A method, device, electronic device and storage medium for obtaining violation identification results

Abstract

The present application discloses a method, device, electronic device and storage medium for obtaining violation identification results, the method comprising: determining a first learning model that has passed the evaluation based on first evaluation data; constructing a second training set and a second test set; training a second learning model based on the second training set, obtaining second evaluation data of the second learning model on the second test set, and determining a second learning model that has passed the evaluation based on the second evaluation data; constructing a third training set and a third test set; training a third learning model based on the third training set, obtaining third evaluation data of the third learning model on the third test set, and determining a third learning model that has passed the evaluation based on the third evaluation data; generating a violation identification result of the current electronic medical record based on the first learning model that has passed the evaluation, the second learning model that has passed the evaluation, and the third learning model that has passed the evaluation. The present application is conducive to improving the efficiency of obtaining the current violation identification results.

Description

Technical Field

[0001] The present application relates to the field of Internet technology, and in particular to a method, device, electronic device and storage medium for obtaining violation identification results. Background Art

[0002] With the development of medical informatization, a large number of electronic medical records are collected and stored. Electronic medical records promote the seamless circulation and sharing of medical information, break the limitations of traditional paper medical records, make cross-institutional cooperation more convenient, and provide users with more coherent and coordinated services.

[0003] However, the current process of obtaining the current violation identification result of the electronic medical record is cumbersome, which is not conducive to improving the efficiency of obtaining the current violation identification result. The reason is that the existing technology mainly adopts the manual acquisition method to obtain the current violation identification result of the current electronic medical record, and the manual acquisition method consumes a lot of human resources and time resources, is inefficient, increases the acquisition time of the current violation identification result of the current electronic medical record, and is easily affected by subjective factors. Therefore, it is not conducive to improving the efficiency of obtaining the current violation identification result. Summary of the invention

[0004] The embodiments of the present application provide a method, device, electronic device and storage medium for obtaining violation identification results to solve the technical problem that the process of obtaining the current violation identification results of the current electronic medical record is cumbersome and not conducive to improving the efficiency of obtaining the current violation identification results.

[0005] In a first aspect, an embodiment of the present application provides a method for obtaining a violation identification result, which is applied to an electronic device, wherein the electronic device stores a first learning model, a second learning model, and a third learning model, and the method for obtaining a violation identification result includes:

[0006] Acquire compliant electronic medical records, combine the compliant electronic medical records and the compliant labels corresponding to the compliant electronic medical records into compliant samples, acquire illegal electronic medical records, combine the illegal electronic medical records and the illegal labels corresponding to the illegal electronic medical records into illegal samples, combine a plurality of the compliant samples and a plurality of the illegal samples into a data set, and divide the data set into a first training set and a first test set;

[0007] Training the first learning model based on the first training set, obtaining first evaluation data of the first learning model on the first test set, and determining the first learning model that passes the evaluation according to the first evaluation data and a predefined first evaluation method;

[0008] Obtaining a preset violation sequence number identified by the first learning model through evaluation in the first training set, using the identified preset violation sequence number as a target sequence number, combining the target sequence number and a preset medical insurance catalog code corresponding to the target sequence number into a second sample, and constructing a second training set and a second test set based on different second samples;

[0009] Training the second learning model based on the second training set, obtaining second evaluation data of the second learning model on the second test set, and determining the second learning model that passes the evaluation according to the second evaluation data and a predefined second evaluation method;

[0010] Obtain the preset medical insurance catalog code identified by the second learning model through evaluation in the second training set, set the identified preset medical insurance catalog code as the target code, form a third sample with the target sequence number, the target code and the preset violation category, and construct a third training set and a third test set based on different third samples;

[0011] Training the third learning model based on the third training set, obtaining third evaluation data of the third learning model on the third test set, and determining the third learning model that passes the evaluation according to the third evaluation data and a predefined third evaluation method;

[0012] A violation identification result of the current electronic medical record is generated according to the first learning model that has passed the evaluation, the second learning model that has passed the evaluation, the third learning model that has passed the evaluation, and a predefined generation method.

[0013] In a possible implementation manner of the first aspect, the training the first learning model based on the first training set, obtaining first evaluation data of the first learning model on the first test set, and determining, according to the first evaluation data and a predefined first evaluation method, that the first learning model passes the evaluation includes:

[0014] Training the first learning model based on the first training set, and obtaining a prediction label output by the first learning model during the training process;

[0015] Obtaining a first loss value between the predicted label and the preset violation label;

[0016] When the first loss value is less than a preset loss value, outputting the first learning model after training, and obtaining first evaluation data of the first learning model after training on the first test set, wherein the first evaluation data includes a first accuracy rate, a first precision rate, a first recall rate, and a first score;

[0017] The first learning model whose first accuracy is greater than a preset accuracy, whose first precision is greater than a preset precision, whose first recall is greater than a preset recall, and whose first score is greater than a preset score is set as the first learning model that passes the evaluation.

[0018] In a possible implementation manner of the first aspect, training the first learning model based on the first training set and obtaining a prediction label output by the first learning model during the training process includes:

[0019] The first learning model is trained based on the first training set to obtain a preset medical consultation information table, a preset settlement information table, and a preset expense details table in a preset electronic medical record;

[0020] The feature vector of the preset medical information table, the feature vector of the preset settlement information table, and the feature vector of the preset expense details table are concatenated to generate a preset feature vector;

[0021] The preset feature vector is input into the first learning model in the training process, and a prediction label output by the first learning model in the training process based on the preset feature vector is obtained.

[0022] In a possible implementation manner of the first aspect, the training the second learning model based on the second training set, obtaining second evaluation data of the second learning model on the second test set, and determining, according to the second evaluation data and a predefined second evaluation method, that the second learning model passes the evaluation includes:

[0023] Training the second learning model based on the second training set, obtaining the predicted medical insurance catalog code output by the second learning model during the training process, and obtaining a second loss value between the predicted medical insurance catalog code and the preset medical insurance catalog code;

[0024] When the second loss value is less than a preset loss value, outputting the trained second learning model, and obtaining second evaluation data of the trained second learning model on the second test set, wherein the second evaluation data includes a second accuracy rate, a second precision rate, a second recall rate, and a second score;

[0025] The second learning model whose second accuracy is greater than a preset accuracy, whose second precision is greater than a preset precision, whose second recall is greater than a preset recall, and whose second score is greater than a preset score is set as the second learning model that passes the evaluation.

[0026] In a possible implementation manner of the first aspect, the training the third learning model based on the third training set, obtaining third evaluation data of the third learning model on the third test set, and determining the third learning model that passes the evaluation according to the third evaluation data and a predefined third evaluation method, includes:

[0027] Training the third learning model based on the third training set, and obtaining predicted illegal content output by the third learning model during the training process;

[0028] Obtaining a third loss value between the predicted illegal content and the preset illegal content;

[0029] When the third loss value is less than the preset loss value, output the trained third learning model, and obtain third evaluation data of the trained third learning model on the third test set, wherein the third evaluation data includes a third accuracy rate, a third precision rate, a third recall rate, and a third score;

[0030] The third learning model whose third accuracy is greater than the preset accuracy, whose third precision is greater than the preset precision, whose third recall is greater than the preset recall, and whose third score is greater than the preset score is set as the third learning model that passes the evaluation.

[0031] In a possible implementation of the first aspect, generating a violation identification result of the current electronic medical record according to the evaluated first learning model, the evaluated second learning model, the evaluated third learning model, and a predefined generation method includes:

[0032] Obtaining a current label output by the first learning model that has passed the evaluation based on the current feature vector, and when the current label is the violation label, obtaining a current violation sequence number of the current electronic medical record;

[0033] Input the current violation serial number into the second learning model that has passed the evaluation, and obtain the current medical insurance catalog code recognition result output by the second learning model that has passed the evaluation based on the current violation serial number;

[0034] Input the current medical insurance catalog code recognition result into the evaluated third learning model, and obtain the current violation category recognition result output by the evaluated third learning model based on the current medical insurance catalog code recognition result;

[0035] The violation label, the current violation serial number, the current medical insurance catalog code identification result, and the current violation category identification result are combined to generate a violation identification result of the current electronic medical record.

[0036] In a possible implementation of the first aspect, after generating a violation identification result of the current electronic medical record according to the evaluated first learning model, the evaluated second learning model, the evaluated third learning model, and a predefined generation method, the violation identification result acquisition method includes:

[0037] A display area of ​​a display page is obtained, and the violation identification result is displayed through the display area.

[0038] In a second aspect, an embodiment of the present application provides a violation identification result acquisition device, which is applied to an electronic device, wherein the electronic device stores a first learning model, a second learning model, and a third learning model, including:

[0039] A first acquisition module is used to acquire compliant electronic medical records, combine the compliant electronic medical records and the compliant labels corresponding to the compliant electronic medical records into compliant samples, acquire illegal electronic medical records, combine the illegal electronic medical records and the illegal labels corresponding to the illegal electronic medical records into illegal samples, combine multiple compliant samples and multiple illegal samples into a data set, and divide the data set into a first training set and a first test set;

[0040] A first evaluation module is used to train the first learning model based on the first training set, obtain first evaluation data of the first learning model on the first test set, and determine the first learning model that passes the evaluation according to the first evaluation data and a predefined first evaluation method;

[0041] A second acquisition module is used to acquire a preset violation sequence number identified by the first learning model through evaluation in the first training set, use the identified preset violation sequence number as a target sequence number, form a second sample with the target sequence number and a preset medical insurance catalog code corresponding to the target sequence number, and construct a second training set and a second test set based on different second samples;

[0042] A second evaluation module is used to train the second learning model based on the second training set, obtain second evaluation data of the second learning model on the second test set, and determine the second learning model that passes the evaluation according to the second evaluation data and a predefined second evaluation method;

[0043] A construction module is used to obtain the preset medical insurance catalog code identified by the second learning model through evaluation in the second training set, set the identified preset medical insurance catalog code as the target code, form a third sample with the target sequence number, the target code and the preset violation category, and construct a third training set and a third test set based on different third samples;

[0044] A third evaluation module is used to train the third learning model based on the third training set, obtain third evaluation data of the third learning model on the third test set, and determine the third learning model that passes the evaluation according to the third evaluation data and a predefined third evaluation method;

[0045] A generation module is used to generate a violation identification result of the current electronic medical record according to the first learning model that has passed the evaluation, the second learning model that has passed the evaluation, the third learning model that has passed the evaluation, and a predefined generation method.

[0046] In a third aspect, an embodiment of the present application provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the violation identification result acquisition method of the first aspect when executing the computer program.

[0047] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for obtaining violation identification results of the first aspect is implemented.

[0048] In a fifth aspect, an embodiment of the present application provides a computer program product. When the computer program product is run on an electronic device, the electronic device executes the violation identification result acquisition method of the first aspect.

[0049] The beneficial effects of the embodiments of the present application lie in two aspects. On the one hand, based on the first learning model that has passed the evaluation, the second learning model that has passed the evaluation, the third learning model that has passed the evaluation, and a predefined generation method, a violation identification result of the current electronic medical record is generated. Since manual acquisition is not required, the acquisition time of the current violation identification result of the current electronic medical record is reduced, which is beneficial to improving the acquisition efficiency of the current violation identification result of the current electronic medical record. On the other hand, since the first learning model that has passed the evaluation, the second learning model that has passed the evaluation, and the third learning model that has passed the evaluation will not be affected by subjective factors, it is beneficial to improve the reliability of the current violation identification result of the current electronic medical record obtained. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0051] Figure 1An application scenario diagram of the method for obtaining violation identification results provided in an embodiment of the present application;

[0052] Figure 2 It is a flowchart of a method for obtaining violation identification results provided in an embodiment of the present application;

[0053] Figure 3 A flowchart for obtaining prediction labels provided in an embodiment of the present application;

[0054] Figure 4 A schematic block diagram of a violation identification result acquisition device provided in an embodiment of the present application;

[0055] Figure 5 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical solutions and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present application.

[0057] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.

[0058] In addition, in the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0059] In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the fact that they can be implemented by ordinary technicians in this field. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such combination of technical solutions does not exist and is not within the scope of protection required by this application.

[0060] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.

[0061] The method for obtaining violation identification results provided in the embodiments of the present application can be applied to electronic devices such as mobile phones, tablet computers, wearable devices, vehicle-mounted devices, augmented reality (AR) / virtual reality (VR) devices, laptop computers, ultra-mobile personal computers (UMPC), netbooks, personal digital assistants (PDA), etc. The embodiments of the present application do not impose any restrictions on the specific types of electronic devices.

[0062] See also Figure 1 , Figure 1 The application scenario diagram of the violation identification result acquisition method provided in the embodiment of the present application is described in detail as follows:

[0063] The electronic device is connected to the electronic medical record system to obtain the compliant electronic medical records and the non-compliant electronic medical records from the electronic medical record system.

[0064] The preset electronic medical record may come from a data table of the electronic medical record system or from multiple data tables of the database. For ease of explanation, the preset electronic medical record comes from two data tables as an example, which is described in detail as follows:

[0065] The electronic device is connected to the electronic medical record system, and reads Data Table 1 and Data Table 2 through the data interface provided by the electronic medical record system, obtains the compliant electronic medical record from Data Table 1, and obtains the illegal electronic medical record from Data Table 2.

[0066] In an embodiment of the present application, the electronic device can read different data tables at the same time and obtain preset electronic medical records from different data tables.

[0067] See also Figure 2 , Figure 2 It is a flow chart of a method for obtaining violation identification results provided in an embodiment of the present application. The method can be applied to an electronic device that stores a first learning model, a second learning model, and a third learning model.

[0068] like Figure 2 As shown, the method for obtaining violation identification results provided in the embodiment of the present application includes the following steps, which are described in detail as follows:

[0069] S201, obtaining compliant electronic medical records, combining the compliant electronic medical records and the compliant labels corresponding to the compliant electronic medical records into compliant samples, obtaining illegal electronic medical records, combining the illegal electronic medical records and the illegal labels corresponding to the illegal electronic medical records into illegal samples, combining a plurality of the compliant samples and a plurality of the illegal samples into a data set, and dividing the data set into a first training set and a first test set;

[0070] Among them, compliant electronic medical records are electronic medical records that comply with regulations.

[0071] Among them, illegal electronic medical records are electronic medical records that violate regulations.

[0072] Among them, illegal electronic medical records include electronic medical records with duplicate items and electronic medical records with charging prices exceeding the maximum price.

[0073] For the sake of illustration, an example is given below:

[0074] The items in Electronic Medical Record 1 are repeated, and Electronic Medical Record 1 is the illegal electronic medical record.

[0075] The charging price of Electronic Medical Record 2 exceeds the maximum price, and Electronic Medical Record 2 is an illegal electronic medical record.

[0076] S202, training the first learning model based on the first training set, obtaining first evaluation data of the first learning model on the first test set, and determining the first learning model that passes the evaluation according to the first evaluation data and a predefined first evaluation method;

[0077] The step of training the first learning model based on the first training set, obtaining first evaluation data of the first learning model on the first test set, and determining that the first learning model passes the evaluation according to the first evaluation data and a predefined first evaluation method includes:

[0078] Training the first learning model based on the first training set, and obtaining a prediction label output by the first learning model during the training process;

[0079] Obtaining a first loss value between the predicted label and the preset violation label;

[0080] When the first loss value is less than a preset loss value, outputting the first learning model after training, and obtaining first evaluation data of the first learning model after training on the first test set, wherein the first evaluation data includes a first accuracy rate, a first precision rate, a first recall rate, and a first score;

[0081] The first learning model whose first accuracy is greater than a preset accuracy, whose first precision is greater than a preset precision, whose first recall is greater than a preset recall, and whose first score is greater than a preset score is set as the first learning model that passes the evaluation.

[0082] Wherein, a first loss value between the predicted label and the preset violation label is obtained through a cross entropy loss function.

[0083] Exemplarily, the first accuracy is the accuracy of the first learning model after training on the first test set;

[0084] The first accuracy is the accuracy of the first learning model after training on the first test set;

[0085] The first recall rate is a recall rate of the first learning model after training on the first test set;

[0086] The first score is the F1 score of the first learning model after training on the first test set.

[0087] The Chinese name of the F1 score is F1 score, and the English name of the F1 score is F1 Score. The F1 score is an indicator used to measure the accuracy of a binary classification model in machine learning. The F1 score takes into account both precision and recall.

[0088] The F1 score ranges from 0 to 1.

[0089] Among them, the closer the F1 score of the first learning model on the first test set is to 1, the higher the prediction accuracy of the first learning model is, and the closer the F1 score of the first learning model on the first test set is to 0, the lower the prediction accuracy of the first learning model is.

[0090] Among them, the preset accuracy, preset precision, preset recall rate, and preset score are set by the user or by system default and are not restricted here.

[0091] For the sake of illustration, an example is given below:

[0092] For example, the first learning model after training is model A, and the first test set includes compliant samples and noncompliant samples. When model A identifies compliant samples in the first test set, the first evaluation data is as follows:

[0093] The first accuracy is 0.945, the first precision is 0.96, the first recall is 0.97, and the first score is 0.96;

[0094] When the preset accuracy is 0.9, the preset precision is 0.9, the preset recall is 0.9, and the preset score is 0.9, model A is set as the first learning model that passes the evaluation. For example, the first learning model after training is model A, and the first test set includes compliant samples and noncompliant samples. When model A identifies noncompliant samples in the first test set, the first evaluation data is as follows:

[0095] The first accuracy is 0.995, the first precision is 0.996, the first recall is 0.997, and the first score is 0.996;

[0096] When the preset accuracy is 0.9, the preset precision is 0.9, the preset recall is 0.9, and the preset score is 0.9, model A is set as the first learning model that passes the evaluation.

[0097] S203, obtaining a preset violation sequence number identified by the first learning model through evaluation in the first training set, using the identified preset violation sequence number as a target sequence number, combining the target sequence number and the preset medical insurance catalog code corresponding to the target sequence number into a second sample, and constructing a second training set and a second test set based on different second samples;

[0098] S204, training the second learning model based on the second training set, obtaining second evaluation data of the second learning model on the second test set, and determining the second learning model that passes the evaluation according to the second evaluation data and a predefined second evaluation method;

[0099] The step of training the second learning model based on the second training set, obtaining second evaluation data of the second learning model on the second test set, and determining that the second learning model passes the evaluation according to the second evaluation data and a predefined second evaluation method includes:

[0100] Training the second learning model based on the second training set, obtaining the predicted medical insurance catalog code output by the second learning model during the training process, and obtaining a second loss value between the predicted medical insurance catalog code and the preset medical insurance catalog code;

[0101] When the second loss value is less than a preset loss value, outputting the trained second learning model, and obtaining second evaluation data of the trained second learning model on the second test set, wherein the second evaluation data includes a second accuracy rate, a second precision rate, a second recall rate, and a second score;

[0102] The second learning model whose second accuracy is greater than a preset accuracy, whose second precision is greater than a preset precision, whose second recall is greater than a preset recall, and whose second score is greater than a preset score is set as the second learning model that passes the evaluation.

[0103] Among them, the second loss value between the predicted medical insurance catalog code and the preset medical insurance catalog code is obtained through the cross entropy loss function.

[0104] Exemplarily, the second accuracy is the accuracy of the trained second learning model on the second test set;

[0105] The second accuracy is the accuracy of the trained second learning model on the second test set;

[0106] The second recall rate is a recall rate of the second learning model after training on the second test set;

[0107] The second score is the F1 score of the trained second learning model on the second test set.

[0108] Among them, the closer the F1 score of the second learning model on the second test set is to 1, the higher the prediction accuracy of the second learning model is, and the closer the F1 score of the second learning model on the second test set is to 0, the lower the prediction accuracy of the second learning model is.

[0109] For the sake of illustration, an example is given below:

[0110] For example, the second learning model after training is model B. When model B identifies the second sample of the second test set, the second evaluation data is as follows:

[0111] The second accuracy is 0.995, the second precision is 0.996, the second recall is 0.997, and the second score is 0.969;

[0112] When the preset accuracy is 0.8, the preset precision is 0.8, the preset recall is 0.8, and the preset score is 0.8, model B is set as the second learning model that passes the evaluation.

[0113] S205, obtaining the preset medical insurance catalog code identified by the evaluated second learning model in the second training set, setting the identified preset medical insurance catalog code as a target code, combining the target sequence number, the target code and the preset violation category into a third sample, and constructing a third training set and a third test set based on different third samples;

[0114] S206, training the third learning model based on the third training set, obtaining third evaluation data of the third learning model on the third test set, and determining the third learning model that passes the evaluation according to the third evaluation data and a predefined third evaluation method;

[0115] The step of training the third learning model based on the third training set, obtaining third evaluation data of the third learning model on the third test set, and determining the third learning model that passes the evaluation according to the third evaluation data and a predefined third evaluation method includes:

[0116] Training the third learning model based on the third training set, and obtaining predicted illegal content output by the third learning model during the training process;

[0117] Obtaining a third loss value between the predicted illegal content and the preset illegal content;

[0118] When the third loss value is less than the preset loss value, output the trained third learning model, and obtain third evaluation data of the trained third learning model on the third test set, wherein the third evaluation data includes a third accuracy rate, a third precision rate, a third recall rate, and a third score;

[0119] The third learning model whose third accuracy is greater than the preset accuracy, whose third precision is greater than the preset precision, whose third recall is greater than the preset recall, and whose third score is greater than the preset score is set as the third learning model that passes the evaluation.

[0120] Among them, a third loss value between the predicted illegal content and the preset illegal content is obtained through the cross entropy loss function.

[0121] Exemplarily, the third accuracy is the accuracy of the trained third learning model on the third test set;

[0122] The third accuracy rate is the accuracy rate of the trained third learning model on the third test set;

[0123] The third recall rate is the recall rate of the third learning model after training on the third test set;

[0124] The third score is the F1 score of the third learning model after training on the third test set.

[0125] Among them, the closer the F1 score of the third learning model on the third test set is to 1, the higher the prediction accuracy of the third learning model is, and the closer the F1 score of the third learning model on the third test set is to 0, the lower the prediction accuracy of the third learning model is.

[0126] For the sake of illustration, an example is given below:

[0127] For example, the third learning model after training is model C. When model C identifies the third sample of the third test set, the third evaluation data is as follows:

[0128] The third accuracy is 0.995, the third precision is 0.996, the third recall is 0.897, and the third score is 0.909;

[0129] When the preset accuracy is 0.9, the preset precision is 0.9, the preset recall is 0.8, and the preset score is 0.8, model C is set as the third learning model that passes the evaluation.

[0130] S207, generating a violation identification result of the current electronic medical record according to the first learning model that has passed the evaluation, the second learning model that has passed the evaluation, the third learning model that has passed the evaluation, and a predefined generation method.

[0131] The generating of the violation identification result of the current electronic medical record according to the first learning model that has passed the evaluation, the second learning model that has passed the evaluation, the third learning model that has passed the evaluation and the predefined generation method includes:

[0132] Obtaining a current label output by the first learning model that has passed the evaluation based on the current feature vector, and when the current label is the violation label, obtaining a current violation sequence number of the current electronic medical record;

[0133] Input the current violation serial number into the second learning model that has passed the evaluation, and obtain the current medical insurance catalog code recognition result output by the second learning model that has passed the evaluation based on the current violation serial number;

[0134] Input the current medical insurance catalog code recognition result into the evaluated third learning model, and obtain the current violation category recognition result output by the evaluated third learning model based on the current medical insurance catalog code recognition result;

[0135] The violation label, the current violation serial number, the current medical insurance catalog code identification result, and the current violation category identification result are combined to generate a violation identification result of the current electronic medical record.

[0136] For the sake of explanation, the violation identification results are given as follows:

[0137] For example, violation identification result 1: the current label is a violation label, the current violation serial number is 111, the current medical insurance catalog code identification result is 002504040010200, and the current violation category identification result is item duplication.

[0138] For example, violation identification result 2: the current label is a violation label, the current violation serial number is 222, the current medical insurance catalog code identification result is 002504040010300, and the current violation category identification result is exceeding the maximum price.

[0139] Exemplarily, the obtaining of a current label output by the first learning model that has passed the evaluation based on a current feature vector, and when the current label is the violation label, obtaining a current violation sequence number of the current electronic medical record includes:

[0140] Obtain the current electronic medical record, and obtain the current medical consultation information table, current settlement information table, and current expense details table of the current electronic medical record;

[0141] The feature vector of the current medical information table, the feature vector of the current settlement information table, and the feature vector of the current expense details table are concatenated to generate a current feature vector;

[0142] The current feature vector is input into the first learning model that has passed the evaluation, and a current violation sequence number output by the first learning model that has passed the evaluation based on the current feature vector is obtained.

[0143] Among them, the current electronic medical record is the electronic medical record to be identified.

[0144] Among them, the current medical information table is a user information table to be identified.

[0145] The user information table to be identified includes the user gender, user age, unit type, industry, and number of days the user is hospitalized of the current electronic medical record.

[0146] For the sake of illustration, an example is given below:

[0147] For example, in the user information table to be identified, the first row is the gender of the user of the current electronic medical record;

[0148] The second row is the age of the current EMR user;

[0149] The third line is the unit type of the current electronic medical record;

[0150] The fourth row is the industry to which the current electronic medical record belongs;

[0151] The fifth row is the number of days the user of the current electronic medical record has been hospitalized.

[0152] After generating the violation identification result of the current electronic medical record according to the first learning model that has passed the evaluation, the second learning model that has passed the evaluation, the third learning model that has passed the evaluation, and a predefined generation method, the violation identification result acquisition method includes:

[0153] Step A, obtaining a display area of ​​a display page, and displaying the violation identification result through the display area.

[0154] Exemplarily, after generating the violation identification result of the current electronic medical record according to the evaluated first learning model, the evaluated second learning model, the evaluated third learning model and a predefined generation method, the violation identification result acquisition method includes:

[0155] Step B, obtaining a preset storage area, and storing the violation identification result in the storage area.

[0156] Among them, step A can be executed before or after step B, and step A can also be executed simultaneously with step B. The specific execution order is not limited here.

[0157] It should be noted that the direct purpose of this application is to improve the efficiency of obtaining the current violation identification results of the current electronic medical record. Since there is no need for manual acquisition, a large amount of human resources and time resources are saved, thereby reducing the time for obtaining the current violation identification results of the current electronic medical record, which is conducive to improving the efficiency of obtaining the current violation identification results of the current electronic medical record, thereby achieving the direct purpose of this application.

[0158] The beneficial effects of the embodiments of the present application lie in two aspects. On the one hand, based on the first learning model that has passed the evaluation, the second learning model that has passed the evaluation, the third learning model that has passed the evaluation, and a predefined generation method, a violation identification result of the current electronic medical record is generated. Since manual acquisition is not required, the acquisition time of the current violation identification result of the current electronic medical record is reduced, which is beneficial to improving the acquisition efficiency of the current violation identification result of the current electronic medical record. On the other hand, since the first learning model that has passed the evaluation, the second learning model that has passed the evaluation, and the third learning model that has passed the evaluation will not be affected by subjective factors, it is beneficial to improve the reliability of the current violation identification result of the current electronic medical record obtained.

[0159] See also Figure 3 , Figure 3 The flowchart for obtaining the predicted label provided in the embodiment of the present application is described in detail as follows:

[0160] S301, training the first learning model based on the first training set to obtain a preset medical information table, a preset settlement information table, and a preset expense details table in a preset electronic medical record;

[0161] Among them, the preset electronic medical record is an identified electronic medical record.

[0162] Among them, the preset medical information table is an identified user information table.

[0163] Among them, the identified user information table includes the user's gender, user age, unit type, industry, and user's hospitalization days in the preset electronic medical record.

[0164] For the sake of illustration, an example is given below:

[0165] For example, in the identified user information table, the first row is the gender of the user who presets the electronic medical record;

[0166] The second row is the age of the user who presets the electronic medical record;

[0167] The third row is the unit type for the preset electronic medical record;

[0168] The fourth row is the industry to which the preset electronic medical record belongs;

[0169] The fifth row is the number of hospitalization days for the user of the preset electronic medical record.

[0170] S302, concatenating the feature vector of the preset medical information table, the feature vector of the preset settlement information table, and the feature vector of the preset expense details table to generate a preset feature vector;

[0171] S303: Input the preset feature vector into the first learning model in the training process, and obtain a predicted label output by the first learning model in the training process based on the preset feature vector.

[0172] In the embodiment of the present application, by comparing the predicted label with the preset violation sequence number, the accuracy, stability and generalization ability of the first learning model can be understood, providing direction for the optimization and iteration of the first learning model.

[0173] The list processing request is processed based on the target component to obtain the list processing result, and the list processing result is displayed on the target page. Since the target component is based on a custom function, the list processing result can meet the user's personalized needs. Therefore, displaying the list processing result on the target page can improve the user's usage experience and the efficiency of information acquisition.

[0174] For the method of obtaining the violation identification result described in the above embodiment, please refer to Figure 4 , Figure 4 A schematic block diagram of a violation identification result acquisition device provided in an embodiment of the present application, Figure 4 The violation identification result obtaining device 400 shown can be applied to Figure 1 The electronic device in the application scenario diagram shown below takes the electronic device as an example. Figure 4 The violation identification result acquisition device 400 shown is described in detail. The violation identification result acquisition device 400 may include a first acquisition module 401, a first evaluation module 402, a second acquisition module 403, a second evaluation module 404, a construction module 405, a third evaluation module 406, and a generation module 407.

[0175] The first acquisition module 401 is used to acquire compliant electronic medical records, combine the compliant electronic medical records and the compliant labels corresponding to the compliant electronic medical records into compliant samples, acquire illegal electronic medical records, combine the illegal electronic medical records and the illegal labels corresponding to the illegal electronic medical records into illegal samples, combine multiple compliant samples and multiple illegal samples into a data set, and divide the data set into a first training set and a first test set;

[0176] A first evaluation module 402 is used to train the first learning model based on the first training set, obtain first evaluation data of the first learning model on the first test set, and determine the first learning model that passes the evaluation according to the first evaluation data and a predefined first evaluation method;

[0177] A second acquisition module 403 is used to acquire the preset violation sequence number identified by the first learning model through evaluation in the first training set, use the identified preset violation sequence number as the target sequence number, and form a second sample with the target sequence number and the preset medical insurance catalog code corresponding to the target sequence number, and construct a second training set and a second test set based on different second samples;

[0178] A second evaluation module 404 is used to train the second learning model based on the second training set, obtain second evaluation data of the second learning model on the second test set, and determine the second learning model that passes the evaluation according to the second evaluation data and a predefined second evaluation method;

[0179] A construction module 405 is used to obtain the preset medical insurance catalog code identified by the second learning model through evaluation in the second training set, set the identified preset medical insurance catalog code as the target code, form a third sample with the target sequence number, the target code and the preset violation category, and construct a third training set and a third test set based on different third samples;

[0180] A third evaluation module 406 is used to train the third learning model based on the third training set, obtain third evaluation data of the third learning model on the third test set, and determine the third learning model that passes the evaluation according to the third evaluation data and a predefined third evaluation method;

[0181] The generation module 407 is used to generate a violation identification result of the current electronic medical record according to the first learning model that has passed the evaluation, the second learning model that has passed the evaluation, the third learning model that has passed the evaluation, and a predefined generation method.

[0182] It should be noted that the various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0183] The beneficial effects of the embodiments of the present application lie in two aspects. On the one hand, based on the first learning model that has passed the evaluation, the second learning model that has passed the evaluation, the third learning model that has passed the evaluation, and a predefined generation method, a violation identification result of the current electronic medical record is generated. Since manual acquisition is not required, the acquisition time of the current violation identification result of the current electronic medical record is reduced, which is beneficial to improving the acquisition efficiency of the current violation identification result of the current electronic medical record. On the other hand, since the first learning model that has passed the evaluation, the second learning model that has passed the evaluation, and the third learning model that has passed the evaluation will not be affected by subjective factors, it is beneficial to improve the reliability of the current violation identification result of the current electronic medical record obtained.

[0184] See also Figure 5 , Figure 5 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application.

[0185] like Figure 5 As shown, Figure 5 The electronic device 2 includes: at least one processor 20, a memory 21, and a computer program 22 stored in the memory 21 and executable on the at least one processor 20, and when the processor 20 executes the computer program 22, the steps in any of the above-mentioned method embodiments are implemented.

[0186] The electronic device 2 may include, but is not limited to, a processor 20 and a memory 21. Those skilled in the art will appreciate that Figure 5 It is only an example of the electronic device 2 and does not constitute a limitation on the electronic device 2. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, it may also include input and output devices, network access devices, etc.

[0187] The processor 20 is used to run the computer program 22 stored in the memory 21, and implement the following steps when executing the computer program 22:

[0188] Acquire compliant electronic medical records, combine the compliant electronic medical records and the compliant labels corresponding to the compliant electronic medical records into compliant samples, acquire illegal electronic medical records, combine the illegal electronic medical records and the illegal labels corresponding to the illegal electronic medical records into illegal samples, combine a plurality of the compliant samples and a plurality of the illegal samples into a data set, and divide the data set into a first training set and a first test set;

[0189] Training the first learning model based on the first training set, obtaining first evaluation data of the first learning model on the first test set, and determining the first learning model that passes the evaluation according to the first evaluation data and a predefined first evaluation method;

[0190] Obtaining a preset violation sequence number identified by the first learning model through evaluation in the first training set, using the identified preset violation sequence number as a target sequence number, combining the target sequence number and a preset medical insurance catalog code corresponding to the target sequence number into a second sample, and constructing a second training set and a second test set based on different second samples;

[0191] Training the second learning model based on the second training set, obtaining second evaluation data of the second learning model on the second test set, and determining the second learning model that passes the evaluation according to the second evaluation data and a predefined second evaluation method;

[0192] Obtain the preset medical insurance catalog code identified by the second learning model that has passed the evaluation in the second training set, set the identified preset medical insurance catalog code as the target code, and form a third sample with the target serial number, the target code, and the preset violation category, and construct a third training set and a third test set based on different third samples; train the third learning model based on the third training set, obtain the third evaluation data of the third learning model on the third test set, and determine the third learning model that has passed the evaluation based on the third evaluation data and a predefined third evaluation method; generate a violation identification result of the current electronic medical record based on the first learning model that has passed the evaluation, the second learning model that has passed the evaluation, the third learning model that has passed the evaluation, and a predefined generation method.

[0193] In some embodiments, the processor 20 is configured to implement:

[0194] Training the first learning model based on the first training set, and obtaining a prediction label output by the first learning model during the training process;

[0195] Obtaining a first loss value between the predicted label and the preset violation label;

[0196] When the first loss value is less than a preset loss value, outputting the first learning model after training, and obtaining first evaluation data of the first learning model after training on the first test set, wherein the first evaluation data includes a first accuracy rate, a first precision rate, a first recall rate, and a first score;

[0197] The first learning model whose first accuracy is greater than a preset accuracy, whose first precision is greater than a preset precision, whose first recall is greater than a preset recall, and whose first score is greater than a preset score is set as the first learning model that passes the evaluation.

[0198] In some embodiments, the processor 20 is configured to implement:

[0199] The first learning model is trained based on the first training set to obtain a preset medical consultation information table, a preset settlement information table, and a preset expense details table in a preset electronic medical record;

[0200] The feature vector of the preset medical information table, the feature vector of the preset settlement information table, and the feature vector of the preset expense details table are concatenated to generate a preset feature vector;

[0201] The preset feature vector is input into the first learning model in the training process, and a prediction label output by the first learning model in the training process based on the preset feature vector is obtained.

[0202] In some embodiments, the processor 20 is configured to implement:

[0203] Training the second learning model based on the second training set, obtaining the predicted medical insurance catalog code output by the second learning model during the training process, and obtaining a second loss value between the predicted medical insurance catalog code and the preset medical insurance catalog code;

[0204] When the second loss value is less than a preset loss value, outputting the trained second learning model, and obtaining second evaluation data of the trained second learning model on the second test set, wherein the second evaluation data includes a second accuracy rate, a second precision rate, a second recall rate, and a second score;

[0205] The second learning model whose second accuracy is greater than a preset accuracy, whose second precision is greater than a preset precision, whose second recall is greater than a preset recall, and whose second score is greater than a preset score is set as the second learning model that passes the evaluation.

[0206] In some embodiments, the processor 20 is configured to implement:

[0207] Training the third learning model based on the third training set, and obtaining predicted illegal content output by the third learning model during the training process;

[0208] Obtaining a third loss value between the predicted illegal content and the preset illegal content;

[0209] When the third loss value is less than the preset loss value, output the trained third learning model, and obtain third evaluation data of the trained third learning model on the third test set, wherein the third evaluation data includes a third accuracy rate, a third precision rate, a third recall rate, and a third score;

[0210] The third learning model whose third accuracy is greater than the preset accuracy, whose third precision is greater than the preset precision, whose third recall is greater than the preset recall, and whose third score is greater than the preset score is set as the third learning model that passes the evaluation.

[0211] In some embodiments, the processor 20 is configured to implement:

[0212] Obtaining a current label output by the first learning model that has passed the evaluation based on the current feature vector, and when the current label is the violation label, obtaining a current violation sequence number of the current electronic medical record;

[0213] Input the current violation serial number into the second learning model that has passed the evaluation, and obtain the current medical insurance catalog code recognition result output by the second learning model that has passed the evaluation based on the current violation serial number;

[0214] Input the current medical insurance catalog code recognition result into the evaluated third learning model, and obtain the current violation category recognition result output by the evaluated third learning model based on the current medical insurance catalog code recognition result;

[0215] The violation label, the current violation serial number, the current medical insurance catalog code identification result, and the current violation category identification result are combined to generate a violation identification result of the current electronic medical record.

[0216] In some embodiments, the processor 20 is configured to implement:

[0217] A display area of ​​a display page is obtained, and the violation identification result is displayed through the display area.

[0218] The processor 20 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0219] In some embodiments, the memory 21 may be an internal storage unit of the electronic device 2, such as a hard disk or memory of the electronic device 2. In other embodiments, the memory 21 may also be an external storage device of the electronic device 2, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the electronic device 2. Further, the memory 21 may also include both an internal storage unit and an external storage device of the electronic device 2. The memory 21 is used to store an operating system, an application program, a boot loader, data, and other programs, such as the program code of the computer program. The memory 21 may also be used to temporarily store data that has been output or is to be output.

[0220] It should be noted that the information interaction, execution process, etc. between the above-mentioned devices / units are based on the same concept as the method embodiment of the present application. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.

[0221] An embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned method embodiments can be implemented.

[0222] The computer-readable storage medium stores program codes, which can be called by a processor to execute the violation identification result acquisition method described in the above method embodiment.

[0223] The computer-readable storage medium has a storage space for program codes.

[0224] The program code includes the code of any step in the method for obtaining violation identification results described in the above method embodiment.

[0225] For example, the program code is called by the processor and can execute the following steps:

[0226] Acquire compliant electronic medical records, combine the compliant electronic medical records and the compliant labels corresponding to the compliant electronic medical records into compliant samples, acquire illegal electronic medical records, combine the illegal electronic medical records and the illegal labels corresponding to the illegal electronic medical records into illegal samples, combine a plurality of the compliant samples and a plurality of the illegal samples into a data set, and divide the data set into a first training set and a first test set;

[0227] Training the first learning model based on the first training set, obtaining first evaluation data of the first learning model on the first test set, and determining the first learning model that passes the evaluation according to the first evaluation data and a predefined first evaluation method;

[0228] Obtaining a preset violation sequence number identified by the first learning model through evaluation in the first training set, using the identified preset violation sequence number as a target sequence number, combining the target sequence number and a preset medical insurance catalog code corresponding to the target sequence number into a second sample, and constructing a second training set and a second test set based on different second samples;

[0229] Training the second learning model based on the second training set, obtaining second evaluation data of the second learning model on the second test set, and determining the second learning model that passes the evaluation according to the second evaluation data and a predefined second evaluation method;

[0230] Obtain the preset medical insurance catalog code identified by the second learning model through evaluation in the second training set, set the identified preset medical insurance catalog code as the target code, form a third sample with the target sequence number, the target code and the preset violation category, and construct a third training set and a third test set based on different third samples;

[0231] Training the third learning model based on the third training set, obtaining third evaluation data of the third learning model on the third test set, and determining the third learning model that passes the evaluation according to the third evaluation data and a predefined third evaluation method;

[0232] A violation identification result of the current electronic medical record is generated according to the first learning model that has passed the evaluation, the second learning model that has passed the evaluation, the third learning model that has passed the evaluation, and a predefined generation method.

[0233] The specific implementation of the above operations can be found in the previous embodiments, which will not be described in detail here.

[0234] Among them, the computer-readable storage medium can also be an external storage device of the violation identification result acquisition device or electronic device, for example, a plug-in hard disk equipped on the violation identification result acquisition device or electronic device, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), a non-volatile computer-readable medium (non-transitory computer-readable storage medium), etc.

[0235] Since the computer program stored in the computer-readable storage medium can execute any one of the violation identification result acquisition methods provided in the embodiments of the present application, the computer-readable storage medium can achieve the beneficial effects that can be achieved by any one of the violation identification result acquisition methods provided in the embodiments of the present application. For details, please refer to the previous embodiments and will not be repeated here.

[0236] An embodiment of the present application provides a computer program product. When the computer program product is run on an electronic device, the electronic device executes the above-mentioned violation identification result acquisition method.

[0237] The computer program product is loaded into the electronic device and can execute the following steps:

[0238] Acquire compliant electronic medical records, combine the compliant electronic medical records and the compliant labels corresponding to the compliant electronic medical records into compliant samples, acquire illegal electronic medical records, combine the illegal electronic medical records and the illegal labels corresponding to the illegal electronic medical records into illegal samples, combine a plurality of the compliant samples and a plurality of the illegal samples into a data set, and divide the data set into a first training set and a first test set;

[0239] Training the first learning model based on the first training set, obtaining first evaluation data of the first learning model on the first test set, and determining the first learning model that passes the evaluation according to the first evaluation data and a predefined first evaluation method;

[0240] Obtaining a preset violation sequence number identified by the first learning model through evaluation in the first training set, using the identified preset violation sequence number as a target sequence number, combining the target sequence number and a preset medical insurance catalog code corresponding to the target sequence number into a second sample, and constructing a second training set and a second test set based on different second samples;

[0241] Training the second learning model based on the second training set, obtaining second evaluation data of the second learning model on the second test set, and determining the second learning model that passes the evaluation according to the second evaluation data and a predefined second evaluation method;

[0242] Obtain the preset medical insurance catalog code identified by the second learning model through evaluation in the second training set, set the identified preset medical insurance catalog code as the target code, form a third sample with the target sequence number, the target code and the preset violation category, and construct a third training set and a third test set based on different third samples;

[0243] Training the third learning model based on the third training set, obtaining third evaluation data of the third learning model on the third test set, and determining the third learning model that passes the evaluation according to the third evaluation data and a predefined third evaluation method;

[0244] A violation identification result of the current electronic medical record is generated according to the first learning model that has passed the evaluation, the second learning model that has passed the evaluation, the third learning model that has passed the evaluation, and a predefined generation method.

[0245] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium.

[0246] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0247] Based on such understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, which can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of each of the above-mentioned method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may at least include: any entity or device that can carry the computer program code to an electronic device, a recording medium, a computer memory, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), an electrical carrier signal, a telecommunication signal, and a software distribution medium. For example, a USB flash drive, a mobile hard disk, a disk or an optical disk.

[0248] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0249] The above are only preferred embodiments of the present application, and are not intended to limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for obtaining violation identification results, characterized in that: Applied to an electronic device, the electronic device stores a first learning model, a second learning model, and a third learning model, and the method for obtaining the violation identification result includes: Acquire compliant electronic medical records, combine the compliant electronic medical records and the compliant labels corresponding to the compliant electronic medical records into compliant samples, acquire illegal electronic medical records, combine the illegal electronic medical records and the illegal labels corresponding to the illegal electronic medical records into illegal samples, combine a plurality of the compliant samples and a plurality of the illegal samples into a data set, and divide the data set into a first training set and a first test set; Training the first learning model based on the first training set, obtaining first evaluation data of the first learning model on the first test set, and determining the first learning model that passes the evaluation according to the first evaluation data and a predefined first evaluation method; Obtaining a preset violation sequence number identified by the first learning model through evaluation in the first training set, using the identified preset violation sequence number as a target sequence number, combining the target sequence number and a preset medical insurance catalog code corresponding to the target sequence number into a second sample, and constructing a second training set and a second test set based on different second samples; Training the second learning model based on the second training set, obtaining second evaluation data of the second learning model on the second test set, and determining the second learning model that passes the evaluation according to the second evaluation data and a predefined second evaluation method; Obtain the preset medical insurance catalog code identified by the second learning model through evaluation in the second training set, set the identified preset medical insurance catalog code as the target code, form a third sample with the target sequence number, the target code and the preset violation category, and construct a third training set and a third test set based on different third samples; Training the third learning model based on the third training set, obtaining third evaluation data of the third learning model on the third test set, and determining the third learning model that passes the evaluation according to the third evaluation data and a predefined third evaluation method; A violation identification result of the current electronic medical record is generated according to the first learning model that has passed the evaluation, the second learning model that has passed the evaluation, the third learning model that has passed the evaluation, and a predefined generation method.

2. The method for obtaining violation identification results according to claim 1, characterized in that: The step of training the first learning model based on the first training set, obtaining first evaluation data of the first learning model on the first test set, and determining that the first learning model passes the evaluation according to the first evaluation data and a predefined first evaluation method includes: Training the first learning model based on the first training set, and obtaining a prediction label output by the first learning model during the training process; Obtaining a first loss value between the predicted label and the preset violation label; When the first loss value is less than a preset loss value, outputting the first learning model after training, and obtaining first evaluation data of the first learning model after training on the first test set, wherein the first evaluation data includes a first accuracy rate, a first precision rate, a first recall rate, and a first score; The first learning model whose first accuracy is greater than a preset accuracy, whose first precision is greater than a preset precision, whose first recall is greater than a preset recall, and whose first score is greater than a preset score is set as the first learning model that passes the evaluation.

3. The method for obtaining violation identification results according to claim 2, characterized in that: The step of training the first learning model based on the first training set and obtaining a prediction label output by the first learning model during the training process includes: The first learning model is trained based on the first training set to obtain a preset medical consultation information table, a preset settlement information table, and a preset expense details table in a preset electronic medical record; The feature vector of the preset medical information table, the feature vector of the preset settlement information table, and the feature vector of the preset expense details table are concatenated to generate a preset feature vector; The preset feature vector is input into the first learning model in the training process, and a prediction label output by the first learning model in the training process based on the preset feature vector is obtained.

4. The method for obtaining violation identification results according to claim 1, characterized in that: The step of training the second learning model based on the second training set, obtaining second evaluation data of the second learning model on the second test set, and determining that the second learning model passes the evaluation according to the second evaluation data and a predefined second evaluation method includes: Training the second learning model based on the second training set, obtaining the predicted medical insurance catalog code output by the second learning model during the training process, and obtaining a second loss value between the predicted medical insurance catalog code and the preset medical insurance catalog code; When the second loss value is less than a preset loss value, outputting the trained second learning model, and obtaining second evaluation data of the trained second learning model on the second test set, wherein the second evaluation data includes a second accuracy rate, a second precision rate, a second recall rate, and a second score; The second learning model whose second accuracy is greater than a preset accuracy, whose second precision is greater than a preset precision, whose second recall is greater than a preset recall, and whose second score is greater than a preset score is set as the second learning model that passes the evaluation.

5. The method for obtaining violation identification results according to claim 1, characterized in that: The step of training the third learning model based on the third training set, obtaining third evaluation data of the third learning model on the third test set, and determining the third learning model that passes the evaluation according to the third evaluation data and a predefined third evaluation method, includes: Training the third learning model based on the third training set, and obtaining predicted illegal content output by the third learning model during the training process; Obtaining a third loss value between the predicted illegal content and the preset illegal content; When the third loss value is less than the preset loss value, output the trained third learning model, and obtain third evaluation data of the trained third learning model on the third test set, wherein the third evaluation data includes a third accuracy rate, a third precision rate, a third recall rate, and a third score; The third learning model whose third accuracy is greater than the preset accuracy, whose third precision is greater than the preset precision, whose third recall is greater than the preset recall, and whose third score is greater than the preset score is set as the third learning model that passes the evaluation.

6. The method for obtaining violation identification results according to claim 1, characterized in that: The generating of the violation identification result of the current electronic medical record according to the first learning model that has passed the evaluation, the second learning model that has passed the evaluation, the third learning model that has passed the evaluation, and a predefined generation method includes: Obtaining a current label output by the first learning model that has passed the evaluation based on the current feature vector, and when the current label is the violation label, obtaining a current violation sequence number of the current electronic medical record; Input the current violation serial number into the second learning model that has passed the evaluation, and obtain the current medical insurance catalog code recognition result output by the second learning model that has passed the evaluation based on the current violation serial number; Input the current medical insurance catalog code recognition result into the evaluated third learning model, and obtain the current violation category recognition result output by the evaluated third learning model based on the current medical insurance catalog code recognition result; The violation label, the current violation serial number, the current medical insurance catalog code identification result, and the current violation category identification result are combined to generate a violation identification result of the current electronic medical record.

7. The method for obtaining violation identification results according to any one of claims 1 to 6, characterized in that: After generating the violation identification result of the current electronic medical record according to the first learning model that has passed the evaluation, the second learning model that has passed the evaluation, the third learning model that has passed the evaluation, and a predefined generation method, the violation identification result acquisition method includes: A display area of ​​a display page is obtained, and the violation identification result is displayed through the display area.

8. A device for obtaining violation identification results, characterized in that: Applied to an electronic device, the electronic device stores a first learning model, a second learning model, and a third learning model, including: A first acquisition module is used to acquire compliant electronic medical records, combine the compliant electronic medical records and the compliant labels corresponding to the compliant electronic medical records into compliant samples, acquire illegal electronic medical records, combine the illegal electronic medical records and the illegal labels corresponding to the illegal electronic medical records into illegal samples, combine multiple compliant samples and multiple illegal samples into a data set, and divide the data set into a first training set and a first test set; A first evaluation module is used to train the first learning model based on the first training set, obtain first evaluation data of the first learning model on the first test set, and determine the first learning model that passes the evaluation according to the first evaluation data and a predefined first evaluation method; A second acquisition module is used to acquire a preset violation sequence number identified by the first learning model through evaluation in the first training set, use the identified preset violation sequence number as a target sequence number, form a second sample with the target sequence number and a preset medical insurance catalog code corresponding to the target sequence number, and construct a second training set and a second test set based on different second samples; A second evaluation module is used to train the second learning model based on the second training set, obtain second evaluation data of the second learning model on the second test set, and determine the second learning model that passes the evaluation according to the second evaluation data and a predefined second evaluation method; A construction module is used to obtain the preset medical insurance catalog code identified by the second learning model through evaluation in the second training set, set the identified preset medical insurance catalog code as the target code, form a third sample with the target sequence number, the target code and the preset violation category, and construct a third training set and a third test set based on different third samples; A third evaluation module is used to train the third learning model based on the third training set, obtain third evaluation data of the third learning model on the third test set, and determine the third learning model that passes the evaluation according to the third evaluation data and a predefined third evaluation method; A generation module is used to generate a violation identification result of the current electronic medical record according to the first learning model that has passed the evaluation, the second learning model that has passed the evaluation, the third learning model that has passed the evaluation, and a predefined generation method.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method for obtaining violation identification results as described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method for obtaining violation identification results as described in any one of claims 1 to 7 is implemented.

Images