used for acquiring all inquiry data of an inquiry platform; the data screening module is used for marking all the obtained inquiry data by utilizing an open source large model based on the evaluation standard of the inquiry data, and screening the marked inquiry data to obtain first target inquiry data, the evaluation standard of the inquiry data comprises five parts, namely dialogue normality, dialogue integrity, dialogue complexity, dialogue accuracy and dialogue timeliness; and the data updating module is used for updating the inquiry data set by using the screened first target inquiry data. According to the method, the first target inquiry data meeting the requirement is screened out by using the open source large model to update the data set, it is ensured that the inquiry model reflects the newest medical knowledge and clinical practice, doctors are helped to timely obtain the frontier information, and therefore the accuracy and reliability of the model are improved.

Description

Technical Field

[0001] This invention relates to the field of medical data processing, and in particular to an automatic update system, method, computer device, and storage medium for consultation data. Background Technology

[0002] With the widespread application of artificial intelligence technology in the medical field, consultation models have become an important tool for medical consultation and diagnosis. However, existing consultation models still have problems in practical applications, such as inconsistent data quality, lagging knowledge updates, and excessive human intervention. These problems affect the data updates of consultation models, resulting in insufficient accuracy. Therefore, a new consultation model is urgently needed to solve the accuracy problem of consultation models. Summary of the Invention

[0003] This invention provides an automatic update system, method, computer device, and storage medium for consultation data to address the problem of insufficient accuracy in consultation models.

[0004] Firstly, an automatic update system for consultation data is provided, including: The data acquisition module is used to acquire all consultation data from the online consultation platform. The data filtering module is used to label all the consultation data based on the evaluation criteria of the consultation data using an open-source large model, and to filter the labeled consultation data to obtain the first target consultation data. The evaluation criteria of the consultation data include five parts: dialogue normality, dialogue completeness, dialogue complexity, dialogue accuracy, and dialogue timeliness. The data update module is used to update the consultation dataset using the filtered first target consultation data.

[0005] Secondly, a method for automatically updating consultation data is provided, including: Obtain all consultation data from the online consultation platform; Based on the evaluation criteria of the consultation data, all the acquired consultation data are labeled using an open-source large model, and the labeled consultation data is filtered to obtain the first target consultation data. The evaluation criteria of the consultation data include five parts: dialogue normality, dialogue completeness, dialogue complexity, dialogue accuracy, and dialogue timeliness. Update the consultation dataset using the selected first target consultation data.

[0006] Thirdly, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-mentioned automatic consultation data update system or automatic consultation data update method.

[0007] Fourthly, a computer-readable storage medium is provided, which stores a computer program that, when executed by a processor, implements the above-mentioned automatic consultation data update system or method.

[0008] In one scheme implemented by the aforementioned automatic update system, method, computer equipment, and storage medium for consultation data, a data acquisition module is used to acquire all consultation data from the consultation platform; a data filtering module is used to label all acquired consultation data based on evaluation criteria of consultation data using an open-source large model, and to filter the labeled consultation data to obtain first target consultation data, wherein the evaluation criteria of the consultation data include five parts: dialogue normality, dialogue completeness, dialogue complexity, dialogue accuracy, and dialogue timeliness; a data update module is used to update the consultation dataset using the filtered first target consultation data. This application uses an open-source large model to filter out qualified data to train the consultation model and regularly updates the dataset to ensure that the consultation model reflects the latest medical knowledge and clinical practice, helping doctors to obtain the most cutting-edge information in a timely manner, thereby improving the accuracy and reliability of the model. Attached Figure Description

[0009] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a schematic diagram of an application environment for an automatic consultation data update system according to an embodiment of the present invention; Figure 2 This is a schematic diagram of an automatic update system for consultation data in one embodiment of the present invention; Figure 3 This is another flowchart of the automatic update method for consultation data in one embodiment of the present invention; Figure 4 This is a schematic diagram of a computer device according to an embodiment of the present invention. Detailed Implementation

[0011] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0012] It is worth noting that all consultation data used in this application was obtained with the user's authorization and the technical solutions provided in this application were legally implemented.

[0013] This invention provides a method for automatically updating consultation data, which can be applied to, for example... Figure 1 The application environment is shown. Specifically, this automatic consultation data update method is applied in an automatic consultation data update system, which includes, for example,... Figure 1 The client and server shown communicate via a network. The server is used to run various modules, including: a data acquisition module for acquiring all consultation data from the consultation platform; a data filtering module for labeling all acquired consultation data using an open-source large model based on evaluation criteria, and filtering the labeled consultation data to obtain first target consultation data. The evaluation criteria for consultation data include five parts: dialogue normality, dialogue completeness, dialogue complexity, dialogue accuracy, and dialogue timeliness; and a data update module for updating the consultation dataset using the filtered first target consultation data. This application uses an open-source large model to filter out qualified data to train the consultation model and regularly updates the dataset to ensure that the consultation model reflects the latest medical knowledge and clinical practice, helping doctors to obtain the most cutting-edge information in a timely manner, thereby improving the accuracy and reliability of the model. The client, also known as the user terminal, refers to the program that provides local services to the client, corresponding to the server. The client can be installed on, but is not limited to, various personal computers, laptops, smartphones, and tablets. The automatic update system for consultation data in this application is implemented on a server, which can be a standalone server or a server cluster consisting of multiple servers.

[0014] In one embodiment, such as Figure 2 As shown, an automatic update system for consultation data is provided. The automatic update system for consultation data includes: a data acquisition module, a data filtering module, and a data update module. The specific details of each module are as follows: The data acquisition module is used to acquire all consultation data from the consultation platform; Specifically, all consultation data left by patients on the online consultation platform is obtained. This consultation data includes existing online dialogue records between doctors and patients. The consultation data includes the patient's personal information (name, age, gender, phone number, residential address), the patient's medical records (symptoms, past medical history, allergy history, medication), consultation ID, consultation time, patient's chief complaint, attending physician information (name, ID, department, etc.), message ID, message content, message sender, diagnosis, and prescription. The obtained user information is organized, and invalid content is removed. Each piece of consultation data is then organized into the form of a consultation dialogue for subsequent evaluation of the consultation data.

[0015] In this embodiment, by acquiring all consultation data from the consultation platform and selectively filtering the data, the consultation data included in the dataset is of high quality, thereby improving the accuracy and reliability of the model.

[0016] Preferably, after the data acquisition module obtains all consultation data from the consultation platform, it will filter out the required consultation data according to the target field, and organize the filtered consultation data into the format of consultation dialogue to obtain the organized consultation data.

[0017] Specifically, for the obtained consultation data, the required data is filtered according to the required fields (department, time). For example, if it is necessary to organize the consultation data of pediatrics, the data is filtered according to "department: pediatrics", and the filtered consultation data is organized into a format in which each data is a consultation dialogue.

[0018] The data filtering module is used to label all the consultation data based on the evaluation criteria of the consultation data and the open-source large model to filter out the first target consultation data. The evaluation criteria of the consultation data include five parts: dialogue normality, dialogue completeness, dialogue complexity, dialogue accuracy, and dialogue timeliness.

[0019] Specifically, the data filtering module filters and evaluates all acquired consultation data and labels all consultation data. Consultation data that conforms to the training consultation model is labeled as first target consultation data. The evaluation criteria for conforming to the training consultation data are based on five dimensions: dialogue normality, dialogue completeness, dialogue complexity, dialogue accuracy, and dialogue timeliness. The five criteria are determined based on the data evaluation criteria formulated by professionals for each dimension. Based on the evaluation criteria of consultation data, an open-source large model is used to label different dimensions of all consultation data, and first target consultation data that meets the requirements for each dimension and second target consultation data that does not meet the requirements are filtered out. The open-source large model provides the data to be judged and consultation data to professionals for selection. The provided data is organized in a few-shot format. During the process of judging different dimensions of consultation data, different prompts are provided for professionals to filter consultation data of different dimensions.

[0020] It is important to note that the "dialogue normality" refers to whether there are any typos in the patient's or doctor's communication during the consultation data. Typos can confuse the consultation model, especially in highly specialized medical scenarios where different words have different meanings. Therefore, typos can mislead the model's understanding and cause data bias during data filtering. When training with a large open-source model, the model will be configured based on the chosen expert's selection, for example: PROMPT_KEY="dialogue normality", PROMPT=no typos in the user's or doctor's messages during the consultation, RESULT="conclusion: yes or no". Consultation data is filtered based on the dimension of "dialogue normality" according to the choices of professionals. For example, "PROMPT_KEY="dialogue normality", where PROMPT means there are no typos in the user's or doctor's messages in the consultation dialogue, and RESULT="conclusion: yes" means there are typos in the consultation data. "PROMPT_KEY="dialogue normality", where PROMPT means there are no typos in the user's or doctor's messages in the consultation dialogue, and RESULT="conclusion: no" means there are no typos in the consultation data.

[0021] In this embodiment, by judging the normality of the consultation data dialogue, it is determined whether there are typos in the consultation data, thereby improving the consultation model's understanding of the consultation data and ensuring that the consultation data included in the dataset has high quality, thus improving the accuracy and reliability of the model.

[0022] It is important to note that the completeness of the dialogue in the consultation data refers to whether the consultation dialogue includes the complete consultation process. A typical consultation process involves "the doctor understanding the chief complaint, the doctor inquiring in detail about the present illness, past medical history, family history, auxiliary examinations, diagnosis, treatment, and lifestyle suggestions." Sometimes, patients leave midway through the consultation due to personal issues, refusing to continue. This can cause the consultation model to make a diagnostic error, preventing it from accurately understanding the patient's condition and providing professional advice. Therefore, during the screening of consultation data, it is necessary to assess the completeness of the dialogue. The system determines whether the consultation data was obtained according to the principle of "doctor understands the chief complaint, doctor inquires in detail about the present medical history, past medical history, family history, auxiliary examinations, diagnosis, treatment, and lifestyle suggestions." When training with an open-source large model, the system will determine whether the consultation data is complete based on the professional's choice. If "PROMPT_KEY=" dialogue completeness, PROMPT = consultation dialogue includes the complete consultation process: "doctor understands the chief complaint, doctor inquires in detail about the present medical history, past medical history, family history, auxiliary examinations, diagnosis, treatment, and lifestyle suggestions," and RESULT = "conclusion: yes," it means that the consultation data has complete dialogue. If "PROMPT_KEY=" dialogue completeness, PROMPT = consultation dialogue includes the complete consultation process: "doctor understands the chief complaint, doctor inquires in detail about the present medical history, past medical history, family history, auxiliary examinations, diagnosis, treatment, and lifestyle suggestions," and RESULT = "conclusion: no," it means that the consultation data does not have complete dialogue. Consultation data that meets the dialogue completeness criteria is marked as the first target consultation data.

[0023] In this embodiment, the completeness of the consultation data flow is determined by whether the consultation dialogue includes a complete consultation process. This improves the consultation model's understanding of the consultation data, ensures that the consultation data included in the dataset has high quality, and thus improves the accuracy and reliability of the model.

[0024] It is important to note that the dialogue complexity refers to whether the dialogue in the consultation data can be broken down into several relatively complete dialogues. For example, during the consultation, a patient may ask detailed questions about their own health problems, and after addressing their own issues, they may then consult about their child's problems. Although different patients can be consulted during the consultation process, simultaneously consulting with two or more patients in one consultation can train the consultation model to better understand the patients' expressions, thus improving its comprehension ability. However, if a single consultation includes consultations with multiple patients, the consultation model may have limited understanding of language complexity, leading to misunderstandings. If "PROMPT_KEY="dialogue complexity", PROMPT="dialogue can be broken down into several relatively complete dialogues", and RESULT="conclusion: integer", then the complexity of the dialogue is within an acceptable range. If "PROMPT_KEY="dialogue complexity", PROMPT="dialogue can be broken down into several relatively complete dialogues", and RESULT="conclusion: non-integer", then the complexity of the dialogue is within an unacceptable range. Based on the choices of professionals, the dialogue complexity of the consultation data can be judged, and the first target consultation data that meets the requirements can be selected.

[0025] In this embodiment, by determining whether the consultation dialogue can be broken down into several relatively complete dialogues, it is determined whether the dialogue complexity is acceptable, thereby improving the consultation model's understanding of the consultation data and ensuring that the consultation data included in the dataset has high quality, thus improving the accuracy and reliability of the model.

[0026] It is important to note that the accuracy of the dialogue refers to the accuracy of the doctor's messages, including questions and answers, during the consultation, ensuring there are no irrelevant inquiries or incorrect medical knowledge responses. If "PROMPT_KEY="Dialogue Accuracy", PROMPT="Whether there are inappropriate or incorrect messages from the doctor in the consultation", and RESULT="Conclusion: Yes", it indicates that no inappropriate errors occurred during the consultation. If "PROMPT_KEY="Dialogue Accuracy", PROMPT="Whether there are inappropriate or incorrect messages from the doctor in the consultation", and RESULT="Conclusion: No", it indicates that inappropriate errors occurred during the consultation. Based on the choices of professionals, the accuracy of the consultation data is judged, and the primary target consultation data that meets the requirements is selected.

[0027] In this embodiment, the messages from the doctor during the consultation, including questions and answers, are accurate. This helps determine if there are any errors in the medical knowledge provided, improves the consultation model's understanding of the consultation data, and ensures that the consultation data included in the dataset has high quality, thereby improving the model's accuracy and reliability.

[0028] It's important to note that the "timeliness of the dialogue" refers to whether the consultation dialogue involves time-sensitive medical knowledge. For example, during the COVID-19 outbreak, there was a surge in COVID-19 consultations, followed by a sharp decrease afterward. Therefore, during certain periods, the database needs to add relevant consultation dialogues so that the large consultation dialogue model can handle the surge in time-sensitive consultations, alleviating manual workload. If PROMPT_KEY="Dialogue Timeliness", PROMPT="Is the consultation dialogue time-sensitive?", RESULT="Conclusion: Yes, Time-sensitive topic: xxx", then the consultation data is time-sensitive. If PROMPT_KEY="Dialogue Timeliness", PROMPT="Is the consultation dialogue time-sensitive?", RESULT="Conclusion: No", then the consultation data is not time-sensitive. Time-sensitive consultation data will be updated periodically, and consultations for this type of disease will be added to the database. Then, after this period, this type of data will be removed based on the "time-sensitive topic".

[0029] In this embodiment, the complexity of the dialogue is determined by whether it involves time-sensitive medical knowledge. This allows for the updating of time-sensitive medical knowledge, improving the model's understanding of the consultation data and ensuring that the consultation data included in the dataset has high quality, thereby improving the model's accuracy and reliability.

[0030] In this application, professionals provide first-target diagnostic data that meets the requirements under five dimensions: dialogue normality, dialogue completeness, dialogue complexity, dialogue accuracy, and dialogue timeliness, as well as second-target diagnostic data that does not meet the requirements. Specifically, for each of the five dimensions—dialogue normality, dialogue completeness, dialogue complexity, dialogue accuracy, and dialogue timeliness—approximately five pieces of first-target diagnostic data that meet the requirements and five pieces of second-target diagnostic data that do not meet the requirements are selected. The data is evaluated using an open-source large model, and the selected data is used as the basis for subsequent training of the large-scale diagnostic dialogue model, so that the diagnostic data is constantly updated, thereby improving the accuracy of the diagnostic model.

[0031] The data update module is used to train the consultation model using the selected first target consultation data and to update the consultation dataset periodically.

[0032] Specifically, in the data update module, the selected first-target consultation data that meets the requirements is used to train the consultation model. The trained consultation model is then automatically updated according to a preset time. Within the preset time, the consultation model acquires consultation data that meets the criteria of dialogue normality, dialogue completeness, dialogue complexity, dialogue accuracy, and dialogue timeliness, updates the consultation dataset, and uses the updated data to train the consultation model. For consultation data that meets the criteria of timeliness, after the generation of timely consultation data, the corresponding consultation data can be automatically added to train the consultation model, and a timely consultation module can be added to the consultation model. When the consultation model, through data analysis of the consultation data, determines that the number of patients with timely consultation data has significantly decreased, it is considered that the time-sensitive disease is nearing its end. At this point, based on the results of the data analysis, the added timely consultation module can be deleted, and consultation data that meets the criteria of dialogue normality, dialogue completeness, dialogue complexity, dialogue accuracy, and dialogue timeliness can be acquired again to train the consultation model. The consultation model is then updated within the preset time. In this application, as the training data is continuously updated, the amount and quality of the data are also continuously improved, thus obtaining an automatically updated diagnostic model that continuously improves its diagnostic capabilities.

[0033] Preferably, in the data update module, no update time is set for the consultation data with dialogue timeliness, and the corresponding consultation data is deleted after the dialogue timeliness expires.

[0034] Specifically, after time-sensitive data is added to the training dataset, the related dialogue data does not have a fixed update time. One week after the occurrence of the time-sensitive dialogue data, the relevant data is filtered from the database and updated in real-time through a data filtering module after evaluating the normality, completeness, complexity, and accuracy of the dialogue in the consultation data. This updated data is then added to the training dataset. If a time-sensitive event has ended, it needs to be filtered from the training data to determine if the event has ended and then deleted. In this application, as the training data is continuously updated, the data volume and quality also continuously improve, laying the foundation for updating the consultation dataset.

[0035] In this embodiment, the consultation model is trained using the selected first target consultation data and the consultation model is updated regularly. This ensures that the updated consultation model reflects the latest medical knowledge and clinical practice, helping doctors to obtain the most cutting-edge information in a timely manner and improve the diagnosis and treatment of patients.

[0036] In one embodiment, a method for automatically updating consultation data is provided, such as... Figure 3 As shown, the specific steps are as follows: S10: Obtain all consultation data from the consultation platform.

[0037] Specifically, consultation data refers to records generated during medical and health consultations on online consultation platforms. It mainly includes the patient's personal information (name, age, gender, phone number, residential address), the patient's medical records (symptoms, past medical history, allergy history, medication), consultation ID, consultation time, patient's chief complaint, attending physician information (name, ID, department, etc.), message ID, message content, message sender, diagnosis, prescription, etc. This application focuses on the training and updating of the consultation model and does not impose specific restrictions on the specific content of the consultation data obtained from the consultation platform.

[0038] It should be noted that since this application directly obtains the consultation data from the consultation platform, after obtaining all the consultation data from the consultation platform, it is necessary to organize and filter the consultation data to remove irrelevant dialogues and redundant data. Specifically, step S10, that is, after obtaining all the consultation data from the consultation platform, may include the following steps: performing part-of-speech tagging and removing invalid words on all consultation data to obtain valid consultation data, and organizing each piece of consultation data into the form of a consultation dialogue for subsequent evaluation of the consultation data.

[0039] Preferably, after the data acquisition module obtains all consultation data from the consultation platform, it will filter out the required consultation data according to the target field, and organize the filtered consultation data into the format of consultation dialogue to obtain the organized consultation data.

[0040] Specifically, for the obtained consultation data, the required data is filtered according to the required fields (department, time). For example, if it is necessary to organize the consultation data of pediatrics, the data is filtered according to "department: pediatrics", and the filtered consultation data is organized into a format in which each data is a consultation dialogue.

[0041] In this embodiment, by acquiring all consultation data from the consultation platform and selectively filtering the data, the consultation data included in the dataset is of high quality, thereby improving the accuracy and reliability of the model.

[0042] S20: Based on the evaluation criteria of the consultation data, all consultation data are labeled using an open-source large model to filter out the first target consultation data. The evaluation criteria of the consultation data include five parts: normality of dialogue, integrity of dialogue, complexity of dialogue, accuracy of dialogue, and timeliness of dialogue.

[0043] Specifically, to screen and evaluate all acquired consultation data and label all consultation data, consultation data that conforms to the training consultation model is labeled as the first target consultation data. The evaluation criteria for conforming to the training consultation data are based on five dimensions: dialogue normality, dialogue completeness, dialogue complexity, dialogue accuracy, and dialogue timeliness. The five criteria are determined based on the data evaluation criteria formulated by professionals for each dimension. Based on the evaluation criteria of consultation data, an open-source large model is used to label different dimensions of all consultation data, and the first target consultation data that meets the requirements of each dimension is screened out. The open-source large model provides the data to be judged and the consultation data to be selected by professionals. The provided data is organized in a few-shot format. During the process of judging different dimensions of consultation data, different prompts are provided for professionals to filter consultation data of different dimensions.

[0044] The "dialogue normality" refers to whether there are any typos in the patient's or doctor's communication during the consultation data. Typos can confuse the consultation model, especially in highly specialized medical scenarios where different words have different meanings. Therefore, typos can mislead the model's understanding and cause data bias during data filtering. When training with a large open-source model, the model is configured based on the chosen expert's selection, for example: PROMPT_KEY="dialogue normality", PROMPT=no typos in the user's or doctor's messages during the consultation, RESULT="conclusion: yes or no". Consultation data is filtered based on the dimension of "dialogue normality" according to the choices of professionals. For example, "PROMPT_KEY="dialogue normality", where PROMPT means there are no typos in the user's or doctor's messages in the consultation dialogue, and RESULT="conclusion: yes" means there are typos in the consultation data. "PROMPT_KEY="dialogue normality", where PROMPT means there are no typos in the user's or doctor's messages in the consultation dialogue, and RESULT="conclusion: no" means there are no typos in the consultation data.

[0045] In this embodiment, by judging the normality of the consultation data dialogue, it is determined whether there are typos in the consultation data, thereby improving the consultation model's understanding of the consultation data and ensuring that the consultation data included in the dataset has high quality, thus improving the accuracy and reliability of the model.

[0046] Dialogue completeness refers to whether the dialogue in the consultation data includes the complete consultation process. A typical consultation process includes: "Doctor understands the chief complaint, doctor inquires in detail about the present medical history, past medical history, family history, auxiliary examinations, diagnosis, treatment, and lifestyle suggestions." Sometimes, patients leave midway through the consultation due to personal issues, refusing to continue. This can cause the consultation model to make a diagnostic error, preventing it from accurately understanding the patient's condition and providing professional advice. Therefore, in the process of screening consultation data, it is necessary to judge the completeness of the dialogue, determining whether the consultation data follows the process of "Doctor understands the chief complaint, doctor inquires in detail about the present medical history, past medical history, family history, auxiliary examinations, diagnosis, treatment, and lifestyle suggestions." When training with an open-source large model, based on the professional's selection, if "PROMPT_KEY="Dialogue Completeness", PROMPT = Consultation dialogue includes the complete consultation process: "Doctor understands the chief complaint, doctor inquires in detail about the present medical history, past medical history, family history, auxiliary examinations, diagnosis, treatment, and lifestyle suggestions", and RESULT = "Conclusion: Yes", it indicates that the dialogue in the consultation data is complete. If "PROMPT_KEY=" dialogue integrity", PROMPT = the consultation dialogue includes the complete consultation process: "the doctor understands the chief complaint, the doctor inquires in detail about the present medical history, past medical history, family history, auxiliary examinations, diagnosis, treatment, and lifestyle suggestions", RESULT = "conclusion: no" indicates that the consultation data does not have completeness during the consultation. Consultation data that meets the dialogue integrity is marked as the first target consultation data.

[0047] In this embodiment, the completeness of the consultation data flow is determined by whether the consultation dialogue includes a complete consultation process. This improves the consultation model's understanding of the consultation data, ensures that the consultation data included in the dataset has high quality, and thus improves the accuracy and reliability of the model.

[0048] Dialogue complexity refers to whether the dialogue in the consultation data can be broken down into several relatively complete dialogues. For example, during the consultation, a patient asks detailed questions about their own health problems, and after finishing their own questions, they then consult about their child's problems. Although different patients can be consulted during the consultation process, consulting two or more patients simultaneously in one consultation can train the consultation model to better understand the patients' expressions, thus improving the model's comprehension ability. However, if a consultation involves multiple patients in one consultation, it may lead to a limited understanding of language complexity in the consultation model, resulting in misunderstandings. If "PROMPT_KEY="dialogue complexity", PROMPT="dialogue can be broken down into several relatively complete dialogues", and RESULT="conclusion: integer", then the complexity of the dialogue is within an acceptable range. If "PROMPT_KEY="dialogue complexity", PROMPT="dialogue can be broken down into several relatively complete dialogues", and RESULT="conclusion: non-integer", then the complexity of the dialogue is within an unacceptable range. Based on the choices of professionals, the dialogue complexity of the consultation data can be judged, and the first target consultation data that meets the requirements can be selected.

[0049] In this embodiment, by determining whether the consultation dialogue can be broken down into several relatively complete dialogues, it is determined whether the dialogue complexity is acceptable, thereby improving the consultation model's understanding of the consultation data and ensuring that the consultation data included in the dataset has high quality, thus improving the accuracy and reliability of the model.

[0050] Dialogue accuracy refers to the accuracy of the doctor's messages, including questions and answers, during the consultation, ensuring there are no irrelevant inquiries or incorrect medical knowledge responses. If "PROMPT_KEY="Dialogue Accuracy", PROMPT="Whether there are inappropriate or incorrect messages from the doctor in the consultation", and RESULT="Conclusion: Yes", it indicates no inappropriate errors occurred during the consultation. If "PROMPT_KEY="Dialogue Accuracy", PROMPT="Whether there are inappropriate or incorrect messages from the doctor in the consultation", and RESULT="Conclusion: No", it indicates inappropriate errors occurred during the consultation. Based on the choices of professionals, the dialogue accuracy of the consultation data is judged, and the primary target consultation data that meets the requirements is selected.

[0051] In this embodiment, the messages from the doctor during the consultation, including questions and answers, are accurate. This helps determine if there are any errors in the medical knowledge provided, improves the consultation model's understanding of the consultation data, and ensures that the consultation data included in the dataset has high quality, thereby improving the model's accuracy and reliability.

[0052] The timeliness of a consultation refers to whether the consultation involves timely medical knowledge. For example, during the COVID-19 outbreak, there was a surge in COVID-19-related consultations, followed by a sharp decrease after that period. Therefore, during certain special periods, the database needs to add relevant consultation dialogues so that the large consultation dialogue model can handle the surge in timely consultations, alleviating manual pressure. If PROMPT_KEY="Conversation Timeliness", PROMPT="Is the consultation dialogue timely?", RESULT="Conclusion: Yes, Timely Topic: xxx", then the consultation data is timely. If PROMPT_KEY="Conversation Timeliness", PROMPT="Is the consultation dialogue timely?", RESULT="Conclusion: No", then the consultation data is not timely. Timely consultation data will be updated periodically, and consultations for this type of disease will be added to the database. Then, after this period, this type of data will be removed based on the "timely topic".

[0053] In this embodiment, the complexity of the dialogue is determined by whether it involves time-sensitive medical knowledge. This allows for the updating of time-sensitive medical knowledge, improving the model's understanding of the consultation data and ensuring that the consultation data included in the dataset has high quality, thereby improving the model's accuracy and reliability.

[0054] In this application, professionals provide first-target diagnostic data that meets the requirements under five dimensions: dialogue normality, dialogue completeness, dialogue complexity, dialogue accuracy, and dialogue timeliness, as well as second-target diagnostic data that does not meet the requirements. Specifically, for each of the five dimensions, approximately five pieces of first-target diagnostic data that meet the requirements and five pieces of second-target diagnostic data that do not meet the requirements are selected. The data is evaluated using an open-source large model, and the selected data is used as the basis for subsequent training of the large-scale diagnostic dialogue model. This ensures that the diagnostic data is constantly updated, thereby improving the accuracy of the diagnostic model.

[0055] In this example, the system assesses typos, consultation process, timeliness, and completeness of the consultation dialogue within the consultation data to filter out the first target consultation data. This application utilizes a large open-source model for filtering, ensuring that the consultation data included in the dataset meets the requirements for the first target consultation data, thereby improving the accuracy and reliability of the consultation model.

[0056] S30: Train the consultation model using the selected first target consultation data, and update the consultation model.

[0057] Specifically, the first set of target consultation data that meets the requirements is used to update the consultation dataset. The data in the consultation dataset is used to train the consultation model, and the trained consultation model is automatically updated according to a preset time. The consultation model acquires consultation data that meets the requirements of dialogue normality, dialogue completeness, dialogue complexity, dialogue accuracy, and dialogue timeliness within the preset time, and trains the consultation model. For consultation data that meets the requirements of timeliness, after the occurrence of time-sensitive consultation data, the corresponding consultation data can be automatically added to train the consultation model, and a time-sensitive consultation module can be added to the consultation model. When the consultation model believes, through data analysis of the consultation data, that the number of patients with timeliness has decreased significantly, it is considered that the time-sensitive disease has ended. At this time, according to the results of data analysis, the added time-sensitive consultation module can be deleted, and consultation data that meets the requirements of dialogue normality, dialogue completeness, dialogue complexity, dialogue accuracy, and dialogue timeliness can be acquired again to retrain the consultation model. The consultation model is then updated within the preset time. In this application, as the training data is continuously updated, the amount and quality of the data are also continuously improved, thus obtaining an automatically updated diagnostic model that continuously improves its diagnostic capabilities.

[0058] Preferably, in the data update module, no update time is set for the consultation data with dialogue timeliness, and the corresponding consultation data is deleted after the dialogue timeliness expires.

[0059] Specifically, after time-sensitive data is added to the training dataset, the related dialogue data does not have a fixed update time. One week after the occurrence of the time-sensitive dialogue, the relevant data is filtered from the database and updated in real-time through a data filtering module after evaluating the normality, completeness, complexity, and accuracy of the dialogue in the consultation data. This updated data is then added to the training dataset. If a time-sensitive event has ended, it needs to be filtered from the training data to determine if the event has ended and then deleted. In this application, as the training data is continuously updated, the data volume and quality also continuously improve, laying the foundation for updating the consultation data.

[0060] In this embodiment, the consultation model is trained using the selected first target consultation data and the consultation model is updated regularly. This ensures that the updated consultation model reflects the latest medical knowledge and clinical practice, helping doctors to obtain the most cutting-edge information in a timely manner and improve the diagnosis and treatment of patients.

[0061] In one scheme implemented by the aforementioned automatic update system, method, computer equipment, and storage medium for consultation data, a data acquisition module is used to acquire all consultation data from the consultation platform; a data filtering module is used to label all acquired consultation data based on evaluation criteria of consultation data using an open-source large model, and to filter the labeled consultation data to obtain first target consultation data, wherein the evaluation criteria of the consultation data include five parts: dialogue normality, dialogue completeness, dialogue complexity, dialogue accuracy, and dialogue timeliness; a data update module is used to update the consultation dataset using the filtered first target consultation data. This application uses an open-source large model to filter out target consultation records that meet the requirements to train the consultation model, and regularly updates the dataset to ensure that the consultation model reflects the latest medical knowledge and clinical practice, helping doctors to obtain the most cutting-edge information in a timely manner, thereby improving the accuracy and reliability of the model.

[0062] It should be understood that the sequence number of each step in the above embodiments does not imply 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 embodiments of the present invention.

[0063] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 3 As shown, the computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The computer device is used to acquire all consultation data from the online consultation platform; based on the evaluation criteria of the consultation data, it uses an open-source large model to label all the consultation data to filter out first target consultation data. The evaluation criteria for the consultation data include five parts: dialogue normality, dialogue completeness, dialogue complexity, dialogue accuracy, and dialogue timeliness; the selected first target consultation data is used to train the consultation model, and the consultation model is updated regularly. By using an open-source large model to filter out the first target consultation data that meets the requirements to train the consultation model, and by regularly updating the dataset, the consultation model is ensured to reflect the latest medical knowledge and clinical practice, helping doctors to obtain the most cutting-edge information in a timely manner, thereby improving the accuracy and reliability of the model. The network interface of this computer device is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements a system and method for automatically updating medical consultation data.

[0064] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the automatic consultation data update system of the above embodiment, for example... Figure 2 The diagram shown illustrates an automatic consultation data update system, or it can be used to implement the automatic consultation data update method described in the above embodiments, such as... Figure 3 The flowchart illustrating the automatic update method for consultation data in the above embodiment is omitted here to avoid repetition. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in this embodiment of the automatic update system for consultation data, for example... Figure 2 The consultation data shown is automatically updated in each module of the system. To avoid duplication, it will not be described in detail here.

[0065] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When executed by a processor, the computer program implements the automatic consultation data update system of the above embodiment, for example... Figure 2 The diagram shown illustrates an automatic consultation data update system, or it can be used to implement the automatic consultation data update method described in the above embodiments, such as... Figure 3 The flowchart illustrating the automatic update method for consultation data in the above embodiment is omitted here to avoid repetition. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in this embodiment of the automatic update system for consultation data, for example... Figure 2 The consultation data shown is automatically updated in various modules of the system; to avoid duplication, it will not be described again here. The computer-readable storage medium can be non-volatile or volatile.

[0066] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. This computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0067] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0068] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A system for automatically updating medical consultation data, characterized in that: include: Data collection module, used to obtain all consultation data on the consultation platform; A data screening module, for marking all the acquired medical consultation data using an open source big model based on the evaluation criteria of the medical consultation data, and screening the marked medical consultation data to obtain the first target medical consultation data, wherein the evaluation criteria of the medical consultation data include five parts: dialogue normality, dialogue completeness, dialogue complexity, dialogue accuracy, and dialogue timeliness; A data updating module is used to update the medical consultation data set using the screened first target medical consultation data.

2. The automatic updating system for medical consultation data according to claim 1, characterized in that: The system includes a data acquisition module, which is used to obtain all the consultation data from the consultation platform, filter out the required consultation data according to the target field, and organize the filtered consultation data into the format of consultation dialogue to obtain the organized consultation data.

3. The automatic updating system for medical consultation data according to claim 1, characterized in that: The evaluation criteria for the medical consultation data are used to judge the typos, medical consultation process, timeliness of the medical consultation and completeness of the medical consultation dialogue in the medical consultation data, so as to screen out the first target medical consultation data.

4. The automatic updating system for medical consultation data according to claim 1, characterized in that: The evaluation criteria for the medical inquiry data are used to screen the first target medical inquiry data and the second target medical inquiry data, and use the screened first target medical inquiry data and the second target medical inquiry data as model training data, wherein the first target medical inquiry data is standard medical inquiry data and the second target medical inquiry data is non-standard medical inquiry data.

5. The automatic updating system for medical consultation data according to claim 1, characterized in that: The data updating module does not set a corresponding update time for updating the medical consultation data with dialogue timeliness, and deletes the corresponding medical consultation data after the dialogue timeliness has passed.

6. A method for automatically updating medical consultation data, characterized in that: The system according to claims 1 to 5 comprises: Obtain all consultation data from the consultation platform; Based on the evaluation criteria of the medical consultation data, all the acquired medical consultation data are marked using the open source big model, and the marked medical consultation data are screened to obtain the first target medical consultation data, wherein the evaluation criteria of the medical consultation data include five parts: dialogue normality, dialogue completeness, dialogue complexity, dialogue accuracy, and dialogue timeliness; The medical inquiry data set is updated using the filtered first target medical inquiry data.

7. The method for automatically updating medical consultation data according to claim 6, characterized in that: The evaluation criteria for the medical consultation data are used to judge the typos, medical consultation process, timeliness of the medical consultation and completeness of the medical consultation dialogue in the medical consultation data, so as to screen out the first target medical consultation data.

8. The method for automatically updating medical consultation data according to claim 6, characterized in that: The medical inquiry data is updated using the first target medical inquiry data that is screened out, including: not setting a corresponding update time for updating the medical inquiry data with conversation timeliness, and deleting the corresponding medical inquiry data after the conversation timeliness has passed.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the system for automatically updating medical inquiry data according to any one of claims 1 to 5 or the method for automatically updating medical inquiry data according to any one of claims 6 to 8 is implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the system for automatically updating medical inquiry data according to any one of claims 1 to 5 or the method for automatically updating medical inquiry data according to any one of claims 6 to 8 is implemented.

Images