CN116823168A

CN116823168A - Abnormal medical item detection method, device, computer equipment and storage medium

Info

Publication number: CN116823168A
Application number: CN202310802879.0A
Authority: CN
Inventors: 吴开源; 徐啸; 刘小双; 郭建影
Original assignee: Ping An Technology Shenzhen Co Ltd
Current assignee: Ping An Technology Shenzhen Co Ltd
Priority date: 2023-06-30
Filing date: 2023-06-30
Publication date: 2023-09-29

Abstract

The application belongs to the field of intelligent decision making, and relates to an abnormal medical item detection method and related equipment, wherein the method comprises the following steps: inputting the current treatment data into a document embedding model to perform vector conversion operation to obtain a document embedding vector to be checked; calculating the similarity of the embedded vector of the document to be checked and each embedded vector of the history document in the set of embedded vectors of the history document; screening the historical document embedded vectors according to the similarity to obtain a similar document embedded vector set; judging whether abnormal medical project fees exist in the current project fees according to the similar document embedded vector set; if the current project expense has abnormal medical project expense, outputting an expense abnormal signal of the abnormal medical project expense. According to the application, the document embedding model based on the natural language model is utilized to search similar cases, and the suspicious unreasonable medical charges are screened for by utilizing the frequency and the cost conditions of medical charge projects in similar claim settlement cases, so that the workload of subsequent rationality judgment can be reduced.

Description

Abnormal medical item detection method, device, computer equipment and storage medium

Technical Field

The present application relates to the field of intelligent decision making technology in artificial intelligence, and in particular, to a method, apparatus, computer device, and storage medium for detecting abnormal medical items

Background

The examination and verification of the medical fee is a key link in insurance claims relating to injury and illness of people, and is important for improving the service quality of insurance companies. However, since the examination of medical fees requires strong professional knowledge, it is difficult for the claimant to examine the rationality of each fee by combining the information data of the patient in the actual claimant process, so as to achieve scientific, reasonable, fair and fair claimant.

At present, the rationality identification method of medical expense in the industry is mainly based on a rule base, namely, reasonable medication management is carried out by utilizing the rule base; the technology of meters provides a diagnosis irrelevant reasonable cost identification method based on multiple strategies in a patent CN_11580961, and the method constructs a collection of high, medium and low specific charge items and a forward and reverse rule to carry out rationality judgment of diagnosis and treatment charge.

However, the applicant found that the conventional rationality recognition method requires manual establishment of a rule base, and cannot benefit complex situations such as multiple cases of diagnosis complex, so that the conventional rationality recognition method has a problem that the complex situation cases cannot be handled.

Disclosure of Invention

The embodiment of the application aims to provide an abnormal medical item detection method, an abnormal medical item detection device, computer equipment and a storage medium, so as to solve the problem that a traditional rationality identification method cannot handle complex situation cases.

In order to solve the above technical problems, the embodiment of the present application provides a method for detecting an abnormal medical item, which adopts the following technical scheme:

receiving current medical treatment information to be audited, wherein the current medical treatment information comprises N current medical treatment items and current item fees corresponding to the current medical treatment items, and N is an integer greater than or equal to 1;

inputting the current diagnosis data into a document embedding model to perform vector conversion operation to obtain a document embedding vector to be checked;

reading a database, and extracting a history document embedded vector set from the database;

calculating the similarity between the to-be-checked document embedded vector and each history document embedded vector in the history document embedded vector set;

screening the historical document embedded vectors according to the similarity to obtain a similar document embedded vector set;

judging whether abnormal medical project fees exist in the current project fees according to the similar document embedded vector set;

If the abnormal medical project expense exists in the current project expense, outputting an expense abnormal signal of the abnormal medical project expense;

and if the abnormal medical project expense does not exist in the current project expense, outputting an expense error-free signal.

In order to solve the above technical problems, the embodiment of the present application further provides an abnormal medical item detection device, which adopts the following technical scheme:

the data receiving module is used for receiving current medical treatment data to be audited, wherein the current medical treatment data comprise N current medical treatment items and current item fees corresponding to the current medical treatment items, and N is an integer greater than or equal to 1;

the vector conversion module is used for inputting the current treatment data into the document embedding model to perform vector conversion operation so as to obtain an embedding vector of the document to be checked;

the historical document acquisition module is used for reading a database and extracting an embedded vector set of the historical document from the database;

the similarity calculation module is used for calculating the similarity between the to-be-checked document embedded vector and each history document embedded vector in the history document embedded vector set;

the screening module is used for carrying out screening operation on the historical document embedded vectors according to the similarity to obtain a similar document embedded vector set;

The abnormal cost judging module is used for judging whether abnormal medical project cost exists in the current project cost according to the similar document embedded vector set;

the cost anomaly module is used for outputting a cost anomaly signal of the abnormal medical project cost if the abnormal medical project cost exists in the current project cost;

and the expense error-free module is used for outputting expense error-free signals if the abnormal medical project expense does not exist in the current project expense.

In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical schemes:

comprising a memory having stored therein computer readable instructions which when executed by a processor implement the steps of the abnormal medical item detection method as described above.

In order to solve the above technical problems, an embodiment of the present application further provides a computer readable storage medium, which adopts the following technical schemes:

the computer readable storage medium has stored thereon computer readable instructions which when executed by a processor implement the steps of the abnormal medical item detection method as described above.

The application provides an abnormal medical project detection method, which comprises the following steps: receiving current medical treatment information to be audited, wherein the current medical treatment information comprises N current medical treatment items and current item fees corresponding to the current medical treatment items, and N is an integer greater than or equal to 1; inputting the current diagnosis data into a document embedding model to perform vector conversion operation to obtain a document embedding vector to be checked; reading a database, and extracting a history document embedded vector set from the database; calculating the similarity between the to-be-checked document embedded vector and each history document embedded vector in the history document embedded vector set; screening the historical document embedded vectors according to the similarity to obtain a similar document embedded vector set; judging whether abnormal medical project fees exist in the current project fees according to the similar document embedded vector set; if the abnormal medical project expense exists in the current project expense, outputting an expense abnormal signal of the abnormal medical project expense; and if the abnormal medical project expense does not exist in the current project expense, outputting an expense error-free signal. Compared with the prior art, the application utilizes the document embedded model based on the natural language model to search similar cases, more comprehensively utilizes the treatment information of medical claim case, and improves the comparability of the searched similar cases and the current claim case. The frequency and cost conditions of medical charging items in similar claim settlement cases are utilized to primarily screen suspicious unreasonable medical charging, so that the workload of subsequent rationality judgment can be reduced.

Drawings

In order to more clearly illustrate the solution of the present application, a brief description will be given below of the drawings required for the description of the embodiments of the present application, it being apparent that the drawings in the following description are some embodiments of the present application, and that other drawings may be obtained from these drawings without the exercise of inventive effort for a person of ordinary skill in the art.

FIG. 1 is an exemplary system architecture diagram in which the present application may be applied;

FIG. 2 is a flowchart of an abnormal medical item detection method according to an embodiment of the present application;

FIG. 3 is a flow chart of an embodiment prior to step S202 in FIG. 2;

FIG. 4 is a flow chart of one embodiment of step S201 in FIG. 2;

FIG. 5 is a flow chart of an embodiment prior to step S404 in FIG. 4;

FIG. 6 is a flow chart of one embodiment of step S503 in FIG. 5;

FIG. 7 is a flow chart of one embodiment of step S206 of FIG. 2;

fig. 8 is a schematic structural diagram of an abnormal medical item detection device according to a second embodiment of the present application;

FIG. 9 is a schematic structural view of one embodiment of a computer device according to the present application.

Detailed Description

Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs; the terminology used in the description of the applications herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the application; the terms "comprising" and "having" and any variations thereof in the description of the application and the claims and the description of the drawings above are intended to cover a non-exclusive inclusion. The terms first, second and the like in the description and in the claims or in the above-described figures, are used for distinguishing between different objects and not necessarily for describing a sequential or chronological order.

Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the application. The appearances of such phrases in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. Those of skill in the art will explicitly and implicitly appreciate that the embodiments described herein may be combined with other embodiments.

In order to make the person skilled in the art better understand the solution of the present application, the technical solution of the embodiment of the present application will be clearly and completely described below with reference to the accompanying drawings.

As shown in fig. 1, a system architecture 100 may include terminal devices 101, 102, 103, a network 104, and a server 105. The network 104 is used as a medium to provide communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, among others.

The user may interact with the server 105 via the network 104 using the terminal devices 101, 102, 103 to receive or send messages or the like. Various communication client applications, such as a web browser application, a shopping class application, a search class application, an instant messaging tool, a mailbox client, social platform software, etc., may be installed on the terminal devices 101, 102, 103.

The terminal devices 101, 102, 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smartphones, tablet computers, electronic book readers, MP3 players (Moving Picture Experts Group Audio Layer III, dynamic video expert compression standard audio plane 3), MP4 (Moving Picture Experts Group Audio Layer IV, dynamic video expert compression standard audio plane 4) players, laptop and desktop computers, and the like.

The server 105 may be a server providing various services, such as a background server providing support for pages displayed on the terminal devices 101, 102, 103.

It should be noted that, the method for detecting an abnormal medical item provided by the embodiment of the present application is generally executed by a server/terminal device, and accordingly, the device for detecting an abnormal medical item is generally disposed in the server/terminal device.

It should be understood that the number of terminal devices, networks and servers in fig. 1 is merely illustrative. There may be any number of terminal devices, networks, and servers, as desired for implementation.

With continued reference to fig. 2, a flow chart of one embodiment of a method of abnormal medical item detection according to the present application is shown. The abnormal medical item detection method comprises the following steps: step S201, step S202, step S203, step S204, step S205, step S206, step S207, and step S208.

In step S201, current medical treatment data to be audited is received, where the current medical treatment data includes N pieces of current medical treatment items and current item fees corresponding to the current medical treatment items, and N is an integer greater than or equal to 1.

In step S202, the current diagnosis data is input to the document embedding model to perform vector conversion operation, so as to obtain the document embedding vector to be checked.

In the embodiment of the application, a document embedding model based on a natural language model technology is firstly utilized for the current claim case needing to be audited, and the doctor seeing data of a user, such as a discharge section, a medical record and other document data (if the document information exists in a picture form, text contents are required to be extracted by an OCR module), and a multidimensional real vector (embedding vector) is embedded in a coding mode, so that a document embedding vector V of the current case is obtained.

In step S203, a database is read, and a history document embedded vector set E is extracted in the database.

In step S204, the similarity of the document embedding vector to be checked and each of the history document embedding vectors in the history document embedding vector set is calculated.

In step S205, a filtering operation is performed on the history document embedded vectors according to the similarity, so as to obtain a similar document embedded vector set.

In the embodiment of the application, each vector V of the vectors V and E is calculated _i Similarity (L) ₂ Similarity, cosine similarity), the greater the similarity is, the more similar the two vectors are, the similarity is sorted from big to small, and the similarity is selected to be not smaller than a threshold t ₁ Or the case of the front top k, ensuring that at least k similar cases can be selected.

In step S206, it is determined whether there is an abnormal medical item fee in the current item fee based on the similar document embedding vector set.

In the embodiment of the application, suspicious charging items are screened, the number of times c of occurrence in a similar case set of each charging item in the current claim case cost details is counted, and the occurrence of each charging item in the phase is countedAverage cost Avg in a set of similar cases _amount Cost variance Std _amount . For times c less than t ₂ The entries are suspicious entries of class 1 for fees exceeding Avg _amount +t ₃ ×Std _amount Is a type 2 suspicious entry. For suspicious entries of class 1, a cost threshold t is further set ₄ Neglecting that the cost is less than the threshold t ₄ Is a term for the entry of (a). The cost in a suspicious entry of class 1 exceeds a threshold t ₄ And the suspicious entries of class 2 form a final suspicious unreasonable charging item list.

In step S207, if there is an abnormal medical item fee in the current item fee, a fee abnormality signal of the abnormal medical item fee is output.

In step S208, if there is no abnormal medical item fee among the current item fees, a fee error-free signal is output.

In practical application, the achievement of a large language model represented by ChatGPT recently can be fully utilized, the reasonable problem of medical toll collection project can be identified through prompt language engineering splicing, and corresponding reasons are given. Helping the claimant to further audit.

In an embodiment of the present application, there is provided an abnormal medical item detection method including: receiving current medical treatment information to be audited, wherein the current medical treatment information comprises N pieces of current medical treatment items and current item fees corresponding to the current medical treatment items, and N is an integer greater than or equal to 1; inputting the current treatment data into a document embedding model to perform vector conversion operation to obtain a document embedding vector to be checked; reading a database, and extracting a history document embedded vector set from the database; calculating the similarity of the embedded vector of the document to be checked and each embedded vector of the history document in the set of embedded vectors of the history document; screening the historical document embedded vectors according to the similarity to obtain a similar document embedded vector set; judging whether abnormal medical project fees exist in the current project fees according to the similar document embedded vector set; if the current project expense has abnormal medical project expense, outputting an expense abnormal signal of the abnormal medical project expense; if the current project expense does not have abnormal medical project expense, outputting an expense error-free signal. Compared with the prior art, the application utilizes the document embedded model based on the natural language model to search similar cases, more comprehensively utilizes the treatment information of medical claim case, and improves the comparability of the searched similar cases and the current claim case. The frequency and cost conditions of medical charging items in similar claim settlement cases are utilized to primarily screen suspicious unreasonable medical charging, so that the workload of subsequent rationality judgment can be reduced.

With continued reference to fig. 3, a flowchart of one embodiment of fig. 2 prior to step S202 is shown, only the portions relevant to the present application being shown for ease of illustration.

In some optional implementations of the present embodiment, before step S202, the method includes: step S301, step S302, and step S303.

In step S301, a history claim case is obtained.

In step S302, the history claim case is input to the document embedding model to perform a vector conversion operation, so as to obtain a history document embedding vector set.

In step S303, the history document embedded vector set is stored in the database.

In the embodiment of the application, historical claim cases are collected, and a document embedding model based on a natural language model technology is utilized to embed a multidimensional real vector (EMBedding vector) into the user's treatment data, such as discharge bars, medical records and other document data (if the treatment data exists in a picture form, text contents are required to be extracted by using an OCR module). And sequentially processing the historical claim settlement cases to obtain a document embedded vector library.

E＝[[ID ₁ ，V ₁ ]，[ID ₂ ，V ₂ ]，...，[ID _n ，V _n ]]

Each embedded vector is provided with an associated case ID, and case claim information such as treatment information, cost details, charge invoice and the like can be searched through the case ID.

With continued reference to fig. 4, a flowchart of one embodiment of step S201 in fig. 2 is shown, only the portions relevant to the present application being shown for ease of illustration.

In some optional implementations of the present embodiment, step S201 specifically includes: step S401, step S402, step S403, step S404, step S405, and step S406.

In step S401, a current visit picture to be audited is received.

In step S402, screenshot template data corresponding to the current visit picture is invoked.

In the embodiment of the application, the local database is pre-stored with screenshot templates corresponding to various doctor-seeing pictures, and screenshot template data corresponding to the current doctor-seeing picture can be called in the local database.

In the embodiment of the application, the screenshot template data is preset with a semantic segmentation area and a template matching area which are divided according to the occupation ratio of the keywords.

In step S403, an image capturing operation is performed on the current diagnosis picture based on the captured template data, so as to obtain a semantic segmentation captured image and a template matching captured image.

In step S404, the semantic segmentation screenshot is input to the image segmentation model to perform semantic segmentation operation, so as to obtain a semantic segmentation field.

In the embodiment of the application, the principle of depth semantic segmentation comprises the following steps:

1) Downsampling+upsampling: convlusion+deconvlusion/Resize;

2) Multi-scale feature fusion: feature point-by-point addition/feature channel dimension splicing;

3) Obtaining a pixel level segment map: judging the category of each pixel point

The depth semantic segmentation algorithm adopts deeplabv3 in deeplab series as an identity card segmentation algorithm, and combines various full-field segmentation data sets of the diagnosis data to realize full-field diagnosis data segmentation. The kernel of the deep series algorithm employs a hole convolution (dlated/Atrous Convolution). The hole convolution is in fact a common convolution kernel with a few holes inserted in between. Hole convolution at different sampling rates can effectively capture multi-scale information. And taking the semantic segmentation area as the input of the model, obtaining a full-field mask diagram of the diagnosis picture in the segmentation model, and finding out a rectangular frame with the maximum outline according to the value of l abe in the mask to obtain a corresponding key field.

In step S405, a template matching screenshot is subjected to a similarity matching operation in the screenshot template data, so as to obtain a template matching field.

In the embodiment of the application, the similarity matching operation refers to searching a part which is most matched (similar) with screenshot template data in the template matching screenshot template data, obtaining a corresponding area diagram through the highest matching position, carrying out gray level binarization on the picture, and searching the largest outline to obtain a largest rectangular frame area, namely an identity card number area.

In step S406, the semantic segmentation field and the template matching field are input to the text recognition model for text recognition operation, so as to obtain the current diagnosis data.

In the embodiment of the application, all key fields of the picture to be treated can be obtained through an identification card recognition algorithm based on the combination of depth semantic segmentation and template matching. And sending each field into a character recognition model to perform character recognition, so as to obtain the final required diagnosis key information.

In the embodiment of the application, the region of interest is acquired by adopting the image interception operation to increase the duty ratio of the target image, so that the segmentation accuracy of the image segmentation model is improved.

With continued reference to fig. 5, a flowchart of one embodiment of fig. 4 prior to step S404 is shown, only the portions relevant to the present application being shown for ease of illustration.

In some optional implementations of the present embodiment, before step S404, the method includes: step S501, step S502, and step S503.

In step S501, a training data set is acquired, the training data set including a plurality of input images, a target object in each input image, and a rectangular region corresponding to the target object in each input image.

In the embodiment of the present application, for the image segmentation model in the foregoing embodiment, the embodiment of the present application further includes a training method for the image segmentation model, which is to be noted that, training of the image segmentation model may be performed in advance according to the acquired training data set, and then, when image segmentation is required each time, the image segmentation model may be performed, without training the image segmentation model each time when image segmentation of the target object is required.

In an embodiment of the present application, the training data set may include a plurality of input images, a target object in each input image, and a rectangular region of the target object in each input image. The input image may be an image including a target object.

In the embodiment of the present application, the number of input images may not be limited. As an optional implementation manner, the number of the input images may be multiple, each input image is labeled with a corresponding target object, and a rectangular area of the target object corresponding to each input image, and the initial model may be trained according to each input image and the target object and the rectangular area labeled for each input image, so as to improve accuracy of the image segmentation model obtained after training.

In step S502, an image segmentation network is acquired, where the image segmentation network includes a first sub-network and a second sub-network, the first sub-network is used for outputting a target object in an image, and the second sub-network is used for outputting a rectangular area corresponding to the target object in the image.

In the embodiment of the application, when the image segmentation model is obtained through training, an image segmentation network can be constructed, wherein the image segmentation network can comprise a first sub-network for outputting a target object in an image and a second sub-network for outputting a rectangular area corresponding to the target object in the image.

In the embodiment of the application, the image segmentation network can be constructed according to the deeplabv3+ semantic image segmentation model. The deeplabv3+ semantic image segmentation model is a deep learning model for image semantic segmentation, and aims to assign semantic tags to each pixel of an input image so as to realize segmentation of a target object in the image. The output of ASPP structure of the deeplabv3+ semantic image segmentation model typically has one output branch, which is used to output the target object.

In the embodiment of the application, when the image segmentation network is constructed according to the deeplabv3+ semantic image segmentation model, another output branch can be led out from the output of the ASPP structure of the original deeplabv3+ network, namely, a second sub-network is led out after the Encoder network, the second sub-network can be a CNN neural network, and the original output branch is used as a first sub-network, so that the construction of the image segmentation model can be completed.

In the embodiment of the application, the Encoder network in the deeplabv3+ semantic image segmentation model is generally provided with an ASPP hole convolution structure so as to extract object information in an image and output the object information to the Decode network. Therefore, the above second sub-network can be led out from the output of the ASPP hole convolution structure, and the second sub-network can be a CNN neural network, so as to output the rectangular area of the target object according to the information output by the ASPP structure.

In the embodiment of the application, when the image segmentation network is constructed based on the deeplabv3+ semantic image segmentation model, the background network part in the deeplabv3+ semantic image segmentation model can be replaced by a mobiletv 2 network in consideration of the large volume of the deeplabv3+ semantic image segmentation model and the large operation amount in operation, which possibly causes the blocking in operation when the deeplabv3+ semantic image segmentation model is applied to mobile terminals such as mobile phones and the like. The mobile network is a lightweight CNN network mainly applied to a mobile terminal, and comprises a depthwise convolution and a pointwise convolution of 1x1 convolution, and the structure separates the spatial correlation and the channel correlation, so that compared with the traditional convolution, the calculated amount and parameters are greatly reduced, the constructed image segmentation network is based on the mobile network, and the image segmentation model obtained by subsequent training can avoid the blocking during the operation when the image segmentation model runs on the mobile terminal.

In step S503, the image segmentation network is trained according to the training data set, and an image segmentation model is obtained.

In the embodiment of the application, the electronic equipment can train the image segmentation network by utilizing the acquired training data set so as to train to obtain the image segmentation model which can realize the output of the target object in the input image and the rectangular area corresponding to the target object according to the input image. The electronic equipment can perform iterative training on the image segmentation network according to the constructed total loss function and the training data set, and finally obtain an image segmentation model through training by utilizing the total loss function.

In the embodiment of the application, in the iterative training process, parameters of the structure of the image segmentation network are continuously changed, and the image segmentation network after the final iterative training can output a result with a smaller total loss function value, so that the parameters of the obtained image segmentation network can output a target object in an input image and a rectangular area corresponding to the target object according to the input image.

With continued reference to fig. 6, a flowchart of one embodiment of step S503 in fig. 5 is shown, only the portions relevant to the present application being shown for ease of illustration.

In some optional implementations of this embodiment, step S503 specifically includes: step S601 and step S602.

In step S601, a loss function of the image segmentation network is acquired, the loss function comprising a cross entropy loss characterizing the first subnetwork and a regression loss of the second subnetwork.

In an embodiment of the present application, the loss function of the image segmentation network may be as follows:

Total_loss＝Segmentation_loss+Detection_loss

in the embodiment of the application, the segment_loss represents the cross entropy loss of the first sub-network, the detection_loss represents the regression loss of the second sub-network, and the total_loss represents the Total loss of the whole image Segmentation network.

In step S602, training the image segmentation network according to the loss function and the training data set by using a back propagation algorithm until the image segmentation network converges, so as to obtain an image segmentation model.

In the embodiment of the application, after the electronic device obtains the total loss function of the result output by the image segmentation network, training can be performed under a tensorflow training frame according to the total loss function and the training data, and the trained image segmentation model can output a mask image of a target object in an input image and a rectangular area corresponding to the target object according to the input image.

In the embodiment of the application, under the tensorflow training framework, model parameters can be trained by using a back propagation algorithm, and gradient descent is used on all parameters so as to minimize the value of a loss function of an image segmentation network on training data. It can be understood that iterative training is repeatedly performed, so that the image segmentation model obtained by final training can output a result (a target object and a rectangular area) according to the input image in the training data set, and the difference between the result and the label (the target object and the rectangular area) of the marked input image is minimum.

In the embodiment of the application, an Adam optimizer can be used for carrying out iterative training on the image segmentation network until the image segmentation network converges, and the converged image segmentation network is stored to obtain a trained image segmentation model. Adam optimizer combines the advantages of two optimization algorithms, adaGra (adaptive gradient) and RMSProp. The first moment estimate (first moment estimate, i.e., the mean value of the gradient) and the second moment estimate (second moment estimate, i.e., the non-centered variance of the gradient) of the gradient are taken into account together to calculate the update step.

In an embodiment of the present application, the image segmentation network convergence (i.e., the termination condition of the iterative training) may include: the number of iterative training reaches the target number; or the value of the total loss function corresponding to the result output by the image segmentation network meets the set condition.

In the embodiment of the application, the convergence condition is that the loss function is as small as possible, the initial learning rate 1e-3 is used, the learning rate decays along with the cosine of the step number, the batch_size=8, and after training 16 epochs, the convergence can be considered to be completed. The batch_size is understood as a batch parameter, the limit value of the batch_size is the total number of samples in the training set, the epoch refers to the number of times of training by using all samples in the training set, and the value of the epoch is colloquially that the whole data set is looped for several times, and 1 epoch is equal to 1 time of training by using all samples in the training set.

In an embodiment of the present application, the value of the total loss function satisfying the setting condition may include: the value of the total loss function is less than the set threshold. Of course, specific setting conditions are not limited.

In the embodiment of the application, the image segmentation model obtained by training can be stored in the local mobile terminal, and the image segmentation model obtained by training can also be stored in a server which is in communication connection with the electronic equipment, so that the storage space of the electronic equipment is reduced, and the operation efficiency of the electronic equipment is improved.

In the embodiment of the application, the image segmentation model can also acquire new training data periodically or aperiodically, and train and update the image segmentation model.

With continued reference to fig. 7, a flowchart of one embodiment of step S206 of fig. 2 is shown, only the portions relevant to the present application being shown for ease of illustration.

In some optional implementations of this embodiment, step S206 specifically includes: step S701 and step S702, step S207 includes: step S703, step S208 includes: step S704.

In step S701, a similar document average cost and a similar document variance cost corresponding to the current medical item are calculated from the similar document embedding vector set, respectively.

In step S702, the current item cost, the similar document average cost, and the similar document variance cost belonging to the same medical item are acquired, respectively, and the size of the sum of the current item cost and the similar document average cost and the similar document variance cost is compared.

In step S703, if the current item cost is greater than the sum of the average cost of similar documents and the variance cost of similar documents, it is confirmed that the current compared medical item belongs to an abnormal medical item cost, and a cost abnormality signal of the current compared medical item is output.

In step S704, if the current item cost is less than or equal to the sum of the average cost of similar documents and the variance cost of similar documents, it is confirmed that the current compared medical item does not belong to the abnormal medical item cost, and a cost error-free signal of the current compared medical item is output.

It should be emphasized that, to further ensure the privacy and security of the current visit data, the current visit data may also be stored in a node of a blockchain.

The blockchain is a novel application mode of computer technologies such as distributed data storage, point-to-point transmission, consensus mechanism, encryption algorithm and the like. The Blockchain (Blockchain), which is essentially a decentralised database, is a string of data blocks that are generated by cryptographic means in association, each data block containing a batch of information of network transactions for verifying the validity of the information (anti-counterfeiting) and generating the next block. The blockchain may include a blockchain underlying platform, a platform product services layer, an application services layer, and the like.

The embodiment of the application can acquire and process the related data based on the artificial intelligence technology. Among these, artificial intelligence (Artificial Intelligence, AI) is the theory, method, technique and application system that uses a digital computer or a digital computer-controlled machine to simulate, extend and extend human intelligence, sense the environment, acquire knowledge and use knowledge to obtain optimal results.

Artificial intelligence infrastructure technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technologies, operation/interaction systems, mechatronics, and the like. The artificial intelligence software technology mainly comprises a computer vision technology, a robot technology, a biological recognition technology, a voice processing technology, a natural language processing technology, machine learning/deep learning and other directions. Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by computer readable instructions stored in a computer readable storage medium that, when executed, may comprise the steps of the embodiments of the methods described above. The storage medium may be a nonvolatile storage medium such as a magnetic disk, an optical disk, a Read-Only Memory (ROM), or a random access Memory (Random Access Memory, RAM).

It should be understood that, although the steps in the flowcharts of the figures are shown in order as indicated by the arrows, these steps are not necessarily performed in order as indicated by the arrows. The steps are not strictly limited in order and may be performed in other orders, unless explicitly stated herein. Moreover, at least some of the steps in the flowcharts of the figures may include a plurality of sub-steps or stages that are not necessarily performed at the same time, but may be performed at different times, the order of their execution not necessarily being sequential, but may be performed in turn or alternately with other steps or at least a portion of the other steps or stages.

Example two

With further reference to fig. 8, as an implementation of the method shown in fig. 2 described above, the present application provides an embodiment of an abnormal medical item detection apparatus, which corresponds to the method embodiment shown in fig. 2, and which is particularly applicable to various electronic devices.

As shown in fig. 8, the abnormal medical item detection apparatus 200 of the present embodiment includes: an acquisition module 401, an identification module 402, a calculation module 403, a training module 404, and a processing module 405. Wherein:

The data receiving module 210 is configured to receive current medical treatment data to be audited, where the current medical treatment data includes N current medical items and current item fees corresponding to the current medical items, and N is an integer greater than or equal to 1;

the vector conversion module 220 is configured to input the current diagnosis data to the document embedding model to perform vector conversion operation, so as to obtain an embedding vector of the document to be checked;

a history document acquisition module 230 for reading a database, and extracting a history document embedded vector set from the database;

the similarity calculation module 240 is configured to calculate a similarity between the embedded vector of the document to be checked and each of the embedded vectors of the history documents in the set of embedded vectors of the history documents;

the screening module 250 is configured to perform a screening operation on the historical document embedded vectors according to the similarity, so as to obtain a similar document embedded vector set;

an abnormal fee judging module 260 for judging whether abnormal medical project fees exist in the current project fees according to the similar document embedding vector set;

a cost anomaly module 270, configured to output a cost anomaly signal of the cost of the abnormal medical item if the cost of the abnormal medical item exists in the current item cost;

the expense error-free module 280 is configured to output an expense error-free signal if there is no abnormal medical item expense in the current item expense.

In the present embodiment, there is provided an abnormal medical item detection apparatus 200 including: the data receiving module 210 is configured to receive current medical treatment data to be audited, where the current medical treatment data includes N current medical items and current item fees corresponding to the current medical items, and N is an integer greater than or equal to 1; the vector conversion module 220 is configured to input the current diagnosis data to the document embedding model to perform vector conversion operation, so as to obtain an embedding vector of the document to be checked; a history document acquisition module 230 for reading a database, and extracting a history document embedded vector set from the database; the similarity calculation module 240 is configured to calculate a similarity between the embedded vector of the document to be checked and each of the embedded vectors of the history documents in the set of embedded vectors of the history documents; the screening module 250 is configured to perform a screening operation on the historical document embedded vectors according to the similarity, so as to obtain a similar document embedded vector set; an abnormal fee judging module 260 for judging whether abnormal medical project fees exist in the current project fees according to the similar document embedding vector set; a cost anomaly module 270, configured to output a cost anomaly signal of the cost of the abnormal medical item if the cost of the abnormal medical item exists in the current item cost; the expense error-free module 280 is configured to output an expense error-free signal if there is no abnormal medical item expense in the current item expense. Compared with the prior art, the application utilizes the document embedded model based on the natural language model to search similar cases, more comprehensively utilizes the treatment information of medical claim case, and improves the comparability of the searched similar cases and the current claim case. The frequency and cost conditions of medical charging items in similar claim settlement cases are utilized to primarily screen suspicious unreasonable medical charging, so that the workload of subsequent rationality judgment can be reduced.

In some optional implementations of this embodiment, the abnormal medical item detection apparatus 200 further includes: the system comprises a history case acquisition module, a history case vector conversion module and a history document storage module, wherein:

the historical case acquisition module is used for acquiring historical claim case;

the historical case vector conversion module is used for inputting the historical claim case into the document embedding model to perform vector conversion operation to obtain a historical document embedding vector set;

and the history document storage module is used for storing the history document embedded vector set into the database.

In some optional implementations of this embodiment, the data receiving module 210 includes:

the picture acquisition sub-module is used for receiving a current treatment picture to be audited;

the template acquisition sub-module is used for calling screenshot template data corresponding to the current treatment picture;

the image interception sub-module is used for carrying out image interception operation on the current treatment picture based on the screenshot template data to obtain semantic segmentation screenshot and template matching screenshot;

the semantic segmentation sub-module is used for inputting the semantic segmentation screenshot into the image segmentation model to perform semantic segmentation operation to obtain semantic segmentation fields;

The similarity matching sub-module is used for performing similarity matching operation on the template matching screenshot in the screenshot template data to obtain a template matching field;

the character recognition sub-module is used for inputting the semantic segmentation field and the template matching field into the character recognition model to perform character recognition operation, so as to obtain the current treatment data.

In some optional implementations of this embodiment, the data receiving module 210 further includes: training set acquisition module, segmentation network acquisition module and network training module.

Wherein:

the training set acquisition module is used for acquiring a training data set, wherein the training data set comprises a plurality of input images, target objects in each input image and rectangular areas corresponding to the target objects in each input image;

the image segmentation network comprises a first sub-network and a second sub-network, wherein the first sub-network is used for outputting a target object in an image, and the second sub-network is used for outputting a rectangular area corresponding to the target object in the image;

and the network training module is used for training the image segmentation network according to the training data set to obtain an image segmentation model.

In some optional implementations of this embodiment, the network training module specifically includes: the loss function acquisition sub-module and the network training sub-module. Wherein:

a loss function obtaining sub-module, configured to obtain a loss function of the image segmentation network, where the loss function includes a cross entropy loss used to characterize the first sub-network and a regression loss of the second sub-network;

and the network training sub-module is used for training the image segmentation network by utilizing a back propagation algorithm according to the loss function and the training data set until the image segmentation network converges, so as to obtain an image segmentation model.

In some optional implementations of this embodiment, the abnormal charge determination module 260 includes: fei Yongji operator module and expense comparison submodule, the above-mentioned expense exception module includes: a cost anomaly submodule, said cost error-free submodule comprising: a cost-effective sub-module, wherein:

fei Yongji operator module for calculating average cost of similar document and variance cost of similar document corresponding to current medical project according to the embedded vector set of similar document;

the charge comparison sub-module is used for respectively acquiring the current project charge, the average charge of the similar document and the variance charge of the similar document belonging to the same medical project, and comparing the sum of the current project charge and the average charge of the similar document and the variance charge of the similar document;

The cost anomaly submodule is used for confirming that the current compared medical item belongs to the abnormal medical item cost and outputting a cost anomaly signal of the current compared medical item if the current item cost is larger than the sum of the average cost of the similar documents and the variance cost of the similar documents;

and the expense-free sub-module is used for confirming that the currently compared medical item does not belong to abnormal medical item expense if the current item expense is smaller than or equal to the sum of the average expense of the similar documents and the variance expense of the similar documents, and outputting an expense-free signal of the currently compared medical item.

In order to solve the technical problems, the embodiment of the application also provides computer equipment. Referring specifically to fig. 9, fig. 9 is a basic structural block diagram of a computer device according to the present embodiment.

The computer device 6 comprises a memory 61, a processor 62, a network interface 63 communicatively connected to each other via a system bus. It is noted that only computer device 6 having components 61-63 is shown in the figures, but it should be understood that not all of the illustrated components are required to be implemented and that more or fewer components may be implemented instead. It will be appreciated by those skilled in the art that the computer device herein is a device capable of automatically performing numerical calculations and/or information processing in accordance with predetermined or stored instructions, the hardware of which includes, but is not limited to, microprocessors, application specific integrated circuits (Application Specific Integrated Circuit, ASICs), programmable gate arrays (fields-Programmable Gate Array, FPGAs), digital processors (Digital Signal Processor, DSPs), embedded devices, etc.

The computer equipment can be a desktop computer, a notebook computer, a palm computer, a cloud server and other computing equipment. The computer equipment can perform man-machine interaction with a user through a keyboard, a mouse, a remote controller, a touch pad or voice control equipment and the like.

The memory 61 includes at least one type of readable storage media including flash memory, hard disk, multimedia card, card memory (e.g., SD or DX memory, etc.), random Access Memory (RAM), static Random Access Memory (SRAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), programmable Read Only Memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the storage 61 may be an internal storage unit of the computer device 6, such as a hard disk or a memory of the computer device 6. In other embodiments, the memory 61 may also be an external storage device of the computer device 6, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash Card (Flash Card) or the like, which are provided on the computer device 6. Of course, the memory 61 may also comprise both an internal memory unit of the computer device 6 and an external memory device. In this embodiment, the memory 61 is typically used to store an operating system and various types of application software installed on the computer device 6, such as computer readable instructions of an abnormal medical item detection method. Further, the memory 61 may be used to temporarily store various types of data that have been output or are to be output.

The processor 62 may be a central processing unit (Central Processing Unit, CPU), controller, microcontroller, microprocessor, or other data processing chip in some embodiments. The processor 62 is typically used to control the overall operation of the computer device 6. In this embodiment, the processor 62 is configured to execute computer readable instructions stored in the memory 61 or process data, such as computer readable instructions for executing the abnormal medical item detection method.

The network interface 63 may comprise a wireless network interface or a wired network interface, which network interface 63 is typically used for establishing a communication connection between the computer device 6 and other electronic devices.

The computer equipment provided by the application utilizes the document embedded model based on the natural language model to search similar cases, so that the medical information of medical claim cases is more comprehensively utilized, and the comparability of the searched similar cases and the current claim cases is improved. The frequency and cost conditions of medical charging items in similar claim settlement cases are utilized to primarily screen suspicious unreasonable medical charging, so that the workload of subsequent rationality judgment can be reduced.

The present application also provides another embodiment, namely, a computer-readable storage medium storing computer-readable instructions executable by at least one processor to cause the at least one processor to perform the steps of the abnormal medical item detection method as described above.

The computer readable storage medium provided by the application utilizes the document embedded model based on the natural language model to search similar cases, more comprehensively utilizes the treatment information of the medical claim case, and improves the comparability of the searched similar cases and the current claim case. The frequency and cost conditions of medical charging items in similar claim settlement cases are utilized to primarily screen suspicious unreasonable medical charging, so that the workload of subsequent rationality judgment can be reduced.

From the above description of the embodiments, it will be clear to those skilled in the art that the above-described embodiment method may be implemented by means of software plus a necessary general hardware platform, but of course may also be implemented by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (e.g. ROM/RAM, magnetic disk, optical disk) comprising instructions for causing a terminal device (which may be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to perform the method according to the embodiments of the present application.

It is apparent that the above-described embodiments are only some embodiments of the present application, but not all embodiments, and the preferred embodiments of the present application are shown in the drawings, which do not limit the scope of the patent claims. This application may be embodied in many different forms, but rather, embodiments are provided in order to provide a thorough and complete understanding of the present disclosure. Although the application has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that modifications may be made to the embodiments described in the foregoing description, or equivalents may be substituted for elements thereof. All equivalent structures made by the content of the specification and the drawings of the application are directly or indirectly applied to other related technical fields, and are also within the scope of the application.

Claims

1. A method for detecting an abnormal medical item, comprising the steps of:

2. The abnormal medical item detection method according to claim 1, further comprising, before the step of reading the database and extracting the set of history document embedded vectors from the database, the steps of:

acquiring historical claim cases;

inputting the historical claim case into the document embedding model to perform vector conversion operation to obtain the historical document embedding vector set;

And storing the history document embedded vector set into the database.

3. The method of claim 1, wherein the step of receiving current medical treatment data to be audited comprises the steps of:

receiving a current visit picture to be audited;

invoking screenshot template data corresponding to the current visit picture;

performing image interception operation on the current diagnosis picture based on the screenshot template data to obtain a semantic segmentation screenshot and a template matching screenshot;

inputting the semantic segmentation screenshot into an image segmentation model to perform semantic segmentation operation to obtain a semantic segmentation field;

performing similar matching operation on the template matching screenshot in the screenshot template data to obtain a template matching field;

and inputting the semantic segmentation field and the template matching field into a character recognition model to perform character recognition operation, so as to obtain the current treatment data.

4. The abnormal medical item detection method according to claim 3, wherein before the step of inputting the semantic segmentation screenshot into an image segmentation model for semantic segmentation operation, the method further comprises:

Acquiring a training data set, wherein the training data set comprises a plurality of input images, the target object in each input image and a rectangular area corresponding to the target object in each input image;

the method comprises the steps that an image segmentation network is obtained, the image segmentation network comprises a first sub-network and a second sub-network, the first sub-network is used for outputting a target object in an image, and the second sub-network is used for outputting a rectangular area corresponding to the target object in the image;

and training the image segmentation network according to the training data set to obtain the image segmentation model.

5. The abnormal medical item detection method according to claim 4, wherein the step of training the image segmentation network according to the training data set to obtain the image segmentation model specifically comprises:

obtaining a loss function of the image segmentation network, the loss function comprising a cross entropy loss for characterizing the first subnetwork and a regression loss of the second subnetwork;

and training the image segmentation network by using a back propagation algorithm according to the loss function and the training data set until the image segmentation network converges to obtain the image segmentation model.

6. The abnormal medical item detection method according to claim 1, wherein the step of judging whether abnormal medical item fees exist in the current item fees according to the similar document embedding vector set comprises the following steps:

calculating average cost and variance cost of the similar documents corresponding to the current medical project according to the similar document embedded vector set;

respectively acquiring current project cost, similar document average cost and similar document variance cost belonging to the same medical project, and comparing the current project cost with the sum of the similar document average cost and the similar document variance cost;

the step of outputting a cost anomaly signal of the abnormal medical project cost if the abnormal medical project cost exists in the current project cost, specifically comprises the following steps:

if the current item cost is greater than the sum of the average cost of the similar documents and the variance cost of the similar documents, confirming that the current compared medical item belongs to the abnormal medical item cost, and outputting a cost abnormal signal of the current compared medical item;

The step of outputting a cost error-free signal if the abnormal medical project cost does not exist in the current project cost, specifically comprising the following steps:

if the current item cost is smaller than or equal to the sum of the average cost of the similar documents and the variance cost of the similar documents, confirming that the current compared medical item does not belong to the abnormal medical item cost, and outputting a cost error-free signal of the current compared medical item.

7. An abnormal medical item detection device, comprising:

8. The abnormal medical item detection apparatus of claim 7, wherein the apparatus further comprises:

the historical case vector conversion module is used for inputting the historical claim case into the document embedding model to perform vector conversion operation to obtain the historical document embedding vector set;

9. A computer device comprising a memory having stored therein computer readable instructions which when executed implement the steps of the abnormal medical item detection method of any of claims 1 to 6.

10. A computer readable storage medium having stored thereon computer readable instructions which when executed by a processor implement the steps of the abnormal medical item detection method according to any one of claims 1 to 6.