Quality detection method, system and equipment for special disease database and storage medium

Abstract

The invention relates to a quality detection method, system and device for a special disease database and a storage medium, and belongs to the technical field of data quality detection.The method comprises the steps that the field type of the special disease database to be detected is obtained; constructing a plurality of questions used for obtaining field results, and constructing the questions into a prompt question bank; a field result in the special disease database is read, an original medical text of the field result is obtained, and part of the original medical text serves as training text data; obtaining a field result of the training text data to obtain a target answer; inputting the training text data, the target answer and the question into a quality detection model to obtain a trained quality detection model; and inputting original medical text data and questions into the trained quality detection model, outputting a field result by the quality detection model, and judging the consistency of the field result and the field result in the special disease database. And quality inspection is performed on the special disease database through the quality detection model, so that the manpower consumption is reduced.

Description

Background Art

[0002] At present, a large amount of medical records are stored in the database of the hospital system. Doctors can promote the improvement of medical level by studying all medical records under a certain disease. With the continuous development of hospital informatization, in order to provide an authoritative and accurate source of information for medical professionals, researchers, patients and other related groups, so as to better understand the relevant knowledge and information of specific diseases, a disease-specific database has been established.

[0003] A disease-specific database refers to a medical database or knowledge base established for a specific disease or disease type. It usually contains professional knowledge and information on the disease's clinical characteristics, diagnostic criteria, treatment methods, drug information, prognosis evaluation, etc. It can help doctors diagnose diseases and make treatment decisions, provide researchers with disease data and research progress, and help patients understand and manage their own diseases.

[0004] In the context of the rapid development of medical information technology, disease-specific databases serve as an important support for clinical research and diagnosis and treatment. The data quality is directly related to the accuracy of scientific research results and the reliability of clinical decision-making. At present, the quality detection method of disease-specific databases is generally manual sampling, which requires a lot of manpower and material resources.

[0005] To this end, the present invention provides a quality detection method, system, device and storage medium for a disease-specific database to solve the above-mentioned technical problems. Summary of the invention

[0006] The object of the present invention is to provide a quality detection method, system, device and storage medium for a disease-specific database, so as to solve any of the technical problems mentioned above.

[0007] In order to achieve the above object, the present invention provides the following technical solutions: In a first aspect, a quality detection method for a disease-specific database is provided, wherein the disease-specific database stores field results of several field types, and the method comprises: Get the field type of the disease database to be tested; Construct several questions to obtain field results under a certain field type; Manage several questions that obtain field results under this field type in a unified manner and build them into a prompt question bank; Read the field results under the field type in the disease-specific database, obtain the original medical text corresponding to the field results, verify the accuracy of the field results, and select the original text data corresponding to some of the verified field results as training text data; Read the training text data, and obtain the field results corresponding to the constructed question, to obtain the target answer; Input the training text data, the target answer, and the questions in the prompt question bank that obtain the field results into the pre-built quality detection model, train the quality detection model, and obtain a trained quality detection model; The original medical text data and the questions about the field results under the field type are input into the trained quality detection model. The quality detection model outputs the field results, and the consistency of the field results with the field results in the special disease database is judged to determine the accuracy of the field results in the special disease database, thereby detecting the quality of the special disease database.

[0008] Furthermore, several questions are constructed to obtain the field results under a certain field type, where: For each field type in the disease-specific database, construct at least three questions to obtain field results.

[0009] Further, the field result under the field type in the disease-specific database is read, and the original medical text corresponding to the field result is obtained, including the following specific steps: Use the database connection library or API to establish a connection with the disease-specific database, and execute query statements to obtain field results in the disease-specific database; Through the association relationship or index in the disease-specific database, the original medical text corresponding to the field result is found, and the obtained original medical text is output.

[0010] Furthermore, the pre-constructed quality detection model is the Qwen1.5-14b model.

[0011] Furthermore, after inputting the training text data, the target answer and the questions in the prompt question bank for obtaining the field results into the pre-built quality detection model, training the quality detection model and obtaining the trained quality detection model, the following specific steps are also included: Input the training text data and the questions in the prompt question bank that get the field results into the trained Qwen1.5-14b model, and the trained Qwen1.5-14b model outputs the predicted field results; Through the loss function, the target answer and the predicted field result are calculated to obtain the calculation result; The Qwen1.5-14b model is optimized based on the calculation results using the Adam optimizer; Repeat the above steps until the loss function converges, the training of the Qwen1.5-14b model is completed, and the optimized Qwen1.5-14b model is obtained.

[0012] Furthermore, before the original medical text data and the questions in the prompt question bank for obtaining the field results under the field type are input into the trained quality inspection model, and the trained quality inspection model outputs the predicted field results, the method further includes: Set the training parameters of the Qwen1.5-14b model, including batch_size, epoch, and gradient accumulation.

[0013] Furthermore, the loss function is the cross entropy loss function, and the expression of the cross entropy loss function is:

[0014] In the formula, represents the loss function, Indicates the number of categories, represents the true value in the labeled data, Represents the model prediction value.

[0015] In a second aspect, a quality detection system for a disease-specific database is provided, the system comprising: A data field acquisition unit, used to acquire the field type of the disease-specific database to be detected; A question construction unit is used to construct a number of questions for obtaining field results under a certain field type; The prompt question bank construction unit is used to uniformly manage several questions that obtain field results under this field type and construct them into a prompt question bank; A data extraction unit is used to read the field results under the field type in the disease-specific database, obtain the original medical text corresponding to the field results, verify the accuracy of the field results, and select the original text data corresponding to some of the field results that have passed the verification as training text data; A target answer acquisition unit is used to read the training text data and obtain the field results corresponding to the training text data under the constructed question to obtain the target answer; The quality detection model training unit is used to input the training text data, the target answer and the questions in the prompt question bank for obtaining the field results into the pre-built quality detection model, train the quality detection model, and obtain the trained quality detection model; The quality detection unit is used to input the original medical text data and the problem of obtaining the field results under the field type into the trained quality detection model. The quality detection model outputs the field results, and judges the consistency of the field results with the field results in the special disease database to determine the accuracy of the field results in the special disease database, thereby detecting the quality of the special disease database.

[0016] According to a third aspect, an electronic device includes: at least one processor; and a memory communicatively coupled to the at least one processor; The memory stores computer program instructions that can be executed by at least one processor, and the computer program instructions are executed by the at least one processor so that the at least one processor can execute the quality detection method of the disease-specific database described in the above aspects.

[0017] In a fourth aspect, a non-transitory computer-readable storage medium is provided, wherein the non-transitory computer-readable storage medium stores computer instructions, wherein the computer instructions enable the computer to execute the quality detection method for a disease-specific database as described in the above aspects.

[0018] Beneficial effects of the present invention: The present invention trains a quality detection model, inputs the original medical text data and the questions for obtaining field results into the trained quality detection model, and the quality detection model outputs the field results. The consistency of the field results with the field results in the special disease database is judged to determine the accuracy of the field results in the special disease database, thereby detecting the quality of the special disease database, realizing automatic detection of the quality of the special disease database, improving the efficiency of quality detection of the special disease database, and greatly reducing labor costs and time costs.

[0019] These and other objects, features and advantages of the present invention will be fully reflected in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 A schematic diagram of the method flow of the quality detection method of the disease-specific database of the present invention is shown. DETAILED DESCRIPTION

[0021] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are only examples, and those skilled in the art can think of other obvious variations. The basic principles of the present invention defined in the following description can be applied to other embodiments, variations, improvements, equivalents, and other technical solutions that do not deviate from the spirit and scope of the present invention.

[0022] It should be understood by those skilled in the art that, in the disclosure of the specification, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device including a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices. Therefore, the above terms should not be understood as limiting the present invention.

[0023] It is to be understood that the term "one" should be understood as "at least one" or "one or more", that is, in one embodiment, the number of an element may be one, while in another embodiment, the number of the element may be multiple, and the term "one" should not be understood as a limitation on the quantity.

[0024] refer to Figure 1 A quality detection method for a disease-specific database according to a preferred embodiment of the present invention will be described in detail below, wherein the disease-specific database stores field results of several field types. It can be understood that the field types may be "onset symptoms", "onset time", etc., wherein the field results corresponding to "onset symptoms" are fever, cough, limb pain, etc., and the field results corresponding to "onset time" are November 2023, June 2023, April 2023, etc. The method comprises: Get the field type of the disease database to be tested; Construct several questions to obtain field results under a certain field type; Several questions for obtaining field results under this field type are uniformly managed and constructed into a prompt question bank.

[0025] In this embodiment, the disease-specific database is a database that records the original medical text related to lung cancer. One field type corresponds to one prompt question bank, which avoids the situation where multiple field types correspond to one prompt question bank and there is no answer (field result) to the question. It is now necessary to detect the "onset time" to detect whether the field results of the "onset time" recorded in the disease-specific database are accurate. Among them, the field result of the patient's "onset time" is read in the disease-specific database, and the field result is: "November 2023". Get the original medical text corresponding to the field result. The field result read in the disease-specific database is the field result of the patient's "onset time", and the original medical text corresponding to the traversed field result is: "The patient was diagnosed with lung cancer in November 2023". Construct several questions to obtain the field results of the "onset time". In this embodiment, the questions constructed to obtain the field results of the "onset time" are: "When is the onset time of the patient's lung cancer?", "When did the patient suffer from lung cancer?", "When did the patient discover the relevant symptoms of lung cancer?". It is understandable that when constructing questions to obtain field results under a certain field type, it is necessary to ensure that the field results obtained by several questions are consistent.

[0026] Read the field results under this field type in the disease-specific database, obtain the original medical text corresponding to the field result, verify the accuracy of the field result, and select the original text data corresponding to some of the verified field results as training text data; it can be understood that the disease-specific database contains field results of many field types, and the field results in the disease-specific database are extracted from the corresponding original medical text. For example: the description in the original medical text is: "The patient was diagnosed with lung cancer in November 2023", then in the diagnosis time field in the lung cancer disease-specific database, the field result of this original medical text is: "November 2023".

[0027] At this time, the disease-specific database has not been quality-checked, and the accuracy of the field results in it cannot be guaranteed. Therefore, the field results of the disease-specific database cannot be used directly to train the model. Before training, a manual quality inspection step needs to be added to check whether the field results are accurate and ensure the accuracy of the training data.

[0028] Read the training text data, and obtain the field results corresponding to the constructed question, to get the target answer.

[0029] Input the training text data, target answers, and questions in the prompt question bank that get field results into the pre-built quality inspection model, train the quality inspection model, and obtain a trained quality inspection model. It can be understood that when training the quality inspection model, each training data contains a training text data, a question that gets a field result, and a target answer to the relevant field result. For example, the original text data is: "The patient was diagnosed with lung cancer in November 2023", and the question that gets the field result is: "When did the patient's lung cancer develop?", and the target answer is: "November 2023".

[0030] In this embodiment, when training the quality inspection model, only one large model can be trained, or three large models can be trained. If one large model is trained, the three questions for obtaining field results constructed in the prompt question bank are input into the same large model. If three large models are trained, the three questions for obtaining field results are input into three different large models respectively.

[0031] The original medical text data and the questions about the field results under the field type are input into the trained quality detection model. The quality detection model outputs the field results, and the consistency of the field results with the field results in the special disease database is judged to determine the accuracy of the field results in the special disease database, thereby detecting the quality of the special disease database. If they are consistent, it means that the field results in the special disease database are accurate. If they are inconsistent, the field results in the special disease database are recorded and wait for manual review.

[0032] Exemplarily, several questions are constructed to obtain field results under a certain field type, where: For each field type in the disease-specific database, construct at least three questions to obtain field results.

[0033] Exemplarily, reading the field result under the field type in the disease-specific database and obtaining the original medical text corresponding to the field result includes the following specific steps: Use the database connection library or API to establish a connection with the disease-specific database, and execute query statements to obtain field results in the disease-specific database; Through the association relationship or index in the disease-specific database, the original medical text corresponding to the field result is found, and the obtained original medical text is output.

[0034] In this embodiment, the pre-constructed quality detection model is the Qwen1.5-14b model.

[0035] After inputting the training text data, the target answer and the questions in the prompt question bank for obtaining the field results into the pre-built quality inspection model, training the quality inspection model and obtaining the trained quality inspection model, the following specific steps are also included: Input the training text data and the questions in the prompt question bank that get the field results into the trained Qwen1.5-14b model, and the trained Qwen1.5-14b model outputs the predicted field results; Through the loss function, the target answer and the predicted field result are calculated to obtain the calculation result; The Qwen1.5-14b model is optimized based on the calculation results using the Adam optimizer; Repeat the above steps until the loss function converges, the training of the Qwen1.5-14b model is completed, and the optimized Qwen1.5-14b model is obtained.

[0036] Before the training text data and the questions in the prompt question bank for obtaining the field results under the field type are input into the trained quality inspection model, the trained quality inspection model outputs the predicted field results, and also includes: Set the training parameters of the Qwen1.5-14b model, including batch_size, epoch, and gradient accumulation. Batch_size (batch size): The batch size determines the number of samples used for each gradient update. A smaller batch_size can adapt to changes in the distribution of training data more quickly, but it is easy to cause instability in the training process; a larger batch_size can stabilize the training process of the Qwen1.5-14b model training, but it requires more memory and longer training time. Epoch (number of training cycles): The number of times the entire training data is traversed. Too many epochs are likely to lead to overfitting, while too few epochs are likely to lead to underfitting. Gradient accumulation: When the GPU memory is insufficient to support large-batch training, the gradient accumulation technique can be used. It allows the gradient to be accumulated over multiple small batches, and then the parameters are updated uniformly, thereby simulating the effect of large-batch training. The above training parameters can be set by technicians in this field according to actual needs.

[0037] It can be understood that the Qwen1.5-14b model is loaded using the transformers library, and the training set is serialized using the Qwen1.5-14b model; the Qwen1.5-14b model is trained using the serialized training set, and the difference between the predicted field result output by the Qwen1.5-14b model and the target answer is calculated using the loss function; the training parameters of the Qwen1.5-14b model are optimized using the Adam optimizer according to the difference calculated by the loss function; until the expected output is generated according to the instructions.

[0038] In this embodiment, the loss function is a cross entropy loss function, and the expression of the cross entropy loss function is:

[0039] In the formula, represents the loss function, Indicates the number of categories, represents the true value in the labeled data, Represents the model prediction value.

[0040] The embodiment of the present invention further provides a quality detection system for a disease-specific database, the system comprising: A data field acquisition unit, used to acquire the field type of the disease-specific database to be detected; A question construction unit is used to construct a number of questions for obtaining field results under a certain field type; The prompt question bank construction unit is used to uniformly manage several questions that obtain field results under this field type and construct them into a prompt question bank; A data extraction unit is used to read the field results under the field type in the disease-specific database, obtain the original medical text corresponding to the field results, verify the accuracy of the field results, and select the original text data corresponding to some of the field results that have passed the verification as training text data; A target answer acquisition unit is used to read the training text data and obtain the field results corresponding to the training text data under the constructed question to obtain the target answer; The quality detection model training unit is used to input the training text data, the target answer and the questions in the prompt question bank for obtaining the field results into the pre-built quality detection model, train the quality detection model, and obtain the trained quality detection model; The quality detection unit is used to input the original medical text data and the problem of obtaining the field results under the field type into the trained quality detection model. The quality detection model outputs the field results, and judges the consistency of the field results with the field results in the special disease database to determine the accuracy of the field results in the special disease database, thereby detecting the quality of the special disease database.

[0041] An embodiment of the present invention further provides an electronic device, comprising: a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus. The communication bus can be used for information transmission between the electronic device and the sensor. The processor can call the logic instructions in the memory to execute the following method: obtain the field type of the special disease database to be detected; construct several questions for obtaining the field results under a certain field type; uniformly manage the several questions for obtaining the field results under the field type and construct them into a prompt question bank; read the field results under the field type in the special disease database, and obtain the original medical text corresponding to the field result, verify the accuracy of the field result, and select the original text data corresponding to some of the field results that have passed the verification as training text data; read the training text data, and obtain the field results corresponding to the training text data under the constructed questions to obtain the target answers; input the training text data, the target answers and the questions for obtaining the field results in the prompt question bank into a pre-constructed quality detection model, train the quality detection model, and obtain a trained quality detection model; input the original medical text data and the questions for obtaining the field results under the field type into the trained quality detection model, the quality detection model outputs the field results, and judges the consistency of the field results with the field results in the special disease database to determine the accuracy of the field results in the special disease database, thereby detecting the quality of the special disease database.

[0042] In addition, the logic instructions in the above-mentioned memory can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program code.

[0043] An embodiment of the present invention provides a non-transitory computer-readable storage medium, which stores computer instructions. The computer instructions enable a computer to execute the method provided by the above method embodiment, for example, including: Obtain the field type of the special disease database to be tested; construct several questions for obtaining field results under a certain field type; uniformly manage several questions for obtaining field results under the field type and construct them into a prompt question bank; read the field results under the field type in the special disease database, and obtain the original medical text corresponding to the field result, verify the accuracy of the field result, and select some original text data corresponding to the field results that have passed the verification as training text data; read the training text data, and obtain the field results corresponding to the training text data under the constructed questions to obtain the target answers; input the training text data, the target answers and the questions for obtaining field results in the prompt question bank into a pre-constructed quality detection model, train the quality detection model, and obtain a trained quality detection model; input the original medical text data and the questions for obtaining field results under the field type into the trained quality detection model, the quality detection model outputs the field results, and judges the consistency of the field results with the field results in the special disease database to determine the accuracy of the field results in the special disease database, thereby detecting the quality of the special disease database.

[0044] It should be understood that the order of execution of the steps in the above embodiment does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

[0045] The above-mentioned embodiment provides an embodiment of a quality detection system for a special disease database. This system and the quality detection method for a special disease database in the above-mentioned embodiments belong to the same inventive concept. For details not described in detail in the embodiment of the quality detection system for a special disease database, please refer to the embodiment of the quality detection method for a special disease database mentioned above.

[0046] The quality detection system for the disease-specific database is a combination of the units and algorithm steps of each example described in the embodiments disclosed herein, and can be implemented by electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0047] Those skilled in the art will appreciate that various aspects of the quality detection method for a disease-specific database can be implemented as a system, method, or program product. Therefore, various aspects of the present disclosure can be specifically implemented in the following forms, namely: a complete hardware implementation, a complete software implementation (including firmware, microcode, etc.), or a combination of hardware and software, which can be collectively referred to as "circuit", "module" or "system".

[0048] It should be understood by those skilled in the art that the embodiments of the present invention described above and shown in the accompanying drawings are only examples and do not limit the present invention. The advantages of the present invention have been fully and effectively achieved. The functional and structural principles of the present invention have been demonstrated and explained in the embodiments, and the embodiments of the present invention may be deformed or modified in any way without departing from the principles.

[0049] The algorithm and display provided herein are not inherently related to any particular computer, virtual system or other device. Various general purpose systems can also be used together with the teachings based on this. According to the above description, it is obvious that the structure required for constructing such systems. In addition, the present invention is not directed to any specific programming language either. It should be understood that various programming languages ​​can be utilized to realize the content of the present invention described herein, and the description of the above specific languages ​​is for disclosing the best mode of the present invention.

[0050] In the description provided herein, a large number of specific details are described. However, it is understood that embodiments of the present invention can be practiced without these specific details. In some instances, well-known methods, structures and techniques are not shown in detail so as not to obscure the understanding of this description.

[0051] Similarly, it should be understood that in order to simplify the present disclosure and help understand one or more of the various inventive aspects, in the above description of exemplary embodiments of the present invention, the various features of the present invention are sometimes grouped together into a single embodiment, figure, or description thereof. More specifically, inventive aspects lie in less than all the features of a single embodiment disclosed above.

[0052] Those skilled in the art will appreciate that the modules in the devices in the embodiments can be adaptively changed and arranged in one or more devices different from the embodiments. The modules or units or components in the embodiments can be combined into one module or unit or component, and in addition they can be divided into multiple submodules or subunits or subassemblies. Except that at least some of such features and / or processes or units are mutually exclusive, all features disclosed in this specification and all processes or units of any method or device disclosed in this manner can be combined in any combination. Unless otherwise clearly stated, each feature disclosed in this specification can be replaced by an alternative feature providing the same, equivalent or similar purpose.

[0053] In addition, those skilled in the art will appreciate that, although some embodiments described herein include certain features included in other embodiments but not other features, the combination of features of different embodiments is meant to be within the scope of the present invention and form different embodiments. For example, any one of the claimed embodiments may be used in any combination.

[0054] The various component embodiments of the present invention can be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. It should be understood by those skilled in the art that a microprocessor or digital signal processor (DSP) can be used in practice to implement some or all functions of some or all components in physical devices such as servers, clients and systems for resource optimization according to embodiments of the present invention. The present invention can also be implemented as a device or apparatus program (e.g., computer program and computer program product) for executing part or all of the methods described herein. Such a program implementing the present invention can be stored on a computer-readable medium, or can have the form of one or more signals. Such a signal can be downloaded from an Internet website, or provided on a carrier signal, or provided in any other form.

Claims

1. A quality detection method for a disease-specific database, wherein the disease-specific database stores field results of several field types, characterized in that: The method comprises: Get the field type of the disease database to be tested; Construct several questions to obtain field results under a certain field type; Manage several questions that obtain field results under this field type in a unified manner and build them into a prompt question bank; Read the field results under the field type in the disease-specific database, obtain the original medical text corresponding to the field results, verify the accuracy of the field results, and select the original text data corresponding to some of the verified field results as training text data; Read the training text data, and obtain the field results corresponding to the constructed question, to obtain the target answer; Input the training text data, the target answer, and the questions in the prompt question bank that obtain the field results into the pre-built quality detection model, train the quality detection model, and obtain a trained quality detection model; The original medical text data and the questions about the field results under the field type are input into the trained quality detection model. The quality detection model outputs the field results, and the consistency of the field results with the field results in the special disease database is judged to determine the accuracy of the field results in the special disease database, thereby detecting the quality of the special disease database.

2. The quality inspection method according to claim 1, characterized in that: Construct several questions to obtain the field results under a certain field type, where: For each field type in the disease-specific database, construct at least three questions to obtain field results.

3. The quality inspection method according to claim 1, characterized in that: Read the field results under the field type in the disease-specific database and obtain the original medical text corresponding to the field results, including the following specific steps: Use the database connection library or API to establish a connection with the disease-specific database, and execute query statements to obtain field results in the disease-specific database; Through the association relationship or index in the disease-specific database, the original medical text corresponding to the field result is found, and the obtained original medical text is output.

4. The quality inspection method according to claim 1, characterized in that: The pre-built quality detection model is the Qwen1.5-14b model.

5. The quality inspection method according to claim 4, characterized in that: After inputting the training text data, the target answer and the questions in the prompt question bank for obtaining the field results into the pre-built quality inspection model, training the quality inspection model and obtaining the trained quality inspection model, the following specific steps are also included: Input the training text data and the questions in the prompt question bank that get the field results into the trained Qwen1.5-14b model, and the trained Qwen1.5-14b model outputs the predicted field results; Through the loss function, the target answer and the predicted field result are calculated to obtain the calculation result; The Qwen1.5-14b model is optimized based on the calculation results using the Adam optimizer; Repeat the above steps until the loss function converges, the training of the Qwen1.5-14b model is completed, and the optimized Qwen1.5-14b model is obtained.

6. The quality inspection method according to claim 5, characterized in that: Before the original medical text data and the questions in the prompt question bank for obtaining the field results under the field type are input into the trained quality inspection model, the trained quality inspection model outputs the predicted field results, and also includes: Set the training parameters of the Qwen1.5-14b model, including batch_size, epoch, and gradient accumulation.

7. The quality inspection method according to claim 6, characterized in that: The loss function is the cross entropy loss function, and the expression of the cross entropy loss function is: In the formula, represents the loss function, Indicates the number of categories, represents the true value in the labeled data, Represents the model prediction value.

8. A quality detection system for a disease-specific database, characterized in that: The system comprises: A data field acquisition unit, used to acquire the field type of the disease-specific database to be detected; A question construction unit is used to construct a number of questions for obtaining field results under a certain field type; The prompt question bank construction unit is used to uniformly manage several questions that obtain field results under this field type and construct them into a prompt question bank; A data extraction unit is used to read the field results under the field type in the disease-specific database, obtain the original medical text corresponding to the field results, verify the accuracy of the field results, and select the original text data corresponding to some of the field results that have passed the verification as training text data; A target answer acquisition unit is used to read the training text data and obtain the field results corresponding to the training text data under the constructed question to obtain the target answer; The quality detection model training unit is used to input the training text data, the target answer and the questions in the prompt question bank for obtaining the field results into the pre-built quality detection model, train the quality detection model, and obtain the trained quality detection model; The quality detection unit is used to input the original medical text data and the problem of obtaining the field results under the field type into the trained quality detection model. The quality detection model outputs the field results, and judges the consistency of the field results with the field results in the special disease database to determine the accuracy of the field results in the special disease database, thereby detecting the quality of the special disease database.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively coupled to the at least one processor; The memory stores computer program instructions that can be executed by at least one processor, and the computer program instructions are executed by the at least one processor so that the at least one processor can execute the quality detection method for a special disease database as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium stores computer instructions, and the computer instructions enable the computer to execute the quality detection method for a disease-specific database as described in any one of claims 1 to 7.

Images