Quality inspection method for inquiry data and related device

Abstract

The invention discloses an inquiry data quality inspection method and a related device, and relates to the technical field of natural language processing, and the method comprises the steps: obtaining a large language model with the capability of carrying out quality inspection on inquiry data through training in advance, and enabling the large language model to serve as an inquiry data quality inspection model, and carrying out the quality inspection of the inquiry data after the inquiry data to be subjected to quality inspection is obtained; according to the method, the inquiry data quality inspection model is called firstly to generate the error information of the inquiry data, then the inquiry data quality inspection model is called, and the inquiry data is repaired based on the error information of the inquiry data, so that quality inspection of the inquiry data can be realized, and the inquiry data after quality inspection is generated. According to the invention, intelligent, efficient and reliable inquiry data quality inspection is realized based on the strong natural language processing capability of the large language model.

Description

Technical Field

[0001] The present application relates to the technical field of natural language processing, and in particular to a medical inquiry data quality inspection method and related devices. Background Art

[0002] As a powerful natural language processing technology, large language models are being widely used in the medical field, mainly in the areas of auxiliary diagnosis, auxiliary treatment, and auxiliary medical research. In order for large language models to better provide intelligent and personalized medical services to doctors and patients, they often need to be pre-trained or supervised fine-tuned using consultation data. However, there are often various problems in consultation data (such as wrong consultation direction, redundant consultation, confusing consultation logic, etc.), which poses a hidden danger to the pre-training or supervised fine-tuning of large language models, and may not be able to improve the natural language processing capabilities of large language models in the medical field. Therefore, it is particularly necessary to conduct quality inspection on consultation data to improve the quality of consultation data.

[0003] The current quality inspection method for medical consultation data is mainly carried out manually, that is, medical personnel manually inspect the medical consultation data. However, there is a large amount of medical consultation data, and the efficiency of manual quality inspection of medical consultation data is low, which undoubtedly brings a huge burden to the work of medical personnel. Moreover, due to the limitation of the accumulation of professional knowledge of medical personnel themselves, the reliability of manual quality inspection of medical consultation data cannot be guaranteed.

[0004] Therefore, how to provide a method for quality inspection of medical consultation data to achieve intelligent, efficient and reliable quality inspection of medical consultation data has become a technical problem that needs to be solved urgently by technical personnel in this field. Summary of the invention

[0005] In view of the above problems, this application provides a method and related devices for quality inspection of medical consultation data to achieve intelligent, efficient and reliable quality inspection of medical consultation data, thereby improving the quality of medical consultation data. The specific scheme is as follows:

[0006] The first aspect of the present application provides a method for quality inspection of medical consultation data, comprising:

[0007] Obtain the consultation data to be inspected;

[0008] Based on the medical inquiry data, the medical inquiry data quality inspection model is called to generate error information of the medical inquiry data; the medical inquiry data quality inspection model is a large language model obtained through training and capable of performing quality inspection on the medical inquiry data, and the error information of the medical inquiry data is used to indicate the quality inspection results of each round of medical inquiry dialogue in the medical inquiry data;

[0009] Based on the medical inquiry data and the error information of the medical inquiry data, the medical inquiry data quality inspection model is called to generate repaired medical inquiry data, and the repaired medical inquiry data is the medical inquiry data after quality inspection.

[0010] In a possible implementation, before calling the medical inquiry data quality inspection model based on the medical inquiry data to generate error information of the medical inquiry data, the method further includes:

[0011] A medical inquiry data error location instruction format template and a medical inquiry data repair instruction format template are generated based on an instruction generation model, wherein the instruction generation model is a large language model obtained through training and has the ability to generate the medical inquiry data error location instruction format template and the medical inquiry data repair instruction format template.

[0012] In a possible implementation, the generating of the inquiry data error location instruction format template and the inquiry data repair instruction format template based on the instruction generation model includes:

[0013] Obtain a preconfigured automatic generation instruction template for medical inquiry data quality inspection instructions, wherein the automatic generation instruction template for medical inquiry data quality inspection instructions includes a goal of the automatic generation task for medical inquiry data quality inspection instructions, a slot for filling description information of the medical inquiry data quality inspection task, a slot for filling examples of the medical inquiry data quality inspection task, and a slot for filling annotation specifications and regulations;

[0014] Fill the description information of the medical inquiry data quality inspection task into the description information filling slot of the medical inquiry data quality inspection task, fill the medical inquiry data quality inspection task example into the medical inquiry data quality inspection task example filling slot, and fill the annotation specification regulations into the annotation specification regulations filling slot to obtain the medical inquiry data quality inspection instruction automatic generation instruction;

[0015] The medical inquiry data quality inspection instruction automatic generation instruction is input into the instruction generation model to obtain the medical inquiry data error location instruction format template and the medical inquiry data repair instruction format template generated by the instruction generation model.

[0016] In a possible implementation, calling the medical inquiry data quality inspection model based on the medical inquiry data to generate error information of the medical inquiry data includes:

[0017] Obtaining a medical inquiry data error location instruction format template, wherein the medical inquiry data error location instruction format template includes medical inquiry data error location requirement information and a medical inquiry data filling slot;

[0018] Filling the medical inquiry data into the medical inquiry data filling slot to obtain a medical inquiry data error locating instruction;

[0019] The medical inquiry data error location instruction is input into the medical inquiry data quality inspection model to obtain error information of the medical inquiry data generated by the medical inquiry data quality inspection model.

[0020] In a possible implementation, calling the medical inquiry data quality inspection model based on the medical inquiry data and the error information of the medical inquiry data to generate the repaired medical inquiry data includes:

[0021] Obtaining a consultation data repair instruction format template, wherein the consultation data repair instruction format template includes consultation data repair requirement information, consultation data filling slots, and error information filling slots;

[0022] Filling the medical inquiry data into the medical inquiry data filling slot, filling the error information of the medical inquiry data into the error information filling slot, and obtaining a medical inquiry data repair instruction;

[0023] The medical inquiry data repair instruction is input into the medical inquiry data quality inspection model to obtain the repaired medical inquiry data generated by the medical inquiry data quality inspection model.

[0024] In a possible implementation, after calling the medical inquiry data quality inspection model based on the medical inquiry data to generate error information of the medical inquiry data, and calling the medical inquiry data quality inspection model based on the medical inquiry data and the error information of the medical inquiry data to generate repaired medical inquiry data, the method further includes:

[0025] Determining the accuracy of locating errors in the medical inquiry data based on the error information of the medical inquiry data, and determining the qualified rate of repairing the medical inquiry data based on the repaired medical inquiry data;

[0026] If the error location accuracy of the inquiry data does not meet the preset conditions, and / or the repair qualification rate of the inquiry data does not meet the preset conditions, the instruction generation model is iteratively optimized.

[0027] In a possible implementation, the iteratively optimizing the instruction generation model includes:

[0028] Integrating the medical inquiry data error location instruction format template to obtain an integrated medical inquiry data error location instruction format template;

[0029] Integrating the medical inquiry data repair instruction format template based on the repaired medical inquiry data to obtain an integrated medical inquiry data repair instruction format template;

[0030] The instruction generation model is iteratively optimized based on the integrated medical inquiry data error location instruction format template and the integrated medical inquiry data repair instruction format template.

[0031] A second aspect of the present application provides a medical inquiry data quality inspection device, comprising:

[0032] An acquisition unit, used for acquiring the medical consultation data to be inspected;

[0033] A medical inquiry data error locating unit, configured to call a medical inquiry data quality inspection model based on the medical inquiry data to generate error information of the medical inquiry data; the medical inquiry data quality inspection model is a large language model obtained through training and capable of performing quality inspection on medical inquiry data, and the error information of the medical inquiry data is used to indicate the quality inspection results of each round of medical inquiry dialogue in the medical inquiry data;

[0034] The medical inquiry data repairing unit is used to call the medical inquiry data quality inspection model based on the medical inquiry data and the error information of the medical inquiry data to generate repaired medical inquiry data, where the repaired medical inquiry data is the medical inquiry data after quality inspection.

[0035] In a possible implementation, the device further includes:

[0036] An instruction generation unit is used to generate a medical inquiry data error location instruction format template and a medical inquiry data repair instruction format template based on an instruction generation model before calling a medical inquiry data quality inspection model based on the medical inquiry data to generate error information of the medical inquiry data. The instruction generation model is a large language model obtained through training and has the ability to generate the medical inquiry data error location instruction format template and the medical inquiry data repair instruction format template.

[0037] In a possible implementation, the instruction generation unit is specifically configured to:

[0038] Obtain a preconfigured automatic generation instruction template for medical inquiry data quality inspection instructions, wherein the automatic generation instruction template for medical inquiry data quality inspection instructions includes a goal of the automatic generation task for medical inquiry data quality inspection instructions, a slot for filling description information of the medical inquiry data quality inspection task, a slot for filling examples of the medical inquiry data quality inspection task, and a slot for filling annotation specifications and regulations;

[0039] Fill the description information of the medical inquiry data quality inspection task into the description information filling slot of the medical inquiry data quality inspection task, fill the medical inquiry data quality inspection task example into the medical inquiry data quality inspection task example filling slot, and fill the annotation specification regulations into the annotation specification regulations filling slot to obtain the medical inquiry data quality inspection instruction automatic generation instruction;

[0040] The medical inquiry data quality inspection instruction automatic generation instruction is input into the instruction generation model to obtain the medical inquiry data error location instruction format template and the medical inquiry data repair instruction format template generated by the instruction generation model.

[0041] In a possible implementation, the medical inquiry data error locating unit is specifically configured to:

[0042] Obtaining a medical inquiry data error location instruction format template, wherein the medical inquiry data error location instruction format template includes medical inquiry data error location requirement information and a medical inquiry data filling slot;

[0043] Filling the medical inquiry data into the medical inquiry data filling slot to obtain a medical inquiry data error locating instruction;

[0044] The medical inquiry data error location instruction is input into the medical inquiry data quality inspection model to obtain error information of the medical inquiry data generated by the medical inquiry data quality inspection model.

[0045] In a possible implementation, the medical consultation data repair unit is specifically used to:

[0046] Obtaining a consultation data repair instruction format template, wherein the consultation data repair instruction format template includes consultation data repair requirement information, consultation data filling slots, and error information filling slots;

[0047] Filling the medical inquiry data into the medical inquiry data filling slot, filling the error information of the medical inquiry data into the error information filling slot, and obtaining a medical inquiry data repair instruction;

[0048] The medical inquiry data repair instruction is input into the medical inquiry data quality inspection model to obtain the repaired medical inquiry data generated by the medical inquiry data quality inspection model.

[0049] In a possible implementation, the device further includes:

[0050] a determination unit, configured to, after calling the medical inquiry data quality inspection model based on the medical inquiry data to generate error information of the medical inquiry data, and, based on the medical inquiry data and the error information of the medical inquiry data, calling the medical inquiry data quality inspection model to generate repaired medical inquiry data, determine the accuracy of error location of the medical inquiry data based on the error information of the medical inquiry data, and determine the qualified rate of repair of the medical inquiry data based on the repaired medical inquiry data;

[0051] The iterative optimization unit is used to iteratively optimize the instruction generation model if the error location accuracy of the inquiry data does not meet the preset conditions and / or the qualified rate of repair of the inquiry data does not meet the preset conditions.

[0052] In a possible implementation, the iterative optimization unit is specifically used to:

[0053] Integrating the medical inquiry data error location instruction format template to obtain an integrated medical inquiry data error location instruction format template;

[0054] Integrating the medical inquiry data repair instruction format template based on the repaired medical inquiry data to obtain an integrated medical inquiry data repair instruction format template;

[0055] The instruction generation model is iteratively optimized based on the integrated medical inquiry data error location instruction format template and the integrated medical inquiry data repair instruction format template.

[0056] The third aspect of the present application provides a computer program product, including computer-readable instructions. When the computer-readable instructions are executed on an electronic device, the electronic device implements the medical inquiry data quality inspection method of the first aspect or any implementation of the first aspect.

[0057] A fourth aspect of the present application provides an electronic device, comprising at least one processor and a memory connected to the processor, wherein:

[0058] The memory is used to store computer programs;

[0059] The processor is used to execute the computer program so that the electronic device can implement the medical inquiry data quality inspection method of the first aspect or any implementation manner of the first aspect.

[0060] In a fifth aspect, the present application provides a computer-readable storage medium, which carries one or more computer programs. When the one or more computer programs are executed by an electronic device, the electronic device can implement the medical inquiry data quality inspection method of the above-mentioned first aspect or any implementation of the first aspect.

[0061] By means of the above technical solution, the medical inquiry data quality inspection method and related devices provided by the present application include: a large language model with the ability to perform quality inspection on medical inquiry data is obtained in advance through training as a medical inquiry data quality inspection model, then after obtaining the medical inquiry data to be quality inspected, the medical inquiry data quality inspection model is first called to generate error information of the medical inquiry data, and then the medical inquiry data quality inspection model is called to repair the medical inquiry data based on the error information of the medical inquiry data, thereby realizing quality inspection of the medical inquiry data and generating medical inquiry data after quality inspection. The present application realizes intelligent, efficient and reliable quality inspection of medical inquiry data based on the powerful natural language processing capability of the large language model. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] The above and other features, advantages and aspects of the embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the accompanying drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and the originals and elements are not necessarily drawn to scale.

[0063] Figure 1A flowchart of a method for quality inspection of medical consultation data provided in an embodiment of the present application;

[0064] Figure 2 A flowchart of a method for generating a diagnosis data error location instruction format template and a diagnosis data repair instruction format template based on an instruction generation model disclosed in an embodiment of the present application;

[0065] Figure 3 A schematic diagram of the structure of a medical inquiry data quality inspection device provided in an embodiment of the present application;

[0066] Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0067] The following describes the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. The terms used in the implementation method section of the present application are only used to explain the specific embodiments of the present application, and are not intended to limit the present application.

[0068] The embodiments of the present application are described below in conjunction with the accompanying drawings. Those skilled in the art will appreciate that, with the development of technology and the emergence of new scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0069] The terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and need not be used to describe a specific order or sequential order. It should be understood that the terms used in this way can be interchangeable under appropriate circumstances, which is only to describe the distinction mode adopted by the objects of the same attributes when describing in the embodiments of the present application. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, so that the process, method, system, product or equipment comprising a series of units need not be limited to those units, but may include other units that are not clearly listed or inherent to these processes, methods, products or equipment.

[0070] As a powerful natural language processing technology, large language models are being widely used in the medical field, mainly in the areas of auxiliary diagnosis, auxiliary treatment, and auxiliary medical research. In order for large language models to better provide intelligent and personalized medical services to doctors and patients, they often need to be pre-trained or supervised fine-tuned using consultation data. However, there are often various problems in consultation data (such as wrong consultation direction, redundant consultation, confusing consultation logic, etc.), which poses a hidden danger to the pre-training or supervised fine-tuning of large language models, and may not be able to improve the natural language processing capabilities of large language models in the medical field. Therefore, it is particularly necessary to conduct quality inspection on consultation data to improve the quality of consultation data.

[0071] The current quality inspection method for medical consultation data is mainly carried out manually, that is, medical personnel manually inspect the medical consultation data. However, there is a large amount of medical consultation data, and the efficiency of manual quality inspection of medical consultation data is low, which undoubtedly brings a huge burden to the work of medical personnel. Moreover, due to the limitation of the accumulation of professional knowledge of medical personnel themselves, the reliability of manual quality inspection of medical consultation data cannot be guaranteed.

[0072] In order to solve the above problems, the embodiment of the present application provides a method for quality inspection of medical inquiry data. The method for quality inspection of medical inquiry data of the embodiment of the present application is described in detail below with reference to the accompanying drawings.

[0073] Reference Figure 1 , Figure 1 A flowchart of a method for quality inspection of medical consultation data provided in an embodiment of the present application is shown in FIG. Figure 1 As shown, a method for quality inspection of medical inquiry data provided in an embodiment of the present application may include steps S101 to S103, and these steps are described in detail below.

[0074] S101: Obtaining the medical consultation data to be inspected;

[0075] In this application, the medical inquiry data to be quality-checked can be actively initiated by the patient online and generated by human-computer interaction, or can be collected by a voice collection device when the patient has multiple rounds of interaction with the doctor offline. This application does not impose any restrictions on this. In one possible implementation, the medical inquiry data includes multiple rounds of medical inquiry dialogues, each round of medical inquiry dialogue contains a question-answer pair, which includes a question (or answer) from the patient and a reply (or inquiry) from the doctor.

[0076] S102: Based on the medical inquiry data, calling the medical inquiry data quality inspection model to generate error information of the medical inquiry data; the medical inquiry data quality inspection model is a large language model obtained through training and capable of performing quality inspection on the medical inquiry data, and the error information of the medical inquiry data is used to indicate the quality inspection results of each round of medical inquiry dialogue in the medical inquiry data;

[0077] In the present application, as a possible implementation method, the ability of the medical inquiry data quality inspection model to perform quality inspection on the medical inquiry data includes the ability to locate errors in the medical inquiry data, generate error information for the medical inquiry data, and the ability to repair the medical inquiry data using the error information of the medical inquiry data.

[0078] In the present application, it is necessary to pre-train the large language model so that it has the ability to locate errors in the medical inquiry data, generate error information for the medical inquiry data, and repair the medical inquiry data using the error information of the medical inquiry data. Specifically, medical inquiry data quality inspection model training data can be pre-constructed, and the large language model can be trained based on the medical inquiry data quality inspection model training data.

[0079] The medical consultation data quality inspection model training data may include medical consultation data samples, medical consultation data error information corresponding to the medical consultation data samples, and medical consultation data error information corresponding to the medical consultation data samples. The training method may refer to the current traditional large language model training method, which will not be described in detail here.

[0080] As mentioned in the above content, in this application, the medical consultation data includes multiple rounds of medical consultation dialogues, each round of medical consultation dialogue includes a question-answer pair, and the question-answer pair includes a question (or answer) from the patient and a reply (or consultation) from the doctor. In this application, the error information of the medical consultation data is used to indicate the quality inspection results of each round of medical consultation dialogues in the medical consultation data.

[0081] In a possible implementation, the quality inspection result of each round of consultation dialogue may include the error type, error location, specification reference, and error cause.

[0082] Among them, the error type is cited from the annotation standard regulations pre-defined by medical experts, and the error location indicates the error fragment existing in the current round of consultation dialogue (the error fragment can refer to either a word or a sentence, generally a continuous text that does not cross sentences in the current round of consultation dialogue). The standard reference shows that the large language model has a basis for locating errors in the current round of consultation dialogue, which increases the credibility of error location. Based on the standard reference, the cause of the error gives guidelines or guidance suggestions for improving consultation instructions, laying the foundation for reference of error information for subsequent consultation repairs.

[0083] For ease of understanding, this application provides an example of error information of medical consultation data, as follows:

[0084] “Round 1: Contains 1 patient’s question (or answer), 1 doctor’s response (or consultation)

[0085] Quality inspection results:

[0086] # Error type: Error type in [Marking Standards]

[0087] # Error location: The original text segment where an error occurred during the consultation process

[0088] # Standard reference: trigger and directly reference [marking standard regulations] specific regulations

[0089] # Cause of error: Provide guidelines or guidance to improve the quality of consultation

[0090] # Error type: Error type in [Marking Standards]

[0091] # Error location: The original text segment where an error occurred during the consultation process

[0092] # Standard reference: trigger and directly reference [marking standard regulations] specific regulations

[0093] # Cause of error: Provide guidelines or guidance to improve the quality of consultation

[0094] Round 2: Contains 1 patient's question (or answer), 1 doctor's response (or consultation)

[0095] Quality inspection results:

[0096] # Error type: Error type in [Marking Standards]

[0097] # Error location: The original text segment where an error occurred during the consultation process

[0098] # Standard reference: trigger and directly reference [marking standard regulations] specific regulations

[0099] # Cause of error: Provide guidelines or guidance to improve the quality of consultation

[0100] Round 3: Contains 1 question (or answer) from the patient and 1 response (or consultation) from the doctor

[0101] Quality inspection results:

[0102] # Error type: Error type in [Marking Standards]

[0103] # Error location: The original text segment where an error occurred during the consultation process

[0104] # Standard reference: trigger and directly reference [marking standard regulations] specific regulations

[0105] # Error reason: Provide guidelines or guidance to improve the quality of consultation"

[0106] S103: Based on the medical inquiry data and the error information of the medical inquiry data, the medical inquiry data quality inspection model is called to generate repaired medical inquiry data, where the repaired medical inquiry data is the medical inquiry data after quality inspection.

[0107] It should be noted that the repaired medical consultation data remains unchanged in terms of the order of questions and answers, the order of arrangement of doctor and patient roles, and the structure compared to the medical consultation data to be quality inspected.

[0108] The medical inquiry data quality inspection method provided in this embodiment includes: obtaining a large language model capable of quality inspection of medical inquiry data through training in advance as a medical inquiry data quality inspection model, then after obtaining the medical inquiry data to be quality inspected, first calling the medical inquiry data quality inspection model to generate error information of the medical inquiry data, then calling the medical inquiry data quality inspection model to repair the medical inquiry data based on the error information of the medical inquiry data, thereby realizing quality inspection of the medical inquiry data and generating medical inquiry data after quality inspection. This embodiment realizes intelligent, efficient and reliable quality inspection of medical inquiry data based on the powerful natural language processing capability of the large language model.

[0109] The interaction between the user and the large language model is based on instructions (Prompt), and the design of the instructions directly affects the instructions output by the large language model. A good instruction can guide the large language model to generate more accurate and expected results, while inappropriate instructions may cause the large language model to output biased or misleading content. In this application, the prompt engineering staff can manually design and optimize the query data error location instruction format template and the query data repair instruction format template to implement the above steps S102 and S103.

[0110] Taking into account the complexity of the task of quality inspection of medical inquiry data, manually designing and optimizing the medical inquiry data error location instruction format template and the medical inquiry data repair instruction format template is time-consuming and requires high professional knowledge. Therefore, in the present application, as a possible implementation method, the large language model can be trained in advance to enable it to have the ability to generate the medical inquiry data error location instruction format template and the medical inquiry data repair instruction format template, and use it as an instruction generation model. Then, before calling the medical inquiry data quality inspection model based on the medical inquiry data to generate error information of the medical inquiry data, the medical inquiry data error location instruction format template and the medical inquiry data repair instruction format template can be generated based on the instruction generation model.

[0111] In another embodiment of the present application, a specific implementation method for generating a diagnosis data error location instruction format template and a diagnosis data repair instruction format template based on an instruction generation model is described.

[0112] Reference Figure 2 , Figure 2 The present invention discloses a method for generating a diagnosis data error location instruction format template and a diagnosis data repair instruction format template based on an instruction generation model. Figure 2 As shown, the method may include the following steps:

[0113] S201: Obtain a pre-configured instruction template for automatically generating a quality inspection instruction for medical inquiry data, wherein the instruction template for automatically generating a quality inspection instruction for medical inquiry data includes a target for automatically generating a quality inspection instruction for medical inquiry data, a slot for filling description information of the quality inspection task for medical inquiry data, a slot for filling examples of the quality inspection task for medical inquiry data, and a slot for filling annotation specifications and rules;

[0114] Among them, the goal of the automatic generation of medical inquiry data quality inspection instructions is used to describe the specific requirements and background information of the automatic generation of medical inquiry data quality inspection instructions, which can help the large language model understand the intention and expected output type of the automatic generation of medical inquiry data quality inspection instructions. The description information filling slot of the medical inquiry data quality inspection task is used to fill in the description information of the medical inquiry data quality inspection task; the medical inquiry data quality inspection task example filling slot is used to fill in the medical inquiry data quality inspection task example, and the annotation specification filling slot is used to fill in the annotation specification.

[0115] It should be noted that the description information of the medical inquiry data quality inspection task is used to describe the medical inquiry data quality inspection task, and may include the execution characteristics of the medical inquiry data quality inspection task, the execution constraint information of the medical inquiry data quality inspection task, the output result constraint information of the medical inquiry data quality inspection task, etc.

[0116] The medical inquiry data quality inspection task examples refer to the medical inquiry data examples before and after manual quality inspection processing; by providing medical inquiry data quality inspection task examples in the automatic generation instructions of medical inquiry data quality inspection instructions, the large language model can be activated to understand the medical inquiry data quality inspection tasks, and then design and output medical inquiry data error location instructions and medical inquiry data repair instructions by itself.

[0117] Labeling specifications refer to reference documents used when manually performing quality inspection of medical consultation data, and can be pre-defined by medical experts.

[0118] For ease of understanding, an example of an automatic generation instruction template for a medical inquiry data quality inspection instruction is provided in the present application embodiment, as follows:

[0119] "Assuming that you are a Prompt engineer, please generate and output the instructions for locating errors in the inquiry data and repairing the inquiry data according to the [Description of the inquiry data quality inspection task] and [Example of the inquiry data quality inspection task] and the [Annotation specification regulations].

[0120] The description information of the medical consultation data quality inspection task is: [xxx];

[0121] An example of a medical consultation data quality inspection task is: [xxx];

[0122] The marking regulations are: [xxx]. ”

[0123] Among them, "Assuming that you are a Prompt engineer, please generate and output the inquiry data error location instructions and inquiry data repair instructions according to [the description of the inquiry data quality inspection task] and [the example of the inquiry data quality inspection task] and refer to [the annotation specifications]." is the goal of the automatic generation of inquiry data quality inspection instructions. "The description of the inquiry data quality inspection task is: [xxx]" fills the slot for the description of the inquiry data quality inspection task, "the example of the inquiry data quality inspection task is: [xxx]" fills the slot for the example of the inquiry data quality inspection task, and "the annotation specifications are: [xxx]" fills the slot for the annotation specifications.

[0124] For ease of understanding, this application provides an example of description information of a medical inquiry data quality inspection task, as follows:

[0125] "[Task Description]

[0126] # 1. Features of the simulated doctor quality inspection and consultation dialogue flow

[0127] 1.1 Quality inspection dialogue has a time sequence, from the beginning to the end of the dialogue.

[0128] 1.2 The quality inspection dialogue is single, that is, it will only be conducted in rounds of questions and answers.

[0129] 1.3 The quality inspection dialogue is one-way. During the current round of quality inspection dialogue, only the previous round of dialogue will be referred to, and subsequent rounds of consultation will not be referred to.

[0130] # 2. Keep the evaluation dimensions, evaluation standards, standard explanations, and specific explanation items in the [Labeling Standards Regulations]

[0131] 2.1 Evaluation dimensions are: not excellent, excellent, outrageous

[0132] 2.2 Evaluation criteria are detailed classifications of evaluation dimensions

[0133] 2.2.1 A single round of conversation means that each round of conversation includes one patient consultation or response and one doctor response or consultation.

[0134] 2.2.2 Error types include: redundant inquiry, missing inquiry, wrong inquiry direction, confusing inquiry logic, wrong conclusion paradigm, lack of practicality, insufficient detail, redundant conclusion, confusing logic, wrong diagnosis recommendation, too many diagnosis recommendations, too few diagnosis recommendations, unreasonable diagnosis recommendations, wrong department recommendation, wrong examination and test recommendation, lack of necessary guarantee, large medical risk, incorrect medical knowledge, violation of ethical laws and morals, lack of humanistic care, incorrect expression, outrageous inquiry, outrageous diagnosis, outrageous suggestion, and other outrageous

[0135] 2.3 Standard elaboration is a detailed explanation of the evaluation criteria

[0136] 2.4 Specific interpretation items refer to the detailed explanation of the standard. For details, please refer to [Marking Standards Regulations]

[0137] # 3. Quality inspection result output

[0138] 3.1 If an error occurs during the consultation process, detailed information about the error needs to be output, including the following three:

[0139] 3.1.1 Error type: The error type needs to be identified. The error type directly refers to the predefined types in the evaluation criteria.

[0140] 3.1.2 Error location: The original text segment of the error is required. The segment can be a medical term or a sentence.

[0141] 3.1.3 Cause of error: To give the basis for the error, you can directly quote the marking regulations.

[0142] 3.2 For the same multiple rounds of consultation, there may be many types of errors, all of which need to be located

[0143] 3.3 Output format example"

[0144] S202: Fill the description information of the medical inquiry data quality inspection task into the description information filling slot of the medical inquiry data quality inspection task, fill the medical inquiry data quality inspection task example into the medical inquiry data quality inspection task example filling slot, and fill the annotation specification regulations into the annotation specification regulations filling slot to obtain the medical inquiry data quality inspection instruction automatic generation instruction;

[0145] S203: Input the medical inquiry data quality inspection instruction automatic generation instruction into the instruction generation model to obtain the medical inquiry data error location instruction format template and the medical inquiry data repair instruction format template generated by the instruction generation model.

[0146] In the present application, the instruction generation model can be a trained initial instruction generation model, or a instruction generation model optimized by the latest iteration. After the automatic instruction generation instructions between the medical consultation data are input into the instruction generation model, the instruction generation model can output the medical consultation data error location instruction format template and the medical consultation data repair instruction format template.

[0147] In one possible implementation, the instruction format template for locating the error in the inquiry data includes the requirement information for locating the error in the inquiry data and the filling slot for the inquiry data; wherein the requirement information for locating the error in the inquiry data may include task objectives, quality inspection guidelines, labeling specifications and regulations, output format and quality inspection process.

[0148] For ease of understanding, the present application provides an example of a format template for a medical inquiry data error location instruction generated by an instruction generation model, as follows:

[0149] "[Mission Objective]

[0150] Simulate the role of a doctor, do not violate the [Quality Inspection Guidelines], strictly follow the [Quality Inspection Process], and refer to the [Quality Inspection Examples] to review and evaluate a series of multi-round medical dialogue consultations [xxx] in a sequential dialogue flow, conduct quality inspection on each round of questions and answers, and comprehensively evaluate the entire consultation process based on the development trend of the dialogue, and identify possible errors or defects from multiple dimensions.

[0151] [Quality Inspection Guidelines]

[0152] 1. Based on the given user description and consultation dialogue flow, focus on evaluating whether the doctor’s (machine agent’s) answer complies with the following guidelines:

[0153] When starting a conversation review, pay attention to the initial interaction between the patient and the doctor to confirm whether the doctor has fully asked the patient's basic information and main symptoms;

[0154] Each subsequent round of review must be based on the previous conversation content to ensure the coherence, logic, and clinical rationality of the entire consultation process;

[0155] The breadth and depth of the interview, including whether all necessary questions are covered;

[0156] The relevance of the answer to the question itself;

[0157] Identification of potentially serious illnesses and prompt recommendations for medical attention;

[0158] Humanistic care and emotional understanding;

[0159] The user description and consultation dialogue flow are sequential;

[0160] Quality inspection requires full-process dialogue and analysis of errors in context.

[0161] A single-round conversation means that each round of conversation includes one patient consultation or response and one doctor response or consultation.

[0162] The error types include: redundant inquiry, missing inquiry, wrong inquiry direction, confusing inquiry logic, wrong conclusion paradigm, lack of practicality of conclusion, insufficient detail of conclusion, redundant conclusion, confusing logic of conclusion, wrong diagnosis recommendation, too many diagnosis recommendations, too few diagnosis recommendations, unreasonable diagnosis recommendations, wrong department recommendation, wrong examination and test recommendation, lack of necessary guarantee, large medical risk, wrong medical knowledge, violation of ethics, law and morality, lack of humanistic care, incorrect expression, outrageous inquiry, outrageous diagnosis, outrageous suggestion, and other outrageous

[0163] 2. For any errors found, carefully record the error type, error location, and error explanation, and ensure that:

[0164] The error type is consistent with the type defined in [Marking Specifications];

[0165] The error is accurately located, pointing to a specific inappropriate statement or omission in the dialogue;

[0166] The error explanation is clear and rigorous, and directly refers to the instructions in the [Marking Specifications Regulations].

[0167] [Marking Standards Regulations]

[0168] [Output format]

[0169] [Quality Inspection Process]

[0170] 1. Review each round of consultation dialogue one by one and evaluate each answer of the doctor (machine agent) one by one.

[0171] 2. For each doctor’s response, use the classification of [labeling regulations] to give the specific error type.

[0172] 3. If multiple error types occur, they need to be marked and explained separately.

[0173] 4. For situations that are still unclear or require the user to provide additional information, record the context and suggest possible follow-up inquiry directions.

[0174] 5. Finally, provide overall recommendations and give guidelines or guidance to improve the quality of consultation.

[0175] 6. Give the quality inspection results according to [output format]. ”

[0176] Among them, "a series of multi-round medical dialogue consultation [xxx]" is the consultation data filling slot.

[0177] In one possible implementation, the consultation data repair instruction format template includes consultation data repair requirement information, consultation data filling slots, and error information filling slots, wherein the consultation data repair requirement information may include task objectives, repair methodology, result output format, correct repair examples, and repair processes.

[0178] For ease of understanding, the present application provides an example of a consultation data repair instruction format template generated by an instruction generation model, as follows:

[0179] "[Mission Objective]

[0180] 1. Given the original multi-round consultation [xxx] and error improvement suggestions [xxx], follow the repair methodology, strictly follow the repair process to carefully check each round of dialogue in the consultation dialogue, and complete the error repair of the consultation dialogue.

[0181] [Restoration Methodology]

[0182] First, in order to comply with the instruction output and constraints, we need to do the following steps:

[0183] 1. Analyze the current [original multi-round consultation] dialogue flow, timing information and format characteristics:

[0184] 1.1 Dialogue flow: The core is to analyze each round of interaction between patients and doctors and understand how doctors guide the conversation to obtain information.

[0185] 1.2 Sequence information: Confirm the symptom information obtained in each round of communication, ensure that the logical relationship between them is clear and orderly, and the previous and subsequent contents are consistent.

[0186] 1.3 Format features: Observe the length, structure, and tone of the conversation, such as the number of words per question and the level of detail in the answers.

[0187] 2. Next, use [Error Improvement Suggestions]:

[0188] 2.1 Parse the JSON list and clarify the four fields of each element: error type, error location, regulation reference, and suggested inquiry.

[0189] 2.2 For each error, understand its possible impact on the consultation process and determine the necessary corrective actions.

[0190] 3. Next, refer to the following steps for specific execution:

[0191] 3.1 Read and analyze the current consultation flow.

[0192] 3.2 Use the information in [Error Improvement Suggestions] to locate the specific point where the error occurred.

[0193] 3.3 According to the error type, refer to the regulations and conduct correctness check and correction.

[0194] 3.4 Apply the recommended inquiries, modify or add necessary inquiry steps, and ensure that all necessary health information collection is covered.

[0195] 3.5 Format the modified dialog flow to ensure that it complies with the original style and requirements.

[0196] 3.6 Review the entire revised consultation dialogue flow to ensure logical consistency in timing, completeness and accuracy of information.

[0197] [Result output format]

[0198] ```

[0199] Patient: xxx

[0200] Doctor: xxx

[0201] Patient: xxx

[0202] Doctor: xxx

[0203] ```

[0204] [Correct repair example]

[0205] [Repair process]

[0206] 1. Read the current conversation process;

[0207] 2.FOR each error in error list:

[0208] Locate the conversation turn where the error occurred;

[0209] Refer to the regulations and understand the impact of errors;

[0210] According to the suggested inquiries, the incorrect content of the doctor's consultation should be deleted or corrected, and the patient's reply should also be deleted or corrected accordingly;

[0211] Organize the revised conversations to avoid duplication of replies and ensure consistency of format;

[0212] 3.ENDFOR

[0213] 4. Repair or regenerate the conclusion according to the [Conclusion Generation Paradigm].

[0214] 5. Verify the consistency and accuracy of the information in the adjusted dialogue flow;

[0215] 6. Refer to [Result Output Format] to output the final consultation dialogue flow. The repair process does not need to be output. "

[0216] Among them, "Original multi-round consultation [xxx]" is the slot filled with consultation data, and "Error improvement suggestions [xxx]" is the slot filled with error information.

[0217] After generating the inquiry data error location instruction format template and the inquiry data repair instruction format template based on the instruction generation model, the inquiry data error location instruction format template can be used to call the inquiry data quality inspection model based on the inquiry data to generate error information of the inquiry data, and the inquiry data repair instruction format template can be used to call the inquiry data quality inspection model based on the inquiry data and the error information of the inquiry data to generate repaired inquiry data.

[0218] In a possible implementation, calling the medical inquiry data quality inspection model based on the medical inquiry data to generate error information of the medical inquiry data includes:

[0219] Obtaining a medical inquiry data error location instruction format template, wherein the medical inquiry data error location instruction format template includes medical inquiry data error location requirement information and a medical inquiry data filling slot;

[0220] Filling the medical inquiry data into the medical inquiry data filling slot to obtain a medical inquiry data error locating instruction;

[0221] The medical inquiry data error location instruction is input into the medical inquiry data quality inspection model to obtain error information of the medical inquiry data generated by the medical inquiry data quality inspection model.

[0222] In a possible implementation, calling the medical inquiry data quality inspection model based on the medical inquiry data and the error information of the medical inquiry data to generate the repaired medical inquiry data includes:

[0223] Obtaining a consultation data repair instruction format template, wherein the consultation data repair instruction format template includes consultation data repair requirement information, consultation data filling slots, and error information filling slots;

[0224] Filling the medical inquiry data into the medical inquiry data filling slot, filling the error information of the medical inquiry data into the error information filling slot, and obtaining a medical inquiry data repair instruction;

[0225] The medical inquiry data repair instruction is input into the medical inquiry data quality inspection model to obtain the repaired medical inquiry data generated by the medical inquiry data quality inspection model.

[0226] In the present application, in order to further improve the effect of the quality inspection of the medical inquiry data, the medical inquiry data quality inspection model can be called based on the medical inquiry data to generate error information of the medical inquiry data, and the medical inquiry data quality inspection model can be called based on the medical inquiry data and the error information of the medical inquiry data to generate the repaired medical inquiry data. Then, the accuracy of error location of the medical inquiry data is determined based on the error information of the medical inquiry data, and the qualified rate of medical inquiry data repair is determined based on the repaired medical inquiry data; if the accuracy of error location of the medical inquiry data does not meet the preset conditions, and / or the qualified rate of medical inquiry data repair does not meet the preset conditions, the instruction generation model is iteratively optimized.

[0227] In one possible implementation, a threshold for the accuracy of error location of medical inquiry data and a threshold for the pass rate of medical inquiry data repair can be preset. If the accuracy of error location of medical inquiry data does not reach the threshold for the accuracy of error location of medical inquiry data, it means that the accuracy of error location of medical inquiry data does not meet the preset conditions. If the pass rate of medical inquiry data repair does not reach the threshold for the pass rate of medical inquiry data repair, it means that the pass rate of medical inquiry data repair does not meet the preset conditions.

[0228] In a possible implementation, the accuracy of error location of medical consultation data can be determined by the following formula:

[0229] Accuracy = number of correctly predicted errors / total number of predicted errors × 100%

[0230] The qualified rate of consultation data repair can be determined by the following formula:

[0231] Qualified rate = qualified after consultation and repair / total number of consultation and repair × 100%

[0232] In a possible implementation, the iteratively optimizing the instruction generation model includes:

[0233] Integrating the medical inquiry data error location instruction format template to obtain an integrated medical inquiry data error location instruction format template;

[0234] Integrating the medical inquiry data repair instruction format template based on the repaired medical inquiry data to obtain an integrated medical inquiry data repair instruction format template;

[0235] The instruction generation model is iteratively optimized based on the integrated medical inquiry data error location instruction format template and the integrated medical inquiry data repair instruction format template.

[0236] It should be noted that in the present application, the medical inquiry data error location instruction format template can be integrated manually to obtain an integrated medical inquiry data error location instruction format template; and the medical inquiry data repair instruction format template can be integrated based on the repaired medical inquiry data to obtain an integrated medical inquiry data repair instruction format template.

[0237] A method for quality inspection of medical inquiry data provided in an embodiment of the present application is introduced above. A device for executing the method for quality inspection of medical inquiry data will be introduced below.

[0238] See also Figure 3 , Figure 3 This is a schematic diagram of the structure of a medical inquiry data quality inspection device provided in an embodiment of the present application. Figure 4 As shown, the medical inquiry data quality inspection device comprises:

[0239] An acquisition unit 11 is used to acquire the medical inquiry data to be inspected;

[0240] The medical inquiry data error locating unit 12 is used to call the medical inquiry data quality inspection model based on the medical inquiry data to generate error information of the medical inquiry data; the medical inquiry data quality inspection model is a large language model obtained through training and capable of performing quality inspection on the medical inquiry data, and the error information of the medical inquiry data is used to indicate the quality inspection results of each round of medical inquiry dialogue in the medical inquiry data;

[0241] The medical inquiry data repairing unit 13 is used to call the medical inquiry data quality inspection model based on the medical inquiry data and the error information of the medical inquiry data to generate repaired medical inquiry data, where the repaired medical inquiry data is the medical inquiry data after quality inspection.

[0242] In a possible implementation, the device further includes:

[0243] An instruction generation unit is used to generate a medical inquiry data error location instruction format template and a medical inquiry data repair instruction format template based on an instruction generation model before calling a medical inquiry data quality inspection model based on the medical inquiry data to generate error information of the medical inquiry data. The instruction generation model is a large language model obtained through training and has the ability to generate the medical inquiry data error location instruction format template and the medical inquiry data repair instruction format template.

[0244] In a possible implementation, the instruction generation unit is specifically configured to:

[0245] Obtain a preconfigured automatic generation instruction template for medical inquiry data quality inspection instructions, wherein the automatic generation instruction template for medical inquiry data quality inspection instructions includes a goal of the automatic generation task for medical inquiry data quality inspection instructions, a slot for filling description information of the medical inquiry data quality inspection task, a slot for filling examples of the medical inquiry data quality inspection task, and a slot for filling annotation specifications and regulations;

[0246] Fill the description information of the medical inquiry data quality inspection task into the description information filling slot of the medical inquiry data quality inspection task, fill the medical inquiry data quality inspection task example into the medical inquiry data quality inspection task example filling slot, and fill the annotation specification regulations into the annotation specification regulations filling slot to obtain the medical inquiry data quality inspection instruction automatic generation instruction;

[0247] The medical inquiry data quality inspection instruction automatic generation instruction is input into the instruction generation model to obtain the medical inquiry data error location instruction format template and the medical inquiry data repair instruction format template generated by the instruction generation model.

[0248] In a possible implementation, the medical inquiry data error locating unit is specifically configured to:

[0249] Obtaining a medical inquiry data error location instruction format template, wherein the medical inquiry data error location instruction format template includes medical inquiry data error location requirement information and a medical inquiry data filling slot;

[0250] Filling the medical inquiry data into the medical inquiry data filling slot to obtain a medical inquiry data error locating instruction;

[0251] The medical inquiry data error location instruction is input into the medical inquiry data quality inspection model to obtain error information of the medical inquiry data generated by the medical inquiry data quality inspection model.

[0252] In a possible implementation, the medical consultation data repair unit is specifically used to:

[0253] Obtaining a consultation data repair instruction format template, wherein the consultation data repair instruction format template includes consultation data repair requirement information, consultation data filling slots, and error information filling slots;

[0254] Filling the medical inquiry data into the medical inquiry data filling slot, filling the error information of the medical inquiry data into the error information filling slot, and obtaining a medical inquiry data repair instruction;

[0255] The medical inquiry data repair instruction is input into the medical inquiry data quality inspection model to obtain the repaired medical inquiry data generated by the medical inquiry data quality inspection model.

[0256] In a possible implementation, the device further includes:

[0257] a determination unit, configured to, after calling the medical inquiry data quality inspection model based on the medical inquiry data to generate error information of the medical inquiry data, and, based on the medical inquiry data and the error information of the medical inquiry data, calling the medical inquiry data quality inspection model to generate repaired medical inquiry data, determine the accuracy of error location of the medical inquiry data based on the error information of the medical inquiry data, and determine the qualified rate of repair of the medical inquiry data based on the repaired medical inquiry data;

[0258] The iterative optimization unit is used to iteratively optimize the instruction generation model if the error location accuracy of the inquiry data does not meet the preset conditions and / or the qualified rate of repair of the inquiry data does not meet the preset conditions.

[0259] In a possible implementation, the iterative optimization unit is specifically used to:

[0260] Integrating the medical inquiry data error location instruction format template to obtain an integrated medical inquiry data error location instruction format template;

[0261] Integrating the medical inquiry data repair instruction format template based on the repaired medical inquiry data to obtain an integrated medical inquiry data repair instruction format template;

[0262] The instruction generation model is iteratively optimized based on the integrated medical inquiry data error location instruction format template and the integrated medical inquiry data repair instruction format template.

[0263] The present application also provides an electronic device in an embodiment. Figure 4 As shown, it shows a schematic diagram of the structure of an electronic device suitable for implementing the embodiment of the present application. The electronic device in the embodiment of the present application may include but is not limited to fixed terminals such as mobile phones, laptops, PDAs (personal digital assistants), PADs (tablet computers), desktop computers, etc. Figure 4 The electronic device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0264] like Figure 4 As shown, the electronic device may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage device 608 to a random access memory (RAM) 603. When the electronic device is powered on, various programs and data required for the operation of the electronic device are also stored in the RAM 603. The processing device 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0265] Typically, the following devices may be connected to the I / O interface 605: an input device 606 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 607 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 608 including, for example, a memory card, a hard disk, etc.; and a communication device 609. The communication device 609 may allow the electronic device to communicate with other devices wirelessly or by wire to exchange data. Although Figure 4 An electronic device having various devices is shown, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.

[0266] An embodiment of the present application also provides a computer program product including computer-readable instructions. When the computer-readable instructions are executed on an electronic device, the electronic device implements any one of the medical inquiry data quality inspection methods provided in the embodiments of the present application.

[0267] A computer-readable storage medium is also provided in an embodiment of the present application. The storage medium carries one or more computer programs. When the one or more computer programs are executed by an electronic device, the electronic device can implement any one of the medical inquiry data quality inspection methods provided in the embodiment of the present application.

[0268] It should also be noted that the device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed over multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the drawings of the device embodiments provided by the present application, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines.

[0269] Through the description of the above implementation mode, the technicians in the field can clearly understand that the present application can be implemented by means of software plus necessary general hardware, and of course, it can also be implemented by special hardware including special integrated circuits, special CPUs, special memories, special components, etc. In general, all functions completed by computer programs can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be various, such as analog circuits, digital circuits or special circuits. However, for the present application, software program implementation is a better implementation mode in more cases. Based on such an understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a readable storage medium, such as a computer floppy disk, a U disk, a mobile hard disk, a ROM, a RAM, a disk or an optical disk, etc., including a number of instructions to enable a computer device (which can be a personal computer, a training device, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0270] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments may be implemented in the form of a computer program product.

[0271] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from a website site, a computer, a training device, or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, training device, or data center. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a training device, a data center, etc. that includes one or more available media integrations. The available medium may be a magnetic medium, (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive (SSD)), etc.

Claims

1. A method for quality inspection of medical consultation data, characterized in that: include: Obtain the consultation data to be inspected; Based on the medical inquiry data, calling a medical inquiry data quality inspection model to generate error information of the medical inquiry data; The medical inquiry data quality inspection model is a large language model obtained through training and capable of performing quality inspection on medical inquiry data, and the error information of the medical inquiry data is used to indicate the quality inspection results of each round of medical inquiry dialogue in the medical inquiry data; Based on the medical inquiry data and the error information of the medical inquiry data, the medical inquiry data quality inspection model is called to generate repaired medical inquiry data, and the repaired medical inquiry data is the medical inquiry data after quality inspection.

2. The method according to claim 1, characterized in that Before calling the medical inquiry data quality inspection model based on the medical inquiry data to generate error information of the medical inquiry data, the method further includes: A medical inquiry data error location instruction format template and a medical inquiry data repair instruction format template are generated based on an instruction generation model, wherein the instruction generation model is a large language model obtained through training and has the ability to generate the medical inquiry data error location instruction format template and the medical inquiry data repair instruction format template.

3. The method according to claim 2, characterized in that The method of generating a diagnosis data error location instruction format template and a diagnosis data repair instruction format template based on an instruction generation model includes: Obtain a preconfigured automatic generation instruction template for medical inquiry data quality inspection instructions, wherein the automatic generation instruction template for medical inquiry data quality inspection instructions includes a goal of the automatic generation task for medical inquiry data quality inspection instructions, a slot for filling description information of the medical inquiry data quality inspection task, a slot for filling examples of the medical inquiry data quality inspection task, and a slot for filling annotation specifications and regulations; Fill the description information of the medical inquiry data quality inspection task into the description information filling slot of the medical inquiry data quality inspection task, fill the medical inquiry data quality inspection task example into the medical inquiry data quality inspection task example filling slot, and fill the annotation specification regulations into the annotation specification regulations filling slot to obtain the medical inquiry data quality inspection instruction automatic generation instruction; The medical inquiry data quality inspection instruction automatic generation instruction is input into the instruction generation model to obtain the medical inquiry data error location instruction format template and the medical inquiry data repair instruction format template generated by the instruction generation model.

4. The method according to claim 2, characterized in that: The calling the medical inquiry data quality inspection model based on the medical inquiry data to generate error information of the medical inquiry data includes: Obtaining a medical inquiry data error location instruction format template, wherein the medical inquiry data error location instruction format template includes medical inquiry data error location requirement information and a medical inquiry data filling slot; Filling the medical inquiry data into the medical inquiry data filling slot to obtain a medical inquiry data error locating instruction; The medical inquiry data error location instruction is input into the medical inquiry data quality inspection model to obtain error information of the medical inquiry data generated by the medical inquiry data quality inspection model.

5. The method according to claim 2, characterized in that: The calling the medical inquiry data quality inspection model based on the medical inquiry data and the error information of the medical inquiry data to generate the repaired medical inquiry data includes: Obtaining a consultation data repair instruction format template, wherein the consultation data repair instruction format template includes consultation data repair requirement information, consultation data filling slots, and error information filling slots; Filling the medical inquiry data into the medical inquiry data filling slot, filling the error information of the medical inquiry data into the error information filling slot, and obtaining a medical inquiry data repair instruction; The medical inquiry data repair instruction is input into the medical inquiry data quality inspection model to obtain the repaired medical inquiry data generated by the medical inquiry data quality inspection model.

6. The method according to claim 2, characterized in that After calling the medical inquiry data quality inspection model based on the medical inquiry data to generate error information of the medical inquiry data, and calling the medical inquiry data quality inspection model based on the medical inquiry data and the error information of the medical inquiry data to generate repaired medical inquiry data, the method further includes: Determining the accuracy of locating errors in the medical inquiry data based on the error information of the medical inquiry data, and determining the qualified rate of repairing the medical inquiry data based on the repaired medical inquiry data; If the error location accuracy of the inquiry data does not meet the preset conditions, and / or the repair qualification rate of the inquiry data does not meet the preset conditions, the instruction generation model is iteratively optimized.

7. The method according to claim 6, characterized in that The iterative optimization of the instruction generation model includes: Integrating the medical inquiry data error location instruction format template to obtain an integrated medical inquiry data error location instruction format template; Integrating the medical inquiry data repair instruction format template based on the repaired medical inquiry data to obtain an integrated medical inquiry data repair instruction format template; The instruction generation model is iteratively optimized based on the integrated medical inquiry data error location instruction format template and the integrated medical inquiry data repair instruction format template.

8. A medical inquiry data quality inspection device, characterized in that: include: An acquisition unit, used for acquiring the medical consultation data to be inspected; A medical inquiry data error locating unit, configured to call a medical inquiry data quality inspection model based on the medical inquiry data to generate error information of the medical inquiry data; the medical inquiry data quality inspection model is a large language model obtained through training and capable of performing quality inspection on medical inquiry data, and the error information of the medical inquiry data is used to indicate the quality inspection results of each round of medical inquiry dialogue in the medical inquiry data; The medical inquiry data repairing unit is used to call the medical inquiry data quality inspection model based on the medical inquiry data and the error information of the medical inquiry data to generate repaired medical inquiry data, where the repaired medical inquiry data is the medical inquiry data after quality inspection.

9. A computer program product, characterized in that It includes computer-readable instructions, and when the computer-readable instructions are executed on an electronic device, the electronic device implements the medical inquiry data quality inspection method as described in any one of claims 1 to 7.

10. An electronic device, characterized in that: The method comprises at least one processor and a memory connected to the processor, wherein: The memory is used to store computer programs; The processor is used to execute the computer program so that the electronic device can implement the medical inquiry data quality inspection method as described in any one of claims 1 to 7.

11. A computer-readable storage medium, characterized in that: The storage medium carries one or more computer programs, and when the one or more computer programs are executed by an electronic device, the electronic device can implement the medical inquiry data quality inspection method as described in any one of claims 1 to 7.

Images