A method, device and storage medium for detecting the rationality of a treatment project
By using a method based on word vector transformation and Euclidean distance calculation, the problem of inaccurate detection of the rationality of treatment items is solved, and a more accurate judgment of the rationality of treatment items is achieved, supporting the accuracy of medical insurance claims.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- PING AN TECH (SHENZHEN) CO LTD
- Filing Date
- 2023-04-19
- Publication Date
- 2026-07-21
AI Technical Summary
Current technology cannot accurately detect the rationality of treatment items, making it impossible to accurately identify reasonable and unreasonable treatment items during the medical insurance claims process.
By using pre-collected sample disease and treatment names, word vector transformation and Euclidean distance calculation are employed to determine the Euclidean distance threshold between treatment items and diseases, thereby assessing the rationality of treatment items.
It improves the accuracy of testing the rationality of treatment items, enabling it to accurately distinguish between reasonable and unreasonable treatment items, and supporting the precision of the medical insurance claims process.
Smart Images
Figure CN116823491B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the fields of digital healthcare and computer technology, and in particular to a method, apparatus, and storage medium for detecting the rationality of treatment programs. Background Technology
[0002] With the development of modern society, the coverage rate of medical insurance is increasing. The medical insurance system generally follows the principle of "treatment first, reimbursement later," processing claims and reimbursements based on the medical insurance data submitted by the customer. Analyzing the medical insurance data to identify unreasonable treatment items and then processing claims for reasonable treatment items is crucial in the entire claims process.
[0003] However, the rationality testing for treatment programs is not accurate enough, and it cannot accurately detect which treatment programs are rational and which are irrational. Summary of the Invention
[0004] In view of this, the present invention provides a method for detecting the rationality of treatment programs, the main purpose of which is to solve the current problem of being unable to accurately detect irrational treatment programs.
[0005] To address the above problems, this application provides a method for detecting the rationality of a treatment program, comprising:
[0006] Based on the names of several sample diseases and the names of several sample treatment items corresponding to each sample disease obtained in advance, the Euclidean distance threshold corresponding to each sample disease is determined.
[0007] Based on the target disease and the target treatment item to be detected, determine the target Euclidean distance between the target treatment item and the target disease;
[0008] Based on the target Euclidean distance and the target Euclidean distance threshold corresponding to the target disease, the target treatment item is determined to be a reasonable treatment item or an unreasonable treatment item, so as to obtain the target detection result.
[0009] Optionally, the step of determining the Euclidean distance threshold corresponding to each sample disease based on the names of several sample diseases and the names of several sample treatment items corresponding to each sample disease, obtained in advance, specifically includes:
[0010] Based on the pre-trained target word vector model, word vector conversion is performed on the name of each sample disease and the name of the sample treatment item corresponding to the sample disease to obtain the sample disease word vector and the word vector of several sample treatment items corresponding to the sample disease.
[0011] Based at least on the sample disease word vector and the word vector of each sample treatment item corresponding to the sample disease, determine the first Euclidean distance between each treatment item and the sample disease;
[0012] Based on the first Euclidean distance corresponding to each disease sample, determine the Euclidean distance threshold corresponding to each disease sample.
[0013] Optionally, determining the target Euclidean distance between the target treatment and the target disease based on the target disease and the target treatment to be detected specifically includes:
[0014] Based on the pre-trained target word vector model, word vector conversion is performed on the disease name of the target disease and the project name of the target treatment project to obtain the target disease word vector and the target treatment project word vector corresponding to the target disease.
[0015] Based on the word vectors of the target disease and the target treatment item, the target Euclidean distance between the target treatment item and the target disease is calculated.
[0016] Optionally, determining the target Euclidean distance between the target treatment and the target disease based on the target disease to be detected specifically includes:
[0017] Based on the target disease, a preset first mapping relationship is found, and a target first distance coefficient corresponding to the target disease is obtained;
[0018] Based on the target treatment item, a preset second mapping relationship is found to obtain the target second distance coefficient corresponding to the target treatment item;
[0019] Based on the target disease word vector, the target treatment item word vector, the target first distance coefficient, and the target second distance coefficient, the target Euclidean distance between the target treatment item and the target disease is calculated.
[0020] Optionally, before finding the first mapping relationship and / or the second mapping relationship, the method further includes:
[0021] Based on the word vectors of each sample disease and the word vectors of each sample treatment item, the first distance coefficient corresponding to each sample disease and the second distance coefficient corresponding to each sample treatment item are calculated respectively.
[0022] Based on each sample disease and its corresponding first distance coefficient, a first mapping relationship containing the correspondence between diseases and the first distance coefficient is constructed;
[0023] Based on the treatment items for each sample and the corresponding second distance coefficient, a second mapping relationship is constructed that includes the correspondence between treatment items and the second distance coefficient.
[0024] Optionally, the step of calculating the target Euclidean distance between the target treatment item and the target disease based on the target disease word vector, the target treatment item word vector, the target first distance coefficient, and the target second distance coefficient specifically includes:
[0025] Based on the word vectors of the target disease and the word vectors of the target treatment items, calculate the initial Euclidean distance between the target treatment items and the target disease.
[0026] Based on the initial Euclidean distance, the first distance coefficient, and the second distance coefficient, the target Euclidean distance between the target treatment item and the target disease is calculated using a predetermined target calculation formula.
[0027] Optionally, determining whether a target treatment item is a reasonable or unreasonable treatment item based on the target Euclidean distance and the target Euclidean distance threshold corresponding to the target disease specifically includes:
[0028] Compare the target Euclidean distance with the target Euclidean distance threshold;
[0029] If the target Euclidean distance is determined to be less than or equal to the target Euclidean distance threshold, the target treatment item is determined to be a reasonable treatment item;
[0030] If the target Euclidean distance is determined to be greater than the target Euclidean distance threshold, the target treatment item is determined to be an unreasonable treatment item.
[0031] Optionally, the method further includes pre-training the target word vector conversion model, specifically including:
[0032] Based on the names of the diseases in each sample and the names of several corresponding treatment items, the initial word vector conversion model is used to convert the names of the disease and treatment items to obtain the initial word vectors of the disease and treatment items.
[0033] Based on the word vectors of the diseases in each initial sample and the word vectors of the treatment items in each initial sample, several current sample treatment items corresponding to each sample disease are determined.
[0034] Based on the differences between the label treatment items corresponding to each sample disease and the current treatment items, the model parameters in the initial word vector conversion model are adjusted until the training conditions are met, at which point training stops, and the trained target word vector conversion model is obtained.
[0035] To address the aforementioned problems, this application provides an apparatus for detecting the rationality of a treatment program, comprising:
[0036] The first determining module is used to determine the Euclidean distance threshold corresponding to each sample disease based on the names of several sample diseases obtained in advance and the names of several sample treatment items corresponding to each sample disease.
[0037] The second determining module is used to determine the target Euclidean distance between the target treatment item and the target disease based on the target disease and the target treatment item to be detected.
[0038] The detection module is used to determine whether the target treatment item is a reasonable treatment item or an unreasonable treatment item based on the target Euclidean distance and the target Euclidean distance threshold corresponding to the target disease, so as to obtain the target detection result.
[0039] To address the aforementioned problems, this application provides a storage medium storing a computer program that, when executed by a processor, implements the steps of the method for detecting the rationality of a treatment program as described above.
[0040] To address the aforementioned problems, this application provides an electronic device, comprising at least a memory and a processor, wherein the memory stores a computer program, and the processor, when executing the computer program in the memory, implements the steps of the method for detecting the rationality of a treatment program as described above.
[0041] The method, apparatus, and storage medium for detecting the rationality of treatment items in this application, by determining the target Euclidean distance threshold corresponding to the disease, can subsequently accurately judge the target Euclidean distance between the target treatment item and the target disease based on the target Euclidean distance threshold, thereby realizing the judgment of whether the target treatment item is rational based on Euclidean distance and improving the accuracy of the judgment results.
[0042] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and in order to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description
[0043] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:
[0044] Figure 1 This is a flowchart illustrating a method for detecting the rationality of a treatment program according to an embodiment of this application;
[0045] Figure 2 This is a structural block diagram of an apparatus for detecting the rationality of a treatment program, according to another embodiment of this application.
[0046] Figure 3 This is a structural block diagram of an electronic device according to another embodiment of this application. Detailed Implementation
[0047] Various embodiments and features of this application are described herein with reference to the accompanying drawings.
[0048] It should be understood that various modifications can be made to the embodiments described herein. Therefore, the above description should not be considered as limiting, but merely as an example of embodiments. Other modifications within the scope and spirit of this application will be apparent to those skilled in the art.
[0049] The accompanying drawings, which are included in and form part of this specification, illustrate embodiments of the present application and, together with the general description of the present application given above and the detailed description of the embodiments given below, serve to explain the principles of the present application.
[0050] These and other features of this application will become apparent from the following description of preferred forms of embodiments given as non-limiting examples, with reference to the accompanying drawings.
[0051] It should also be understood that although this application has been described with reference to some specific examples, those skilled in the art can certainly implement many other equivalent forms of this application.
[0052] The above and other aspects, features and advantages of this application will become more apparent when taken in conjunction with the accompanying drawings and in view of the following detailed description.
[0053] Specific embodiments of this application are described thereafter with reference to the accompanying drawings; however, it should be understood that the claimed embodiments are merely examples of this application, which can be implemented in various ways. Well-known and / or repeated functions and structures are not described in detail to avoid unnecessary or redundant details that could obscure the application. Therefore, the specific structural and functional details claimed herein are not intended to be limiting, but merely serve as the basis and representative basis for the claims to teach those skilled in the art to use this application in a variety of substantially any suitable detailed structures.
[0054] This specification may use the phrases “in one embodiment,” “in another embodiment,” “in yet another embodiment,” or “in other embodiments,” all of which may refer to one or more of the same or different embodiments according to this application.
[0055] This application provides a method for detecting the rationality of a treatment program, which can be specifically applied to electronic devices such as terminals and servers. Figure 1 The method described in this embodiment includes the following steps:
[0056] Step S101: Based on the names of several sample diseases and the names of several sample treatment items corresponding to each sample disease obtained in advance, determine the Euclidean distance threshold corresponding to each sample disease.
[0057] In this step, a target word vector conversion model can be pre-trained. This model is then used to convert the names of each sample disease and each sample treatment item into word vectors, resulting in sample disease word vectors and sample treatment item word vectors. Based on the word vectors of each sample treatment item and the sample disease, the first Euclidean distance between each sample treatment item and each sample disease can be calculated. This allows for the determination of Euclidean distance limits for each sample disease based on these first Euclidean distances. Determining these Euclidean distance limits for each sample disease lays the foundation for accurately determining the appropriateness of the treatment item being tested based on these limits.
[0058] Step S102: Based on the target disease and the target treatment item to be detected, determine the target Euclidean distance between the target treatment item and the target disease;
[0059] In the specific implementation of this step, the target word vector conversion model can be used to convert the name of the target disease and the name of the target treatment item into word vectors to obtain the word vectors of the target disease and the target treatment item. Then, the target Euclidean distance between the target treatment item and the target disease can be calculated using the word vectors of the target disease and the target treatment item.
[0060] Step S103: Based on the target Euclidean distance and the target Euclidean distance threshold corresponding to the target disease, determine whether the target treatment item is a reasonable treatment item or an unreasonable treatment item, so as to obtain the target detection result.
[0061] In the specific implementation process of this step, the target Euclidean distance can be compared with the target Euclidean distance threshold. If the target Euclidean distance is less than or equal to the target Euclidean distance threshold, the target treatment item is determined to be a reasonable treatment item; if the target Euclidean distance is greater than the target Euclidean distance threshold, the target treatment item is determined to be an unreasonable treatment item.
[0062] The method for detecting the rationality of a treatment project in this embodiment determines the target Euclidean distance threshold corresponding to the disease. Subsequently, it can accurately judge the target Euclidean distance between the target treatment project and the target disease based on the target Euclidean distance threshold, thereby realizing the judgment of whether the target treatment project is rational based on Euclidean distance and improving the accuracy of the judgment result.
[0063] This application provides a method for detecting the rationality of a treatment program, including the following steps:
[0064] Step S201: Based on the name of each sample disease and the names of several corresponding label treatment items for each sample disease, the initial word vector conversion model is used to convert the names of each sample disease and the names of each sample treatment item to obtain initial sample disease word vectors and initial sample treatment item word vectors; based on the initial sample disease word vectors and initial sample treatment item word vectors, several current sample treatment items corresponding to each sample disease are determined; based on the differences between the label treatment items corresponding to each sample disease and the current treatment items, the model parameters in the initial word vector conversion model are adjusted until the training conditions are met and training is stopped to obtain the trained target word vector conversion model;
[0065] In this step, treatment data containing sample disease names and treatment items for treating the sample diseases can be obtained from historical claims data. Alternatively, prescription data can be extracted from historical medical records of various hospitals, and then treatment data containing sample disease names and treatment items for treating the sample diseases can be obtained from this prescription data. This yields several sample disease names and their corresponding treatment item names. In this embodiment, treatment items can specifically include drug information, examination or test information, surgical information, etc. For example, if the sample disease name is gastric ulcer obtained from a prescription, the treatment items are drug information and examination information, where the drug information includes omeprazole and magnesium aluminum carbonate, and the examination information is gastroscopy. Similarly, several treatment items corresponding to each sample disease can be extracted from several historical prescription data. Subsequently, a word vector model can be trained based on the names of each sample disease and the corresponding treatment items.
[0066] In this step, when training the word vector model, word vectors can be trained for each disease word unit and treatment item word unit using the Global Vectors for Word Representation (glove) algorithm. The specific process is as follows: The initial word vector conversion model is used to convert the names of each sample disease and each sample treatment item to obtain initial sample disease word vectors and initial sample treatment item word vectors. Specifically, the initial word vector conversion model can be used to convert the names of sample disease A, sample disease B, and sample disease C to obtain the initial sample word vector A' corresponding to sample disease A, the initial sample word vector B' corresponding to sample disease B, and the initial sample word vector C' corresponding to sample disease C. Similarly, the initial word vector conversion model is used to convert the names of sample treatment items a, b, c, and d to obtain the initial sample treatment item word vector a' corresponding to sample treatment item a, b' corresponding to sample treatment item b, c' corresponding to sample treatment item c, and d' corresponding to sample treatment item d. Therefore, based on the initial sample disease word vectors and the initial sample treatment item word vectors, the current sample treatment item corresponding to each sample disease can be determined. For example, the current treatment items corresponding to sample disease A are determined to be sample treatment item a and sample treatment item b; the current treatment items corresponding to sample disease B are determined to be sample treatment item b and sample treatment item c; and the current treatment item corresponding to sample disease C is determined to be sample treatment item d. The labeled treatment items corresponding to sample disease A are treatment item c and treatment item d; the labeled treatment item corresponding to sample disease B is treatment item b; and the labeled treatment item corresponding to sample disease C is treatment item a. It can be determined that the labeled treatment items corresponding to sample diseases A, B, and C differ significantly from the current treatment items, i.e., the accuracy is less than a predetermined threshold. Therefore, the parameters in the initial word vector conversion model can be adjusted until the difference between the determined current treatment items and the labeled treatment items is small, or the accuracy is greater than or equal to the predetermined threshold, or the number of training iterations reaches a predetermined value, thus obtaining the target word vector conversion model.
[0067] In this embodiment, the training conditions can be that the accuracy is greater than a predetermined threshold or the number of training sessions reaches a preset number.
[0068] Step S202: Based on the target word vector model, the names of each sample disease and the names of the sample treatment items corresponding to the sample disease are converted into word vectors to obtain the sample disease word vectors and the word vectors of several sample treatment items corresponding to the sample disease.
[0069] In this step, after obtaining the target word vector model, it can be used to perform word vector conversion on the names of each sample disease and each sample treatment item. Alternatively, during the training of the target word vector model, the word vector conversion results for each sample disease and each sample treatment item can be stored, for example, in a pre-defined storage area. Therefore, when the word vectors for the target disease and the target treatment item are needed, the word vector conversion results for the sample disease and sample treatment item names during the training phase can be directly obtained from the pre-defined storage area.
[0070] Step S203: Determine the first Euclidean distance between each treatment item and the sample disease, based at least on the sample disease word vector and the word vector of each sample treatment item corresponding to the sample disease.
[0071] In the specific implementation of this step, the treatment items corresponding to the target disease can be set as the first treatment items, and the treatment items not corresponding to the target disease can be set as the second treatment items. For example, for target disease A (gastric ulcer), treatment items c (gastroscopy) and d (omeprazole) corresponding to target disease A can be set as the first treatment items, and then treatment items a (radiotherapy) and b (chest imaging) not corresponding to target disease A can be set as the second treatment items. Thus, the first Euclidean distance between the first treatment item c and target disease A, and the first Euclidean distance between the first treatment item b and target disease A can be obtained.
[0072] Step S204: Based on the first Euclidean distances corresponding to the same sample disease, determine the Euclidean distance thresholds corresponding to each sample disease.
[0073] In the specific implementation of this step, to make the determined Euclidean distance threshold more accurate, a second Euclidean distance can be further calculated between each treatment item and the non-corresponding sample disease. That is, the Euclidean distance between the target sample disease and the second sample treatment items that do not correspond to the target sample disease is calculated. Specifically, the second Euclidean distance between second sample treatment item a (radiotherapy) and target sample disease A (gastric ulcer), and the second Euclidean distance between second sample treatment item b (chest imaging examination) and target sample disease A (gastric ulcer) are calculated. Finally, based on each first Euclidean distance and each second Euclidean distance, the Euclidean distance threshold corresponding to target sample disease A is determined. Similarly, the Euclidean distance threshold corresponding to each sample disease can be determined based on several first Euclidean distances and several second Euclidean distances corresponding to each sample disease.
[0074] Step S205: Based on the pre-trained target word vector model, the disease name of the target disease and the project name of the target treatment project are converted into word vectors respectively to obtain the target disease word vector and the target treatment project word vector.
[0075] In this step, the electronic device can respond to the user's detection operations for the target treatment item and the target disease, and use the target word vector model to perform word vector conversion on the name of the target treatment item and the name of the target disease. Specifically, the detection operation can be dragging the icon of the target treatment item to a first predetermined area, dragging the icon of the target disease to a second predetermined area, or entering the name of the target disease in the first predetermined area of the display interface, and entering the name of the target treatment item in the second predetermined area of the display interface, etc.
[0076] Step S206: Based on the word vector of the target disease and the word vector of the target treatment item, calculate the target Euclidean distance between the target treatment item and the target disease;
[0077] Step S207: Compare the target Euclidean distance with the target Euclidean distance threshold; if the target Euclidean distance is less than or equal to the target Euclidean distance threshold, determine the target treatment item as a reasonable treatment item; if the target Euclidean distance is greater than the target Euclidean distance threshold, determine the target treatment item as an unreasonable treatment item.
[0078] The method for detecting the rationality of a treatment project in this embodiment determines the target Euclidean distance threshold corresponding to the disease. Subsequently, it can accurately judge the target Euclidean distance between the target treatment project and the target disease based on the target Euclidean distance threshold, thereby realizing the judgment of whether the target treatment project is rational based on Euclidean distance and improving the accuracy of the judgment result.
[0079] In another embodiment of this application, based on the above embodiments, to make the calculated target Euclidean distance more accurate, a first target distance coefficient corresponding to the target disease and a second target distance coefficient corresponding to the target treatment item can be further obtained. Then, based on the word vector of the target disease, the word vector of the target treatment item, the first distance coefficient, and the second distance coefficient, a final target distance threshold is calculated. Specifically, the process is as follows: based on the target disease, a preset first mapping relationship is found to obtain the first target distance coefficient corresponding to the target disease; based on the target treatment item, a preset second mapping relationship is found to obtain the second target distance coefficient corresponding to the target treatment item; based on the word vector of the target disease, the word vector of the target treatment item, the first target distance coefficient, and the second target distance coefficient, the target Euclidean distance between the target treatment item and the target disease is calculated. That is, by finding the first mapping relationship, a first distance coefficient D_disease(i) corresponding to the target disease A (gastric ulcer) is obtained, and then by finding the second mapping relationship, a second distance coefficient D_item(j) corresponding to the target treatment item c (gastroscopy) is obtained. Then, based on the target disease word vector and the target treatment item word vector, the initial Euclidean distance β between the target treatment item and the target disease is calculated; based on the initial Euclidean distance β, the first distance coefficient D_disease(i) and the second distance coefficient D_item(j), the target Euclidean distance between the target treatment item and the target disease is calculated using a predetermined target calculation formula.
[0080] In this embodiment, the target calculation formula is: β'=0.5*(β / D_disease(i)+β / D_item(j)). Where β represents the initial Euclidean distance, and β' represents the target Euclidean distance. D_disease(i) represents the first distance coefficient corresponding to the disease in the (i)th sample, i.e., the target first distance coefficient corresponding to the target disease; D_item(j) represents the second distance coefficient corresponding to the treatment item in the (j)th sample, i.e., the target second distance coefficient corresponding to the target treatment item.
[0081] In this embodiment, before obtaining the target first distance coefficient and the target second distance coefficient, the method further includes: calculating the first distance coefficient corresponding to each sample disease and the second distance coefficient corresponding to each sample treatment item based on the sample disease word vectors and the sample treatment item word vectors for each sample treatment item; constructing the first mapping relationship containing the correspondence between diseases and the first distance coefficients based on each sample disease and the corresponding first distance coefficients; and finally constructing the second mapping relationship containing the correspondence between treatment items and the second distance coefficients based on each sample treatment item and the corresponding second distance coefficients.
[0082] In other words, for each sample disease, the Euclidean distance between each treatment item in the sample data and the sample disease is calculated. Then, the Euclidean distances corresponding to the same sample disease are weighted and the average Euclidean distance is calculated, thus obtaining the average Euclidean distance corresponding to the sample disease, and thus obtaining the first distance coefficient corresponding to each sample disease. Alternatively, for each sample disease, the first Euclidean distance between each treatment item related to the sample disease and the sample disease is calculated, and the second Euclidean distance between each treatment item unrelated to the sample disease and the sample disease is calculated. Then, the average Euclidean distance is calculated for each of the first and second Euclidean distances corresponding to the same sample disease, thus obtaining the distance coefficient corresponding to the sample disease. For example, in the entire sample data, treatment items related to sample disease A include sample treatment item c and sample treatment item d, while treatment items unrelated to sample disease A include sample treatment item a and sample treatment item b. Therefore, we can calculate the first Euclidean distance 1 between sample disease A and sample treatment item c, the first Euclidean distance 2 between sample disease A and sample treatment item d, the second Euclidean distance 3 between sample disease A and sample treatment item a, and the second Euclidean distance 4 between sample disease A and sample treatment item b. Then, we average these distances to obtain the first distance coefficient corresponding to sample disease A. Similarly, we can calculate the first distance coefficient corresponding to each sample disease.
[0083] Similarly, for each sample treatment item, the Euclidean distance between each sample disease and the sample treatment item in the sample data is calculated. Then, the Euclidean distances corresponding to the same sample treatment item are weighted and the average Euclidean distance is calculated, thus obtaining the average Euclidean distance corresponding to the sample treatment item, and thus obtaining the second distance coefficient corresponding to each sample treatment item. That is, for each sample treatment item, the third Euclidean distance between each sample disease related to the sample treatment item and the sample treatment item is calculated, and the fourth Euclidean distance between each sample disease unrelated to the sample treatment item and the sample treatment item is calculated. Then, the average Euclidean distance of each third and fourth Euclidean distance corresponding to the same sample treatment item is calculated, thus obtaining the distance coefficient corresponding to the sample treatment item. For example, in the entire sample data, sample disease C is related to sample treatment item a, and sample diseases A and B are unrelated to sample treatment item a. Therefore, the third Euclidean distance 5 between sample treatment item a and sample disease C, the fourth Euclidean distance 6 between sample treatment item a and sample disease A, and the fourth Euclidean distance 7 between sample treatment item a and sample disease B can be calculated. Finally, the average of the third Euclidean distance 5, the fourth Euclidean distance 6, and the fourth Euclidean distance 7 is calculated to obtain the second distance coefficient corresponding to sample treatment item a. Similarly, the second distance coefficient corresponding to each sample treatment item can be calculated.
[0084] In this embodiment, after obtaining the second distance coefficient corresponding to each sample treatment item and the first distance coefficient corresponding to each sample disease, a first mapping relationship between each disease and the first distance coefficient, and a second mapping relationship between each treatment item and the second distance coefficient can be established.
[0085] In this embodiment, candidate first distance coefficients can be determined based on each first distance coefficient, and candidate second distance coefficients can be determined based on each second distance coefficient. Thus, when no first distance coefficient corresponding to the target disease can be found in the first mapping relationship, the candidate first distance coefficient can be used as the target first distance coefficient. Similarly, when no second distance coefficient corresponding to the target treatment item can be found in the second mapping relationship, the candidate second distance coefficient can be used as the target second distance coefficient.
[0086] In this embodiment, by obtaining the first target distance coefficient corresponding to the target disease and the second target distance coefficient corresponding to the target treatment item, the target Euclidean distance calculated based on the first and second distance coefficients can be made more accurate, which provides a guarantee for accurately determining whether the target treatment item is reasonable based on the target Euclidean distance.
[0087] The method in this embodiment trains a high-dimensional vector representation of diseases and treatment items using a large amount of medical prescription details data, and realizes the calculation of the word vector distance between diseases and treatment items. Subsequently, the rationality of treatment items under different diseases can be quantified based on word vector distance / Euclidean distance. In the single insurance scenario, it has better generalization performance than traditional rules.
[0088] Another embodiment of this application provides an apparatus for detecting the rationality of a treatment program, such as... Figure 2 As shown, it includes:
[0089] The first determining module 11 is used to determine the Euclidean distance threshold corresponding to each sample disease based on the names of several sample diseases obtained in advance and the names of several sample treatment items corresponding to each sample disease.
[0090] The second determining module 12 is used to determine the target Euclidean distance between the target treatment item and the target disease based on the target disease and the target treatment item to be detected.
[0091] The detection module 13 is used to determine whether the target treatment item is a reasonable treatment item or an unreasonable treatment item based on the target Euclidean distance and the target Euclidean distance threshold corresponding to the target disease, so as to obtain the target detection result.
[0092] In this embodiment, the first determining module is specifically used to: based on the pre-trained target word vector model, perform word vector conversion on the name of each sample disease and the name of the sample treatment item corresponding to the sample disease, respectively, to obtain the sample disease word vector and the word vector of several sample treatment items corresponding to the sample disease; determine the first Euclidean distance between each treatment item and the sample disease based at least on the sample disease word vector and the word vector of each sample treatment item corresponding to the sample disease; and determine the Euclidean distance threshold corresponding to each sample disease based on each of the first Euclidean distances corresponding to the same sample disease.
[0093] In this embodiment, the second determining module is specifically used to: based on the pre-trained target word vector model, perform word vector conversion on the disease name of the target disease and the project name of the target treatment project respectively, to obtain the target disease word vector corresponding to the target disease and the target treatment project word vector corresponding to the target treatment project; and calculate the target Euclidean distance between the target treatment project and the target disease based on the target disease word vector and the target treatment project word vector.
[0094] In this embodiment, the second determining module specifically includes: finding a preset first mapping relationship based on the target disease to obtain a target first distance coefficient corresponding to the target disease; finding a preset second mapping relationship based on the target treatment item to obtain a target second distance coefficient corresponding to the target treatment item; and calculating the target Euclidean distance between the target treatment item and the target disease based on the target disease word vector, the target treatment item word vector, the target first distance coefficient, and the target second distance coefficient.
[0095] In this embodiment, the device for detecting the rationality of treatment items further includes a construction module. The construction module is used to: calculate a first distance coefficient corresponding to each sample disease and a second distance coefficient corresponding to each sample treatment item based on the sample disease word vectors of each sample disease and the sample treatment item word vectors of each sample treatment item; construct a first mapping relationship containing the correspondence between diseases and the first distance coefficients based on each sample disease and the corresponding first distance coefficients; and construct a second mapping relationship containing the correspondence between treatment items and the second distance coefficients based on each sample treatment item and the corresponding second distance coefficients.
[0096] In this embodiment, the second determining module is specifically used to: calculate the initial Euclidean distance between the target treatment item and the target disease based on the target disease word vector and the target treatment item word vector; and calculate the target Euclidean distance between the target treatment item and the target disease based on the initial Euclidean distance, the first distance coefficient, and the second distance coefficient using a predetermined target calculation formula.
[0097] In this embodiment, the detection module is specifically used to: compare the target Euclidean distance with the target Euclidean distance threshold; if the target Euclidean distance is less than or equal to the target Euclidean distance threshold, determine that the target treatment item is a reasonable treatment item; if the target Euclidean distance is greater than the target Euclidean distance threshold, determine that the target treatment item is an unreasonable treatment item.
[0098] In this embodiment, the device for detecting the rationality of treatment items further includes a training module for pre-training the target word vector conversion model. Specifically, the training module is used to: convert the names of sample diseases and corresponding treatment items based on the names of each sample disease and its associated labels using an initial word vector conversion model to obtain initial sample disease word vectors and initial sample treatment item word vectors; determine several current sample treatment items corresponding to each sample disease based on the initial sample disease word vectors and initial sample treatment item word vectors; and adjust the model parameters in the initial word vector conversion model based on the differences between the label treatment items corresponding to each sample disease and the current treatment items, until training is stopped when the training conditions are met, thus obtaining the trained target word vector conversion model.
[0099] The device for detecting the rationality of a treatment project in this embodiment determines the target Euclidean distance threshold corresponding to the disease, and then accurately judges the target Euclidean distance between the target treatment project and the target disease based on the target Euclidean distance threshold. This realizes the judgment of whether the target treatment project is rational based on Euclidean distance, and improves the accuracy of the judgment result.
[0100] Another embodiment of this application provides a storage medium storing a computer program, which, when executed by a processor, implements the following method steps:
[0101] Step 1: Based on the names of several sample diseases and the names of several sample treatment items corresponding to each sample disease obtained in advance, determine the Euclidean distance threshold corresponding to each sample disease.
[0102] Step 2: Based on the target disease and the target treatment item to be detected, determine the target Euclidean distance between the target treatment item and the target disease;
[0103] Step 3: Based on the target Euclidean distance and the target Euclidean distance threshold corresponding to the target disease, determine whether the target treatment item is a reasonable treatment item or an unreasonable treatment item, so as to obtain the target detection result.
[0104] The specific implementation process of the above method steps can be found in any of the above embodiments of the method for detecting the rationality of treatment items, and will not be repeated here.
[0105] In this embodiment, the storage medium determines the target Euclidean distance threshold corresponding to the disease, and subsequently can accurately judge the target Euclidean distance between the target treatment item and the target disease based on the target Euclidean distance threshold. This realizes the judgment of whether the target treatment item is reasonable based on the Euclidean distance, and improves the accuracy of the judgment result.
[0106] Another embodiment of this application provides an electronic device, such as... Figure 3 As shown, it includes at least a memory 1 and a processor 2. The memory 1 stores a computer program, and the processor 2 performs the following method steps when executing the computer program in the memory 1:
[0107] Step 1: Based on the names of several sample diseases and the names of several sample treatment items corresponding to each sample disease obtained in advance, determine the Euclidean distance threshold corresponding to each sample disease.
[0108] Step 2: Based on the target disease and the target treatment item to be detected, determine the target Euclidean distance between the target treatment item and the target disease;
[0109] Step 3: Based on the target Euclidean distance and the target Euclidean distance threshold corresponding to the target disease, determine whether the target treatment item is a reasonable treatment item or an unreasonable treatment item, so as to obtain the target detection result.
[0110] The specific implementation process of the above method steps can be found in any of the above embodiments of the method for detecting the rationality of treatment items, and will not be repeated here.
[0111] The electronic device in this embodiment, by determining the target Euclidean distance threshold corresponding to the disease, can then accurately judge the target Euclidean distance between the target treatment item and the target disease based on the target Euclidean distance threshold, thereby realizing the judgment of whether the target treatment item is reasonable based on the Euclidean distance and improving the accuracy of the judgment result.
[0112] The above embodiments are merely exemplary embodiments of this application and are not intended to limit this application. The scope of protection of this application is defined by the claims. Those skilled in the art can make various modifications or equivalent substitutions to this application within its substance and scope of protection, and such modifications or equivalent substitutions should also be considered to fall within the scope of protection of this application.
Claims
1. A method for detecting the rationality of a treatment program, characterized in that, include: Based on the names of several sample diseases and the names of several sample treatment items corresponding to each sample disease obtained in advance, a Euclidean distance threshold corresponding to each sample disease is determined; the Euclidean distance threshold is used to characterize the distance limit between the sample disease and the treatment item. Based on the target disease and the target treatment item to be detected, determine the target Euclidean distance between the target treatment item and the target disease; Based on the target Euclidean distance and the target Euclidean distance threshold corresponding to the target disease, the target treatment item is determined to be a reasonable treatment item or an unreasonable treatment item, so as to obtain the target detection result; The determination of the target Euclidean distance between the target treatment item and the target disease based on the target disease and the target treatment item to be detected specifically includes: Based on the target disease, a preset first mapping relationship is found, and a target first distance coefficient corresponding to the target disease is obtained; Based on the target treatment item, a preset second mapping relationship is found to obtain the target second distance coefficient corresponding to the target treatment item; Based on the word vectors of the target disease and the word vectors of the target treatment items, calculate the initial Euclidean distance between the target treatment items and the target disease; Based on the initial Euclidean distance, the first target distance coefficient, and the second target distance coefficient, the target Euclidean distance between the target treatment item and the target disease is calculated using a predetermined target calculation formula. Before finding the first mapping relationship and / or the second mapping relationship, the method further includes: Based on the word vectors of each sample disease and the word vectors of each sample treatment item, the first distance coefficient corresponding to each sample disease and the second distance coefficient corresponding to each sample treatment item are calculated respectively. Based on each sample disease and its corresponding first distance coefficient, a first mapping relationship containing the correspondence between diseases and the first distance coefficient is constructed; Based on the treatment items for each sample and the corresponding second distance coefficient, a second mapping relationship is constructed that includes the correspondence between treatment items and the second distance coefficient.
2. The method as described in claim 1, characterized in that, The step of determining the Euclidean distance threshold for each sample disease based on the names of several pre-collected sample diseases and the names of several sample treatment items corresponding to each sample disease specifically includes: Based on the pre-trained target word vector model, word vector conversion is performed on the name of each sample disease and the name of the sample treatment item corresponding to the sample disease to obtain the sample disease word vector and the word vector of several sample treatment items corresponding to the sample disease. Based at least on the sample disease word vector and the word vector of each sample treatment item corresponding to the sample disease, determine the first Euclidean distance between each treatment item and the sample disease; Based on the first Euclidean distance corresponding to each disease sample, determine the Euclidean distance threshold corresponding to each disease sample.
3. The method as described in claim 1, characterized in that, The determination of the target Euclidean distance between the target treatment item and the target disease based on the target disease and the target treatment item to be detected specifically includes: Based on the pre-trained target word vector model, word vector conversion is performed on the disease name of the target disease and the project name of the target treatment project to obtain the target disease word vector and the target treatment project word vector corresponding to the target disease. Based on the word vectors of the target disease and the target treatment item, the target Euclidean distance between the target treatment item and the target disease is calculated.
4. The method as described in claim 1, characterized in that, The process of determining whether a target treatment item is a reasonable or unreasonable treatment item based on the target Euclidean distance and the target Euclidean distance threshold corresponding to the target disease specifically includes: Compare the target Euclidean distance with the target Euclidean distance threshold; If the target Euclidean distance is determined to be less than or equal to the target Euclidean distance threshold, the target treatment item is determined to be a reasonable treatment item; If the target Euclidean distance is determined to be greater than the target Euclidean distance threshold, the target treatment item is determined to be an unreasonable treatment item.
5. The method as described in claim 1, characterized in that, The method also includes pre-training a target word vector model, specifically including: Based on the names of the diseases in each sample and the names of several label treatment items corresponding to each disease, the names of the diseases and treatment items in the samples are transformed using an initial word vector model to obtain the initial word vectors of the diseases and treatment items in the samples. Based on the word vectors of the diseases in each initial sample and the word vectors of the treatment items in each initial sample, several current sample treatment items corresponding to each sample disease are determined. Based on the differences between the label treatment items corresponding to each sample disease and the current treatment items, the model parameters in the initial word vector model are adjusted until the training conditions are met, at which point training stops, and the trained target word vector model is obtained.
6. A device for detecting the rationality of a treatment program, characterized in that, include: The first determining module is used to determine the Euclidean distance threshold corresponding to each sample disease based on the names of several sample diseases and the names of several sample treatment items corresponding to each sample disease, which are collected in advance; the Euclidean distance threshold is used to characterize the distance limit between the sample disease and the treatment item. The second determining module is used to determine the target Euclidean distance between the target treatment item and the target disease based on the target disease and the target treatment item to be detected. The detection module is used to determine whether the target treatment item is a reasonable treatment item or an unreasonable treatment item based on the target Euclidean distance and the target Euclidean distance threshold corresponding to the target disease, so as to obtain the target detection result; The second determining module is specifically used to: find a preset first mapping relationship based on the target disease, and obtain a target first distance coefficient corresponding to the target disease; Based on the target treatment item, a preset second mapping relationship is found to obtain the target second distance coefficient corresponding to the target treatment item; Based on the word vectors of the target disease and the word vectors of the target treatment items, calculate the initial Euclidean distance between the target treatment items and the target disease; Based on the initial Euclidean distance, the first target distance coefficient, and the second target distance coefficient, the target Euclidean distance between the target treatment item and the target disease is calculated using a predetermined target calculation formula. The device for detecting the rationality of a treatment program further includes: a construction module, the construction module being used for: Based on the word vectors of each sample disease and the word vectors of each sample treatment item, a first distance coefficient corresponding to each sample disease and a second distance coefficient corresponding to each sample treatment item are calculated respectively. Based on each sample disease and the corresponding first distance coefficient, a first mapping relationship containing the correspondence between the disease and the first distance coefficient is constructed. Based on each sample treatment item and the corresponding second distance coefficient, a second mapping relationship containing the correspondence between the treatment item and the second distance coefficient is constructed.
7. A storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the steps of the method for detecting the rationality of a treatment program as described in any one of claims 1-5.