A method, device, computer equipment and storage medium for entering claims data

Abstract

The embodiments of the present application belong to the field of big data, and relate to a method and device for entering claims information, a computer device, and a storage medium, the method comprising: obtaining claims information to be entered, the claims information comprising a number of images; processing the images through a preset OCR text detection model to obtain corresponding text content, performing a first preprocessing on the text content to generate a target feature vector for each image, so as to classify each image into documents; performing a second preprocessing on the text content based on the document classification result, and inputting the second preprocessing result into a preset rule extractor and a preset neural network model respectively, and obtaining corresponding output results respectively for model integration processing to extract non-standard text elements, and entering the extracted results into the claims system. The present application also relates to blockchain technology, and the privacy information of users in the claims information can be stored in the blockchain. The present application can realize efficient, automated, and intelligent entry of claims information, improve the speed of data entry, and save costs.

Description

Technical Field

[0001] The present application relates to the field of big data technology, and in particular to a method, device, computer equipment and storage medium for entering claims information. Background Art

[0002] During the claims process, customers are required to submit relevant case information to prove the rationality and authenticity of the case. Case information includes customer ID information, medical treatment information, and medical expense information. Medical treatment information includes outpatient medical records, inpatient medical records, pathology reports, etc. Medical expense information includes medical invoices, medical lists, social insurance settlement statements, etc. All of these information need to be entered and archived.

[0003] Traditional data entry is done manually, which takes a long time and is inefficient. It is time-consuming and labor-intensive, prolongs the claims settlement cycle, and affects user experience. Summary of the invention

[0004] The purpose of the embodiments of the present application is to provide a method, device, computer equipment and storage medium for entering claims information to solve the problems of long entry time and low efficiency in the prior art process of entering claims information.

[0005] In order to solve the above technical problems, the present application embodiment provides a method for entering claims data, which adopts the following technical solution:

[0006] A method for entering claims data comprises the following steps:

[0007] Acquiring claims information to be entered, wherein the claims information includes a plurality of images;

[0008] The images are processed by a preset OCR text detection model to obtain text content corresponding to the images, the text content is first preprocessed, a target feature vector of each image is generated according to the first preprocessing result, and then the images are classified according to the target feature vector;

[0009] Based on the document classification result, the text content is subjected to a second preprocessing, and the second preprocessing result is respectively input into a preset rule extractor and a preset neural network model to obtain corresponding output results, and model integration processing is performed according to the output results to extract non-standard text elements, and the extracted results are entered into the claims system to complete the structured entry of claims information.

[0010] Further, after the second preprocessing result is input into the preset rule extractor, the method includes:

[0011] The preset rule extractor sequentially performs element matching word positioning and neighborhood search on the input second preprocessing result to perform named entity recognition on the text content, and obtains the value corresponding to each named entity and the confidence of each value.

[0012] Furthermore, when the preset rule extractor performs a field search on the input second preprocessing result, the method includes:

[0013] Based on the position of the matched named entity in the text content, distance screening, part-of-speech screening and semantic screening operations are sequentially performed on the second preprocessed text content to obtain a complete value corresponding to the named entity.

[0014] Furthermore, after the second preprocessing result is input into the preset neural network model, the method includes:

[0015] Multi-scale sliding window context splicing is performed on the second preprocessing result, and named entity recognition is performed according to the spliced ​​content to obtain recognized named entities and corresponding values ​​as well as confidence levels of each value.

[0016] Furthermore, the step of performing model integration processing according to the output result to extract non-standard text elements includes:

[0017] The confidences of the various values ​​output by the preset rule extractor and the preset neural network model are normalized and compared, and then the normalized confidences of the same named entity recognized by the preset rule extractor and the preset neural network model are compared, and the value with higher confidence is selected as the final value of the named entity, and non-standard text elements are extracted based on the recognized named entities and their final values.

[0018] Furthermore, after the step of inputting the extracted results into the claims system to complete the structured input of claims data, the method further includes:

[0019] In response to the user's tracing request, the extracted non-standard text elements are traced to obtain the positions of the extracted non-standard text elements in the original image, so that the user can verify the entered text content.

[0020] Furthermore, after the step of inputting the extracted results into the claims system to complete the structured input of claims data, the method further includes:

[0021] Acquire new historical claims data, update the preset element locating words and matching rules in the preset rule extractor according to all the historical claims data, train the preset neural network model according to all the historical claims data as model training data, and update the parameters of the preset neural network model.

[0022] In order to solve the above technical problems, the embodiment of the present application further provides a claim data entry device, which adopts the following technical solution:

[0023] A claims data entry device, comprising:

[0024] A data acquisition module, used to acquire the claims data to be entered, wherein the claims data includes a number of images;

[0025] An image classification module is used to process the image through a preset OCR text detection model to obtain text content corresponding to the image, perform a first preprocessing on the text content, generate a target feature vector for each of the images according to the first preprocessing result, and then perform document classification on each of the images according to the target feature vector;

[0026] The data entry module is used to perform a second preprocessing on the text content based on the document classification result, input the second preprocessing result into a preset rule extractor and a preset neural network model respectively, obtain corresponding output results respectively, perform model integration processing according to the output results to extract non-standard text elements, and enter the extracted results into the claims system to complete the structured entry of claims data.

[0027] In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical solution:

[0028] A computer device includes a memory and a processor, wherein the memory stores computer-readable instructions, and the processor implements the steps of the claim data entry method as described above when executing the computer-readable instructions.

[0029] In order to solve the above technical problems, the embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solution:

[0030] A computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the claim data entry method as described above.

[0031] Compared with the prior art, the embodiments of the present application have the following beneficial effects:

[0032] This solution performs document classification on the claim information, and then extracts non-standard text elements from the text content based on the document classification results according to a preset rule extractor and a preset neural network model, thereby performing structured entry of the claim information, achieving efficient, automated and intelligent entry of the claim information, thereby increasing the speed of data entry, saving costs, and at the same time reducing the claim cycle and improving customer experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the scheme in the present application, a brief introduction is given below to the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0034] Figure 1 is an exemplary system architecture diagram to which the present application may be applied;

[0035] Figure 2 A flow chart of an embodiment of a method for entering claims data according to the present application;

[0036] Figure 3 is a schematic structural diagram of an embodiment of a claim data entry device according to the present application;

[0037] Figure 4 is a schematic structural diagram of another embodiment of the claim data entry device according to the present application;

[0038] Figure 5 It is a structural diagram of an embodiment of a computer device according to the present application. DETAILED DESCRIPTION

[0039] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present application; the terms used in the specification of the application herein are only for the purpose of describing specific embodiments and are not intended to limit the present application; the terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of the present application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0040] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0041] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0042] like Figure 1 As shown, the system architecture 100 may include terminal devices 101, 102, 103, a network 104 and a server 105. The network 104 is used to provide a medium for 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 optical fiber cables, etc.

[0043] Users can use terminal devices 101, 102, 103 to interact with server 105 through network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, 103, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0044] Terminal devices 101, 102, 103 can be various electronic devices with display screens and supporting web browsing, including but not limited to smart phones, tablet computers, e-book readers, MP3 players (Moving Picture Experts Group Audio Layer III), MP4 (Moving Picture Experts Group Audio Layer IV) players, laptop computers, desktop computers, etc.

[0045] The server 105 may be a server that provides various services, such as a background server that provides support for web pages displayed on the terminal devices 101 , 102 , and 103 .

[0046] It should be noted that the claim information entry method provided in the embodiment of the present application is generally executed by a server, and accordingly, the claim information entry device is generally set in the server.

[0047] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0048] Continue to refer Figure 2 , shows a flow chart of an embodiment of a method for entering claims data according to the present application. The method for entering claims data comprises the following steps:

[0049] Step S201, obtaining the claim information to be entered, wherein the claim information includes a plurality of images.

[0050] When a user needs to make a claim, he or she initiates a claim data upload request through a mobile phone or computer to the server that handles claim data collection and entry. After responding to the claim data upload request, the server receives the claim data uploaded by the user's mobile phone or computer. These claim data include customer ID information, medical diagnosis and treatment information, and medical expense information required in the claim process. Medical diagnosis and treatment information includes outpatient medical records, inpatient medical records, pathology reports, etc. Medical expense information includes medical invoices, medical lists, social insurance settlement statements, etc. These case claim data need to be collected and stored in the claim system. These claim data are uploaded in the form of pictures. After receiving the claim data, the server stores these pictures in the form of a data set.

[0051] In this embodiment, after the server obtains the claim information, the method further includes:

[0052] Each image is tested to ensure that all images meet the image quality requirements. Specifically, image quality inspection can be performed through one or more of a clarity detection model, a direction correction model, and a copy detection model.

[0053] Among them, the clarity detection model can use the existing image clarity detection model for quality inspection, such as the image clarity detection model based on the gradient operator or the trained convolutional neural network model; the direction correction model can use the projection-based method, the Hough transform-based method, the linear fitting-based method, the Fourier transform-to-frequency domain detection method, etc. for direction correction; the remake detection model can use noise analysis, pixel value deviation detection, etc. for detection. For images that do not meet the detection standards, a prompt message for re-uploading is sent to the user's mobile phone or computer, and the detection is re-performed after receiving the new uploaded image, until all the detections are passed and the subsequent steps are processed.

[0054] In some embodiments, the method further includes: performing deduplication operations on the claim information to remove the same claim information uploaded by the user at different upload entrances due to improper operation and prompting the user to confirm whether any claim information is missed.

[0055] Step S202, processing the image through a preset OCR text detection model to obtain text content corresponding to the image, performing a first preprocessing on the text content, generating a target feature vector for each of the images according to the first preprocessing result, and then classifying each of the images according to the target feature vector.

[0056] In this step, when acquiring text content according to the OCR text detection model, the existing OCR text detection model can be used to implement it.

[0057] When the text content is preprocessed for the first time, the purpose is to perform coarse-grained text content extraction, specifically, the text content is input into the preprocessing module to perform operations such as word form restoration, part-of-speech tagging, and digital regularization, wherein word form restoration is for English words in the text, and part-of-speech tagging is to mark each word as a noun, verb, adjective or other part of speech, and digital regularization is mainly to regularize the numbers involved in the cost in images such as medical invoices in the claim information, so as to facilitate the document classification of the image. Document classification here refers to the classification of images of expense documents (such as diagnosis sheets, invoices, etc.) and identity documents (such as identity cards, driver's licenses, etc.) in the claim information, so as to enable the subsequent targeted entry of the claim information into the claim system. When performing coarse-grained feature extraction, only the key text content that can be used for document classification is extracted as the classification feature.

[0058] When extracting classification features, this embodiment can use the bag-of-words model. Specifically, the text content is segmented and several segmented words are extracted as feature fields. The feature fields are screened to obtain a feature field set, and the distribution of each feature field in the feature field set in the histogram established based on the bag-of-words model is determined to obtain an intermediate feature vector. Finally, the intermediate feature vector is normalized and dimensionally reduced to obtain a target feature vector. The target feature vector is input into the document classification model to obtain the document classification result. For claim materials, they can be specifically divided into different document types such as certificates, medical records, diagnostic materials, invoices, lists, reimbursement certificates, bank cards, etc. The feature fields of different types are different. For example, the feature fields of identity cards include name, number, validity period, etc. The feature fields of medical records include hospital, admission and discharge diagnosis, admission and discharge date, name, etc. The dimension of the intermediate feature vector formed by the bag-of-words model to understand the content of the text is relatively large, so it is necessary to reduce the dimension of the intermediate feature vector to obtain the target feature vector.

[0059] In some embodiments, a convolutional neural network model can also be used to extract classification features, specifically, the text content of each image is spliced ​​and input into the convolutional neural network model to obtain a target feature vector, and the target feature vector is input into the document classification model to obtain the document classification result. Among them, when training the convolutional neural network model, the feature fields of different types of claims information are manually annotated, such as the feature fields of the ID card include name, number, validity period, etc., and the feature fields of the medical record include hospital, admission and discharge diagnosis, admission and discharge date, name, etc. The layout of different materials also has certain characteristics, such as the title of the invoice will be in the middle and upper part of the file. These features are manually defined and interfered. The training of the convolutional neural network model mainly includes the understanding of the content of the text. In addition, the understanding of the document layout of the text can also be trained, and the layout relationship of the feature fields of different materials is established in the continuous learning process.

[0060] In other embodiments, the bag-of-words model and the convolutional neural network model may be combined to extract classification features, specifically by comparing the probability values ​​of the target feature vectors obtained by the two, and selecting the target feature vector with a larger probability value as the final target feature vector.

[0061] Step S203, based on the document classification result, perform a second preprocessing on the text content, input the second preprocessing result into a preset rule extractor and a preset neural network model respectively, obtain corresponding output results respectively, perform model integration processing according to the output results to extract non-standard text elements, and enter the extracted results into the claims system to complete the structured entry of claims information.

[0062] In this step, different from the first preprocessing, the purpose of the second preprocessing of the text content is to perform fine-grained text content extraction based on the document classification results. When performing the second preprocessing, the claim information presented in the form of images is first divided into two image sets of documents according to the document classification results, and the second preprocessing is performed on the two image sets respectively. The difference from the first preprocessing is that when the second preprocessing is performed on the two image sets, there will be differences in the processing methods of the second preprocessing for document images and certificate images, that is, the second preprocessing includes common processing methods (such as word form restoration, part-of-speech tagging, etc.) and processing methods specific to image types (such as digital regularization of document images, etc.).

[0063] In this embodiment, the second preprocessing results are respectively input into the preset rule extractor and the preset neural network model, and the output results are the named entities obtained according to the text content after the second preprocessing, the values ​​corresponding to each named entity, and the confidence of each value.

[0064] In this embodiment, after the second preprocessing result is input into the preset rule extractor, the method includes: performing element matching word positioning and neighborhood search on the input second preprocessing result in sequence through the preset rule extractor to perform named entity recognition on the text content, and obtain the value corresponding to each named entity and the confidence of each value.

[0065] Specifically, the rule extractor extracts key-value pairs from the text content, where the key is the named entity and the value is the value of the named entity, such as name: Zhang San, height: 175cm, etc. These are the named entities and corresponding values ​​obtained by the rule extractor. The rule extractor locates the named entity through element matching words to achieve the purpose of named entity recognition. Several element positioning words are pre-defined in your rule extractor, such as height, discharge diagnosis, etc., which are defined element positioning words. The element positioning words and their synonyms are used to search and match in the second preprocessed text content to identify the named entity.

[0066] In this embodiment, when the preset rule extractor performs a field search on the input second preprocessing result, the method includes: taking the position of the matched named entity in the text content as a reference, performing distance filtering, part of speech filtering and semantic filtering operations on the second preprocessed text content in sequence to obtain a complete value corresponding to the named entity.

[0067] Specifically, after locating the named entity through the element matching word, the value corresponding to the named entity is further extracted through the field search. When performing the field search in this embodiment, the method includes taking the position of the matched named entity in the text content as the reference, and sequentially performing the distance screening, part-of-speech screening and semantic screening steps on the second preprocessed text content to obtain the complete value corresponding to the named entity. Among them, the distance screening is to query multiple elements in space that meet the distance requirements with the matched named entity, generally speaking, several elements that are located close to the matched named entity, the part-of-speech screening is to perform part-of-speech recognition on the multiple elements queried, and then perform element screening based on the matched named entity to obtain the first element, and the semantic screening is to perform semantic recognition and matching on other elements that are adjacent to the first element in space, and use the matched element as the second element, and splice the first element and the second element as the value corresponding to the matched named entity. For example, based on the element matching word "height", the named entity "height" is matched in the second preprocessed text content, and the nearest quantifier in space is further matched after "height", and then the nearest element is further matched based on the quantifier to obtain the unit of the quantifier, and the quantifier and unit are concatenated to obtain the value of height. While outputting the named entity and the corresponding value, the rule extractor will also output the confidence of each value, which is the probability value of correctly identifying the named entity and the corresponding value.

[0068] In some embodiments, the value of a named entity does not necessarily have a unit, so semantic filtering is not necessary in domain search and can be adaptively performed according to actual conditions.

[0069] In some embodiments, if the same named entity is not matched according to the element locating word, fuzzy matching can be further performed by calculating the edit distance between the word segmentation of the second preprocessed text content and the element locating word to achieve named entity recognition. Since the naming of entities is often irregular and there may be multiple variations and spellings, the recall rate of named entity recognition based on a complete match of the element locating word is low. The use of edit distance generalizes from complete matching to fuzzy matching, effectively improving the recognition success rate.

[0070] In this embodiment, after the second preprocessing result is input into the preset neural network model, the method includes: performing multi-scale sliding window context splicing on the second preprocessing result, performing named entity recognition based on the spliced ​​content, and obtaining the recognized named entities and corresponding values ​​and the confidence of each value. Wherein, when the preset neural network model performs named entity recognition based on the spliced ​​content, it performs structured analysis on the text content, and recognizes all entities and corresponding values ​​in the text.

[0071] In this embodiment, the step of performing model integration processing based on the output results to extract non-standard text elements includes: normalizing the confidence of each value output by the preset rule extractor and the preset neural network model, and then comparing the normalized confidence of the same named entity recognized by the preset rule extractor and the preset neural network model, selecting the value with higher confidence as the final value of the named entity, and extracting non-standard text elements based on the recognized named entity and its final value. After the extraction operation is completed, the extraction results are entered into the claims system to complete the structured entry of claims information.

[0072] In this embodiment, after the step of entering the extracted results into the claims system and completing the structured entry of the claims information, the method further includes: tracing the extracted non-standard text elements in response to the user's traceability request, and obtaining the position of the extracted non-standard text elements in the original image, so that the user can verify the entered text content. Correspondingly, the method also includes the step of establishing a correspondence between the text and image positions, specifically determining the position of the text content in the image through target detection, and storing the mapping relationship between the text content and the image position in the form of coordinates. When the user's mouse stays on the entered text interface, in response to the stop action, the corresponding image is automatically obtained and displayed according to the text at the mouse stop position, which is convenient for manual verification of the automatically entered text content. Compared with the existing full manual entry method, it can greatly reduce the entry time and improve the overall timeliness.

[0073] In this embodiment, after the step of entering the extracted results into the claims system and completing the structured entry of the claims information, the method further includes: obtaining new historical claims information, updating the preset element locating words and matching rules in the preset rule extractor according to all the historical claims information, and training the preset neural network model according to all the historical claims information as model training data, and updating the parameters of the preset neural network model. This step can optimize the preset rule extractor and the preset neural network model, wherein the optimization of the rule extractor is to optimize and update the preset element locating words and matching rules according to the historical claims information, and the optimization of the preset neural network model is to optimize and train according to the historical claims information as model training data, and update the model parameters, so that the claims service can be optimized through the optimized preset rule extractor and preset neural network model.

[0074] The above method of the present application can intelligently enter and archive the information submitted by the customer (including customer ID information, medical treatment information, medical expense information, medical treatment information including outpatient medical records, inpatient medical records, pathology reports, etc., medical expense information including medical invoices, medical lists, social security settlement statements, etc.), specifically by classifying the claims information, and then extracting non-standard text elements from the text content based on the document classification results according to a preset rule extractor and a preset neural network model, so as to perform structured entry of the claims information, thereby realizing efficient automation and intelligent entry of claims information, thereby improving the speed of data entry, saving costs, and at the same time reducing the claims cycle and improving customer experience.

[0075] It should be emphasized that in order to further ensure the privacy and security of the above-mentioned claims information, the user privacy information in the above-mentioned claims information can also be stored in a blockchain node.

[0076] The blockchain referred to in this application is a new application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, encryption algorithm, etc. Blockchain is essentially a decentralized database, a string of data blocks generated by cryptographic methods. Each data block contains a batch of network transaction information, which is used to verify the validity of its information (anti-counterfeiting) and generate the next block. Blockchain can include the underlying blockchain platform, platform product service layer, and application service layer.

[0077] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through computer-readable instructions, and the computer-readable instructions can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the aforementioned storage medium can be a non-volatile storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0078] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0079] Further references Figure 3 , as a response to the above Figure 2 In order to realize the method shown in the figure, the present application provides an embodiment of a claim data entry device, and the device embodiment is Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0080] like Figure 3 As shown, the claim data entry device 300 described in this embodiment includes: a data acquisition module 301, an image classification module 302 and a data entry module 303. Among them:

[0081] The data acquisition module 301 is used to acquire the claims data to be entered, and the claims data includes a number of images; the image classification module 302 is used to process the images through a preset OCR text detection model to obtain text content corresponding to the images, perform a first preprocessing on the text content, generate a target feature vector for each image according to the first preprocessing result, and then perform document classification on each image according to the target feature vector; the data entry module 303 is used to perform a second preprocessing on the text content based on the document classification result, input the second preprocessing result into a preset rule extractor and a preset neural network model respectively, obtain corresponding output results respectively, perform model integration processing according to the output results to extract non-standard text elements, and enter the extracted results into the claims system to complete the structured entry of claims data.

[0082] When a user needs to make a claim, the data acquisition module 301 receives a claim data upload request initiated by the user through a mobile phone or computer to the server side that processes the claim data collection and entry, and responds to the claim data upload request to receive the claim data uploaded by the user's mobile phone or computer. These claim data include customer ID information, medical diagnosis and treatment information, and medical expense information required in the claim process. Medical diagnosis and treatment information includes outpatient medical records, inpatient medical records, pathology reports, etc. Medical expense information includes medical invoices, medical lists, social insurance settlement statements, etc. These case claim data need to be collected and stored in the claim system. These claim data are uploaded in the form of pictures. After receiving the claim data, these pictures are stored in the form of data sets to the server side.

[0083] In this embodiment, the data acquisition module 301 is also used to detect each image after obtaining the claim data on the server side to ensure that all images meet the image quality requirements. Specifically, the image quality inspection can be performed through one or more of the clarity detection model, the direction correction model and the remake detection model. Among them, the clarity detection model can use the existing image clarity detection model for quality inspection, such as the image clarity detection model based on the gradient operator or the trained convolutional neural network model; the direction correction model can use a projection-based method, a Hough transform-based method, a linear fitting-based method, a Fourier transform-to-frequency domain detection method, etc. for direction correction; the remake detection model can use noise analysis, pixel value deviation detection, etc. for detection. For images that do not meet the detection standards, a prompt message for re-uploading is sent to the user's mobile phone or computer, and re-detection is performed after receiving the new uploaded image until all detections pass.

[0084] In some embodiments, the information acquisition module 301 is also used to deduplicate the claim information to remove the same claim information uploaded by the user at different upload entrances due to improper operation and prompt the user to confirm whether any claim information is missed.

[0085] In this step, the image classification module 302 may adopt an existing OCR text detection model to obtain text content according to the OCR text detection model.

[0086] When the image classification module 302 performs the first preprocessing on the text content, the purpose is to perform coarse-grained text content extraction, specifically, inputting the text content into the preprocessing module to perform operations such as word form restoration, part-of-speech tagging, and digital regularization, wherein word form restoration is for English words in the text, and part-of-speech tagging is to mark each word as a noun, verb, adjective or other part of speech, and digital regularization is mainly to regularize the numbers involved in the cost in images such as medical invoices in the claim information, so as to facilitate the document classification of the image. Here, document classification refers to classifying the images of expense documents (such as diagnosis sheets, invoices, etc.) and identity documents (such as identity cards, driver's licenses, etc.) in the claim information, so as to subsequently enter the claim information in the claim system in a targeted manner. When performing coarse-grained feature extraction, only the key text content that can be used for document classification is extracted as the classification feature.

[0087] When the image classification module 302 extracts classification features, this embodiment can use a bag-of-words model to extract several words as feature fields after segmenting the text content, screen the feature fields to obtain a feature field set, determine the distribution of each feature field in the feature field set in a histogram established based on the bag-of-words model, obtain an intermediate feature vector, and finally perform normalization and dimensionality reduction on the intermediate feature vector to obtain a target feature vector. The target feature vector is input into the document classification model to obtain the document classification result. For claim materials, they can be specifically divided into different document types such as certificates, medical records, diagnostic materials, invoices, lists, reimbursement certificates, bank cards, etc., and different types have different feature fields. For example, the feature fields of identity cards include name, number, validity period, etc., and the feature fields of medical records include hospital, admission and discharge diagnosis, admission and discharge date, name, etc. The dimension of the intermediate feature vector formed by the bag-of-words model to understand the content of the text is relatively large, so it is necessary to perform dimensionality reduction on the intermediate feature vector to obtain the target feature vector.

[0088] In some embodiments, the image classification module 302 may also use a convolutional neural network model to extract classification features, specifically for splicing the text content of each image and inputting it into the convolutional neural network model to obtain a target feature vector, and inputting the obtained target feature vector into the document classification model to obtain the document classification result. Wherein, when training the convolutional neural network model, the feature fields of different types of claim materials are manually annotated, such as the feature fields of the ID card include name, number, validity period, etc., and the feature fields of the medical record include hospital, admission and discharge diagnosis, admission and discharge date, name, etc. The layout of different materials also has certain characteristics, such as the title of the invoice will be in the middle and upper part of the file, and these features are manually defined and interfered. The training of the convolutional neural network model mainly includes the understanding of the content of the text. In addition, the understanding of the document layout of the text can also be trained, and the layout relationship of the feature fields of different materials is established in the continuous learning process.

[0089] In other embodiments, the image classification module 302 can also be used to extract classification features in combination with the bag-of-words model and the convolutional neural network model, specifically by comparing the probability values ​​of the target feature vectors obtained by the two, and selecting the target feature vector with a larger probability value as the final target feature vector.

[0090] In this embodiment, different from the first preprocessing performed by the image classification module 302, the purpose of the second preprocessing performed by the data entry module 303 on the text content is to perform fine-grained text content extraction based on the document classification results. When performing the second preprocessing, the claim information presented in the form of images is first divided into two image sets of documents according to the document classification results, and the second preprocessing is performed on the two image sets respectively. The difference from the first preprocessing performed by the image classification module 302 is that when the data entry module 303 performs the second preprocessing on the two image sets, there will be differences in the processing methods of the second preprocessing for document images and certificate images, that is, the second preprocessing includes common processing methods (such as word form restoration, part-of-speech tagging, etc.) and processing methods specific to image types (such as digital regularization of document images, etc.).

[0091] In this embodiment, the data entry module 303 inputs the second preprocessing result into the preset rule extractor and the preset neural network model respectively, and the output result is the named entities obtained according to the text content after the second preprocessing, the values ​​corresponding to each named entity, and the confidence of each value.

[0092] In this embodiment, after the data entry module 303 inputs the second preprocessing result into the preset rule extractor, it is specifically used to: perform element matching word positioning and neighborhood search on the input second preprocessing result in sequence through the preset rule extractor to perform named entity recognition on the text content, and obtain the value corresponding to each named entity and the confidence of each value.

[0093] Specifically, the data entry module 303 is used to extract key-value pairs from the text content through the rule extractor, where the key is the named entity and the value is the value of the named entity, such as name: Zhang San, height: 175cm, etc. This is the named entity and the corresponding value obtained by the rule extractor. The rule extractor locates the named entity through the element matching words to achieve the purpose of named entity recognition. Several element locating words are pre-defined in your rule extractor, such as height, discharge diagnosis, etc., which are defined element locating words. The element locating words and their synonyms are used to search and match in the second preprocessed text content to identify the named entity.

[0094] In this embodiment, when the data entry module 303 performs a field search on the second preprocessing result input through the preset rule extractor, it is specifically used to: based on the position of the matched named entity in the text content, perform distance filtering, part of speech filtering and semantic filtering operations on the second preprocessed text content in sequence to obtain the complete value corresponding to the named entity.

[0095] Specifically, after the data entry module 303 locates the named entity through the element matching words, it further extracts the value corresponding to the named entity through the field search. In this embodiment, when the data entry module 303 performs the field search, it is specifically used to use the position of the matched named entity in the text content as a reference, and sequentially perform the distance screening, part-of-speech screening and semantic screening steps on the second preprocessed text content to obtain the complete value corresponding to the named entity. Among them, the distance screening is to query multiple elements that meet the requirements of the distance from the matched named entity in space, generally speaking, several elements that are located close to the matched named entity. The part-of-speech screening is to perform part-of-speech recognition on the multiple elements queried, and then perform element screening based on the matched named entity to obtain the first element. The semantic screening is to perform semantic recognition and matching on other elements that are adjacent to the first element in space, and use the matched element as the second element. The first element and the second element are spliced ​​as the value corresponding to the matched named entity. For example, based on the element matching word "height", the named entity "height" is matched in the second preprocessed text content, and the nearest quantifier in space is further matched after "height", and then the nearest element is further matched based on the quantifier to obtain the unit of the quantifier, and the quantifier and unit are concatenated to obtain the value of height. While outputting the named entity and the corresponding value, the rule extractor will also output the confidence of each value, which is the probability value of correctly identifying the named entity and the corresponding value.

[0096] In some embodiments, the value of a named entity does not necessarily have a unit, so semantic screening is not necessary when the data entry module 303 performs a domain search, and can be adaptively performed according to actual conditions.

[0097] In some embodiments, if the data entry module 303 does not match the same named entity according to the element locating word, the data entry module 303 is further used to perform fuzzy matching by calculating the edit distance between the word segmentation of the second preprocessed text content and the element locating word to achieve named entity recognition. Since the naming of entities is often irregular and there may be multiple variations and spelling forms, the recall rate of named entity recognition based on a complete match of the element locating word is low. The use of the edit distance generalizes from a complete match to a fuzzy match, effectively improving the recognition success rate.

[0098] In this embodiment, after the data entry module 303 inputs the second preprocessing result into the preset neural network model, it is specifically used to perform multi-scale sliding window context splicing on the second preprocessing result, perform named entity recognition based on the splicing content, and obtain the recognized named entities and corresponding values ​​and the confidence of each value. Among them, when the preset neural network model performs named entity recognition based on the splicing content, it performs structured analysis on the text content, and recognizes all entities and corresponding values ​​in the text.

[0099] In this embodiment, the data entry module 303 performs model integration processing based on the output results to extract non-standard text elements, and is specifically used to normalize the confidence of each value output by the preset rule extractor and the preset neural network model, and then compares the normalized confidence of the same named entity recognized by the preset rule extractor and the preset neural network model, selects the value with higher confidence as the final value of the named entity, and extracts non-standard text elements based on the recognized named entity and its final value. After the extraction operation is completed, the extraction results are entered into the claims system to complete the structured entry of claims information.

[0100] In this embodiment, if Figure 4 As shown, the device also includes a traceability module 304, which is used to enter the extracted results into the claims system in the data entry module 303. After completing the structured entry of the claims data, it responds to the user's traceability request to trace the extracted non-standard text elements, and obtains the position of the extracted non-standard text elements in the original image, so that the user can verify the entered text content. Correspondingly, the traceability module 34 is also used to establish a corresponding relationship between the text and image positions, specifically to determine the position of the text content in the image through target detection, and store the mapping relationship between the text content and the image position in the form of coordinates. When the user's mouse stays on the entered text interface, in response to the stop action, the corresponding image is automatically obtained and displayed according to the text at the mouse stop position, which is convenient for manual verification of the automatically entered text content. Compared with the existing full manual entry method, it can greatly reduce the entry time and improve the overall timeliness.

[0101] In this embodiment, if Figure 4 As shown, the device also includes an update module 305, which is used to enter the extracted results into the claims system in the data entry module 303, and after completing the structured entry of the claims data, obtain new historical claims data, update the preset element locating words and matching rules in the preset rule extractor according to all historical claims data, and train the preset neural network model according to all historical claims data as model training data, and update the parameters of the preset neural network model. The update module optimizes the preset rule extractor and the preset neural network model, wherein the optimization of the rule extractor is to optimize and update the preset element locating words and matching rules according to the historical claims data, and the optimization of the preset neural network model is to optimize and train according to the historical claims data as model training data, and update the model parameters, so that the claims service can be optimized through the optimized preset rule extractor and preset neural network model.

[0102] The above-mentioned device of the present application can intelligently enter and archive the information submitted by the customer (including customer ID information, medical diagnosis and treatment information, medical expense information, medical diagnosis and treatment information including outpatient medical records, inpatient medical records, pathology reports, etc., medical expense information including medical invoices, medical lists, social security settlement statements, etc.), specifically by classifying the claim information, and then extracting non-standard text elements from the text content based on the document classification results according to the preset rule extractor and the preset neural network model, so as to perform structured entry of the claim information, realize efficient automation and intelligent entry of claim information, thereby improving the speed of data entry, saving costs, and at the same time reducing the claim cycle and improving customer experience.

[0103] To solve the above technical problems, the present application also provides a computer device. Figure 5 , Figure 5 This is a basic structural block diagram of the computer device in this embodiment.

[0104] The computer device 5 includes a memory 51, a processor 52, and a network interface 53 which are interconnected through a system bus. The memory 51 stores computer-readable instructions, and the processor 52 implements the steps of the claim data entry method described in the above embodiment when executing the computer-readable instructions.

[0105] It should be pointed out that Figure 5 Only a computer device 5 having components 51-53 is shown, but it should be understood that it is not required to implement all the components shown, and more or fewer components may be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (ASIC), programmable gate arrays (FPGA), digital signal processors (DSP), embedded devices, etc.

[0106] The computer device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may interact with a user through a keyboard, a mouse, a remote controller, a touch pad, or a voice control device.

[0107] The memory 51 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, 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, disk, optical disk, etc. In some embodiments, the memory 51 can be an internal storage unit of the computer device 5, such as a hard disk or memory of the computer device 5. In other embodiments, the memory 51 can also be an external storage device of the computer device 5, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (FlashCard), etc. equipped on the computer device 5. Of course, the memory 51 can also include both the internal storage unit of the computer device 5 and its external storage device. In this embodiment, the memory 51 is generally used to store the operating system and various application software installed on the computer device 5, such as computer-readable instructions of the claim data entry method, etc. In addition, the memory 51 can also be used to temporarily store various types of data that have been output or are to be output.

[0108] The processor 52 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chips in some embodiments. The processor 52 is generally used to control the overall operation of the computer device 5. In this embodiment, the processor 52 is used to run the computer-readable instructions stored in the memory 51 or process data, such as running the computer-readable instructions of the claim data entry method.

[0109] The network interface 53 may include a wireless network interface or a wired network interface. The network interface 53 is generally used to establish a communication connection between the computer device 5 and other electronic devices.

[0110] The above-mentioned computer device of the present application can intelligently enter and archive the information submitted by the customer (including customer ID information, medical treatment information, medical expense information, medical treatment information including outpatient medical records, inpatient medical records, pathology reports, etc., medical expense information including medical invoices, medical lists, social security settlement statements, etc.), specifically by classifying the claim information, and then extracting non-standard text elements from the text content based on the document classification results according to the preset rule extractor and the preset neural network model, so as to perform structured entry of the claim information, realize efficient automation and intelligent entry of claim information, thereby improving the speed of data entry, saving costs, and at the same time reducing the claim cycle and improving customer experience.

[0111] The present application also provides another embodiment, namely, providing a computer-readable storage medium, wherein the computer-readable storage medium stores computer-readable instructions, and the computer-readable instructions can be executed by at least one processor to enable the at least one processor to perform the steps of the claim information entry method as described above.

[0112] When the instructions stored in the above-mentioned computer-readable storage medium of the present application are executed, the information submitted by the customer (including customer ID information, medical treatment information, medical expense information, medical treatment information including outpatient medical records, inpatient medical records, pathology reports, etc., medical expense information including medical invoices, medical lists, social insurance settlement statements, etc.) can be intelligently entered and archived. Specifically, the claim information is classified into documents, and then non-standard text elements are extracted from the text content based on the document classification results according to a preset rule extractor and a preset neural network model, so as to perform structured entry of the claim information, thereby realizing efficient automation and intelligent entry of the claim information, thereby improving the speed of data entry, saving costs, and at the same time reducing the claim cycle and improving customer experience.

[0113] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0114] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.

Claims

1. A method for entering claims data, characterized in that: The steps include: Acquiring claims information to be entered, wherein the claims information includes a plurality of images; The images are processed by a preset OCR text detection model to obtain text content corresponding to the images, the text content is first preprocessed, a target feature vector of each image is generated according to the first preprocessing result, and then the images are classified according to the target feature vector; Based on the document classification result, the text content is subjected to a second preprocessing, and the second preprocessing result is respectively input into a preset rule extractor and a preset neural network model to obtain corresponding output results, and model integration processing is performed according to the output results to extract non-standard text elements, and the extracted results are input into the claims system to complete the structured input of claims information; In response to the user's tracing request, the extracted non-standard text elements are traced to obtain the positions of the extracted non-standard text elements in the original image, so that the user can verify the entered text content.

2. The method for entering claims data according to claim 1, characterized in that: After the second preprocessing result is input into the preset rule extractor, the method includes: The preset rule extractor sequentially performs element matching word positioning and neighborhood search on the input second preprocessing result to perform named entity recognition on the text content, and obtains the value corresponding to each named entity and the confidence of each value.

3. The method for entering claims data according to claim 2, characterized in that: When the preset rule extractor performs a field search on the second preprocessing result input, the method includes: Based on the position of the matched named entity in the text content, distance screening, part-of-speech screening and semantic screening operations are sequentially performed on the second preprocessed text content to obtain a complete value corresponding to the named entity.

4. The method for entering claim data according to any one of claims 1 to 3, characterized in that: After the second preprocessing result is input into the preset neural network model, the method includes: Multi-scale sliding window context splicing is performed on the second preprocessing result, and named entity recognition is performed according to the spliced ​​content to obtain recognized named entities and corresponding values ​​as well as confidence levels of each value.

5. The method for entering claim data according to claim 4, characterized in that: The step of performing model integration processing according to the output result to extract non-standard text elements comprises: The confidences of the various values ​​output by the preset rule extractor and the preset neural network model are normalized and compared, and then the normalized confidences of the same named entity recognized by the preset rule extractor and the preset neural network model are compared, and the value with higher confidence is selected as the final value of the named entity, and non-standard text elements are extracted based on the recognized named entities and their final values.

6. The method for entering claim data according to claim 5, characterized in that: After the step of inputting the extracted results into the claims system to complete the structured input of claims data, the method further includes: Acquire new historical claims data, update the preset element locating words and matching rules in the preset rule extractor according to all the historical claims data, train the preset neural network model according to all the historical claims data as model training data, and update the parameters of the preset neural network model.

7. A claim data entry device, characterized in that: include: A data acquisition module, used to acquire the claims data to be entered, wherein the claims data includes a number of images; An image classification module is used to process the image through a preset OCR text detection model to obtain text content corresponding to the image, perform a first preprocessing on the text content, generate a target feature vector for each of the images according to the first preprocessing result, and then perform document classification on each of the images according to the target feature vector; A data entry module, for performing a second preprocessing on the text content based on the document classification result, inputting the second preprocessing result into a preset rule extractor and a preset neural network model respectively, obtaining corresponding output results respectively, performing model integration processing according to the output results to extract non-standard text elements, and inputting the extracted results into the claims system to complete the structured entry of claims data; The tracing module is used to trace the extracted non-standard text elements in response to the user's tracing request, and obtain the position of the extracted non-standard text elements in the original image so that the user can verify the entered text content.

8. A computer device, characterized in that: It comprises a memory and a processor, wherein the memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the claim data entry method as described in any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the steps of the claim data entry method according to any one of claims 1 to 6 are implemented.

Images