WO2019085064A1

WO2019085064A1 - Medical claim denial determination method, device, terminal apparatus, and storage medium

Info

Publication number: WO2019085064A1
Application number: PCT/CN2017/112371
Authority: WO
Inventors: 梁效栋; 戴建云
Original assignee: 平安科技（深圳）有限公司
Priority date: 2017-10-30
Filing date: 2017-11-22
Publication date: 2019-05-09
Also published as: CN107679997A

Abstract

A medical claim denial determination method, a device, a terminal an apparatus, and a storage medium. The medical claim denial determination method comprises: acquiring a claim application request comprising a case ID and current bill information, the current bill information comprising information about at least one current item (S10); acquiring past bill information corresponding to the case ID on the basis of the case ID, the past bill information comprising information about at least one past item (S20); determining whether the current bill information and the past bill information correspond to the same bill on the basis of the information about the at least one current item and the information about the at least one past item (S30); and if so, outputting claim denial information (S40). The automatic medical claim denial determination method determines whether the current bill is the same bill as the past bill corresponding to the case ID, and automatically denies a claim application request for the same bill, thereby increasing the processing accuracy of claim cases, and improving the processing efficiency of claim application requests.

Description

Medical claim refusal method, device, terminal device and storage medium

This patent application is based on the Chinese invention patent application filed on October 30, 2017, with the application number of 201711033230.8, entitled "Medical Claims Reimbursement Method, Device, Terminal Equipment and Storage Medium", and requires its priority.

Technical field

The present application relates to the field of medical claims, and in particular, to a medical claims compensation method, device, terminal device and storage medium.

Background technique

Medical insurance is an insurance policy based on economic losses caused by disease risks. In the process of claim settlement of medical insurance, the insured submits a claim for claim to the business personnel of the insurance institution, and attaches a corresponding medical bill, and the business personnel review the claim application request and the corresponding medical bill, in which the medical bill has been When the historical bill of the claim is duplicated, the claim for the claim is rejected, so as to avoid property damage to the insurance institution. This type of manual review of repetitive bills and claims processing of chargebacks takes a lot of time and has low accuracy and high labor costs.

Summary of the invention

The embodiment of the present application provides a method, a device, a terminal device, and a storage medium method for medical claims compensation, which solve the problem of manually reviewing duplicate bills and performing claims processing methods.

In a first aspect, an embodiment of the present application provides a medical claim compensation method, including:

Obtaining a claim application request, the claim application request includes a case ID and current bill information, and the current bill information includes at least one current item information;

Obtaining historical billing information corresponding to the case ID based on the case ID, the historical billing information including at least one historical item information;

Determining, according to the at least one of the current item information and the at least one historical item information, whether the current billing information and the historical billing information correspond to the same bill;

If the current billing information and the historical billing information correspond to the same bill, the rejecting payment information is output.

In a second aspect, the embodiment of the present application provides a medical claim compensation device, including:

a claim application requesting module, configured to obtain a claim application request, where the claim application request includes a case ID and current bill information, where the current bill information includes at least one current item information;

a historical billing information obtaining module, configured to acquire historical billing information corresponding to the case ID based on the case ID, the historical billing information including at least one historical item information;

The same billing determining module is configured to determine, according to the at least one current item information and the at least one historical item information, whether the current billing information and the historical billing information correspond to the same bill;

The first reject payment information output module is configured to output the reject payment information when the current bill information and the historical bill information correspond to the same bill.

In a third aspect, an embodiment of the present application provides a terminal device, including a memory, a processor, and computer readable instructions stored in the memory and executable on the processor, where the processor executes the computer The following steps are implemented when reading the instruction:

In a fourth aspect, an embodiment of the present application provides a computer readable storage medium, where the computer readable storage medium stores computer readable instructions, and when the computer readable instructions are executed by a processor, the following steps are implemented:

In the medical claim refusal method, device, terminal device and storage medium provided by the embodiment of the present application, the current billing information and the historical billing information are obtained through the claim application request, and the historical billing information is obtained according to the case ID of the claim application request, thereby ensuring history Correspondence between billing information and current billing information. Based on current billing information and historical billing information It is judged whether the current bill and the historical bill correspond to the same bill, so that the judgment of whether the current bill has been compensated is more accurate. Based on the current billing information and the historical billing information, it is judged whether the current bill and the historical bill are corresponding to the same bill, so that the judgment of the same bill is more accurate. When the current bill and the historical bill correspond to the same bill, the current claim application request is rejected, and the reason for rejecting the payment is output, and the automatic billing function of the same bill claim application is realized, the processing efficiency of the claim application is improved, and the accuracy of the claim settlement is improved. .

DRAWINGS

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings used in the description of the embodiments of the present application will be briefly described below. It is obvious that the drawings in the following description are only some embodiments of the present application. Other drawings may also be obtained from those of ordinary skill in the art based on these drawings without the inventive labor.

1 is a flowchart of a method for claiming medical claims in the first embodiment of the present application;

FIG. 2 is a specific flow chart of step S10 of FIG. 1.

FIG. 3 is a specific flow chart of step S12 of FIG. 2.

Figure 4 is a specific flow chart of step S13 of Figure 2.

FIG. 5 is a specific flowchart of step S30 in FIG. 1.

6 is a schematic block diagram of a medical claim refusal device in Embodiment 2 of the present application;

FIG. 7 is a schematic diagram of a terminal device provided in Embodiment 4 of the present application.

Detailed ways

The technical solutions in the embodiments of the present application are clearly and completely described in the following with reference to the drawings in the embodiments of the present application. It is obvious that the described embodiments are a part of the embodiments of the present application, and not all of the embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments of the present application without departing from the inventive scope are the scope of the present application.

Example 1

FIG. 1 is a flow chart showing a method of medical claims refusal in the present embodiment. The medical claim refusal method is applied in the claim system of the insurance institution to improve the efficiency of claim processing of medical insurance and save labor costs. As shown in FIG. 1, the medical claim rejection method includes the following steps:

S10: Acquire a claim application request, the claim application request includes a case ID and current bill information, and the current bill information includes at least one current item information.

Wherein, the claim application request refers to a request for the claim system to perform the claim processing. The case ID is an identifier for uniquely identifying the case. The case ID includes, but is not limited to, the case number of the case to which the claims application request belongs, and the case number can be used to find all the bill information under the case to track the case. The current billing information refers to the detailed information of the current bill to be applied for in the claims application request. The current project information is the value of each specific item in the current bill. Specifically, the current billing information includes, but is not limited to, the billing number of the current bill, the date of admission, the amount of the bill, the type of treatment, and at least one item of information in the hospital. If customer A submits a claim application request through the claim system, after the claim system obtains the claim application request, it obtains all the item information of the current bill of customer A, for example, the current bill includes the bill number 134505, the admission date is 20170809, and the bill amount is 1399RMB, the treatment type is tetanus infection, and the hospital is Shenzhen Zhongchuan Hospital and other project information, so that the claim system can obtain all the bill information in the claim application request.

Specifically, the user may send a claim application request to the claim system through a client (including but not limited to a terminal such as a computer, a smart phone, and a tablet), so that the claim system obtains the claim application request. The user can input the case ID and input the current bill information on the claim page displayed on the client, and should click the confirmation submit button on the claim page to input the claim application request to the claim system. In this embodiment, the current billing information can be input not only by manual input, but also by intelligently identifying the image containing the current bill, and the intelligent identification method can effectively reduce the workload of the user and improve the efficiency of claim processing. Therefore, in a specific embodiment, as shown in FIG. 2, before the step S10 of the medical claim refusal method, the step of intelligently identifying the current billing information includes the following steps:

S11: Get the original billing image.

The original bill image is a billing image or a scanned bill image. The shot billing image is a billing image that is directly obtained after shooting by the shooting device statement. The scanned billing image is a billing image obtained by scanning the bill of the scanning device. The user may initiate a claim application request to the claim system by using the client, and the claim application request carries the original bill image.

S12: Extract the original bill image by using a single detector, a first convolutional layer, and a non-maximum suppression criterion to obtain a current bill image.

Among them, the Single Shot MultiBox Detector (SSD model) is a model that uses a single deep neural network model to achieve target detection and recognition. In this embodiment, the SSD model uses VGG-16 as the basic network, the first convolutional layer is the VGG-16 convolutional layer, and the VGG-16 convolutional layer includes 38*38*512 convolutional layer and 19*19*512 volume. Six layers of convolutional layers, such as 10*10*512 convolutional layer, 5*5*512 convolutional layer, 3*3*512 convolutional layer, 1*1*512 convolutional layer, etc. Corresponds to a detector & classifier. Non-maximum suppression criterion Suppression (hereinafter referred to as the NMS criterion) is an algorithm for suppressing an element that is not a maximum value, and searching for a local maximum value, that is, an algorithm that searches for a maximum value within a domain. The SSD model uses VGG-16 as the basic network, and combines the first convolutional layer and non-maximum suppression criteria to perform billing detection on the original billing image, and uses a rectangular box to mark the position of the bill in the original billing image to obtain the current billing image. Guarantee the efficiency and accuracy of the original bill image detection.

In a specific implementation, as shown in FIG. 3, step S12 specifically includes the following steps:

S121: Normalize the original bill image by using a trained single detector to obtain an initial bill image.

The initial bill image is a bill image obtained after the normalization processing operation. Before extracting the features of any billing image or scanned image, the SSD model for billing detection needs to be pre-trained to directly call the trained SSD model to normalize the original bill image during billing detection. To improve the efficiency and accuracy of billing detection. The normalization process refers to normalizing all the original bill images of the input SSD model into a uniform size, so that the obtained original bill image is more easily detected, and the efficiency and accuracy of billing detection are improved.

In the training process of the SSD model, the positive sample of the original bill image is determined according to the overlap ratio of the default box and the ground turth box, and then the training sample is expanded by operations such as cropping, mirroring, and noise addition. Get more training samples, and train the SSD model based on all the training samples obtained, and improve the accuracy of the trained SSD model for billing detection. Further, in the SSD model training process, all the training samples of the input SSD model need to be normalized to normalize all the training samples into a uniform size to improve the training efficiency of the SSD model.

The specific process of training the SSD model is as follows: First, obtain multiple original billing images for training the SSD model, and mark the real information (ground turth) of all the bills appearing in each original billing image, the real information (ground turth) includes billing number, admission date, bill amount, treatment type, and at least one item information in the hospital for treatment, and uses a rectangular box to identify the billing position to obtain a ground turth box. Then find the overlap ratio (intersection-over-union, hereinafter referred to as IOU) in the default box corresponding to the ground true box. The largest match with the true true box. sample. Then, in the remaining default box, find the default box with all the real box (ground true box) greater than the default value (0.5 in this embodiment) as the real box (ground True box) A matching positive sample. In this way, a default box of corresponding positive samples can be found for a labeled ground true box, which is beneficial to improve training efficiency. The overlap ratio IOU is the overlap ratio between the rectangular box and the default box of the bill position generated by the SSD model, and is used to evaluate the accuracy of the detection.

For the SSD model, according to the granularity of the extracted features of each layer of the VGG-16 network structure, the default box size corresponding to the feature granularity of the layer is determined, so that the default box of the SSD model is extracted. The strategy can cover the scale and position of most of the original billing images, and then according to the overlap ratio of the default box and the ground truth box, the positive and negative samples can be searched for training to obtain the trained SSD model. Therefore, when using the trained SSD model to extract the original bill image to detect the bill, only one feature extraction of the original bill image can obtain the feature map of all the training samples. Therefore, when using the SSD model for bill checking, Can effectively improve the detection efficiency.

S122: Perform multi-scale feature extraction on the initial bill image by using the first convolution layer, obtain a plurality of layer feature maps, and extract a plurality of layer feature maps by using a plurality of default boxes with different ratios to obtain a classification result of each default box.

In this embodiment, the initial bill image outputted by the trained SDD model is sequentially input into each convolution layer for multi-dimensional feature extraction, and a six-layer feature map is obtained, using 1, 1/2, 1/3, The default box of the six ratios 1/4, 2, and 3 extracts the six-level feature map so that the detector & classifier of each layer of the convolution layer outputs the corresponding classification result. Among them, the classification result output by each detector & classifier includes two quantities of category and confidence. It can be understood that a number of default boxes of different ratios are performed on the feature map after feature extraction of the input initial bill image, so that each convolution layer needs to be performed only once during feature extraction. Feature extraction is beneficial to improve feature extraction efficiency.

S123: Selecting the classification result of the default box by using the non-maximum suppression criterion to obtain the current billing image.

Specifically, in the initial bill image detection, a classifier is first created, the classifier gives a fixed size picture, and the classifier determines whether there is a bill in the initial bill image. The classifier is then converted into a detector, that is, a frame of multiple sizes (ie, a window) is generated by sliding a window or other manner on the initial bill image, and the size (ie, Resize) is adjusted to the fixed size, and then detected by the classifier. To output the current bill image, which is an image corresponding to the optimal frame selected from the plurality of frames using the NMS criterion. The sliding window method is used to generate multiple frames (each with a classifier score). The process of suppressing redundant frames by using the NMS criterion is as follows: the scores of all the frames are sorted in descending order, and the highest score and its corresponding box are selected; Traverse the remaining boxes, if the overlap rate (IOU) of the box with the current highest score is greater than a certain threshold, delete the box; continue to select the one with the highest score from the unprocessed box, and repeat the above process.

In this embodiment, after acquiring the feature map of each default box in step S122, the input detector & classifier uses Softmax to calculate the probability value of each default box belonging to all categories. Use the NMS criterion to select the category in which the probability value is the largest as the category of the default box; Traverse the default box. If the overlap ratio between the default box and the target box with the largest classification result is greater than a certain threshold, delete the corresponding default box and repeat the above process. Current billing image.

S13: The current billing image is identified by using the two-way long-term and short-term memory model, the second convolution layer and the translation layer, and the current billing information is obtained.

Among them, the Biddirectional Long Short-Term Memory (BLSTM model) has a forward LSTM and a reverse LSTM in the hidden layer. The forward LSTM captures the above feature information, while the reverse LSTM capture. The following feature information makes it possible to capture more feature information than a unidirectional LSTM. Therefore, the BLSTM model is better than the one-way LSTM model or the one-way RNN model. The second Convolutional Layers can be VGG-16 or other convolutional layers. The translation layer is used to process the character features recognized by the BLSTM. In this embodiment, the BLSTM model, the second convolution layer and the translation layer are used to perform image recognition on the current bill image, which is beneficial to improving the efficiency and accuracy of bill image recognition.

In a specific implementation, as shown in FIG. 4, step S13 specifically includes the following steps:

S131: Cutting the current bill image to obtain a plurality of strip feature maps.

Specifically, the current bill image is input, and the current bill image is switched in the vertical direction to form a plurality of strip feature maps. Among them, the number of strip feature maps depends on the length of the current bill image input. In this embodiment, the current bill image is segmented according to the pixel width of 1 to obtain a plurality of strip feature maps.

S132: Perform feature extraction on the plurality of strip feature maps by using the second convolution layer, and obtain a feature sequence formed by splicing the plurality of strip feature maps from left to right.

In this embodiment, the second Convolutional Layers may be VGG-16 or other convolutional layers, and the second convolutional layer is used to extract features of the input strip features, and convolution in the last layer. On all channels of the layer output, they are spliced column by column from left to right to form a sequence of features. In this embodiment, when the BLSTM model is used for identification, it is not necessary to separately separate each letter, number or text of the current bill image, but directly use the entire current bill image as an input to identify the current bill image. The principle of all characters is that the network structure of the BLSTM model can identify timing information, so that when multiple feature maps with timing information are input, the characters can be directly recognized. Therefore, how many strip features are stitched from left to right to form a sequence of features, depending on the network structure of the BLSTM model to identify the character requirements.

S133: Character recognition is performed on the feature sequence by using the two-way long-term and short-term memory model to obtain character features.

In this embodiment, the bidirectional long-term and short-term memory model (ie, the BLSTM model) is used as the cyclic network layer to perform character recognition on the feature sequence, and multiple features in the feature sequence are composed into one character, compared to the natural language recognition list. The circular network layer of words (such as one-way LSTM or one-way RNN), which directly acquires character features, can effectively improve the efficiency and accuracy of current bill image recognition.

S134: The character layer is processed by using a translation layer to obtain current bill information.

In this embodiment, the translation layer can process the character features recognized by the BLSTM to delete non-character features such as spaces, and then generate the final current bill information.

It can be understood that the claim system uses the BLSTM, the second convolution layer and the translation layer to identify the current bill image to obtain the current bill information corresponding to the current bill image, and the process does not need manual manual input to improve information input. Efficiency; and manual input errors can be avoided to ensure the quality of information input.

S20: Obtain historical billing information corresponding to the case ID based on the case ID, and the historical billing information includes at least one historical item information.

The historical billing information refers to the detailed information corresponding to the historical bill that the claim system has paid before the current time. Historical item information refers to the value of each specific item in any historical bill. Specifically, the historical item information also includes, but is not limited to, the bill number in the historical bill, the date of admission, the amount of the bill, the type of treatment, and at least one historical item information in the hospital. Among them, the bill number is an identifier for identifying each bill. It can be understood that the historical billing information is pre-stored in the database and associated with the case ID to indicate that the historical bill corresponding to the historical billing information has been paid for the case corresponding to the case ID, so as to prevent the insured from using the case. Historical billing information duplicate claims.

Since the case ID is the unique identifier of the case, the historical bill information stored in the database of the claims system can be queried and tracked by the case ID to avoid the occurrence of duplicate claims based on the same bill. Therefore, after obtaining the claim application request input by the user, the claim system may acquire the case ID based on the claim application request, and query and obtain the historical bill information existing in the database of the claim system according to the case ID. Specifically, after obtaining the case ID, the claim system can query whether the historical bill of the same case ID exists in the database of the claim system based on the case ID; if yes, obtain the historical bill information corresponding to the case ID. In this embodiment, the corresponding historical billing information can be obtained based on the case ID, so that the historical billing information is targeted, and the accuracy of the duplicate billing review of the same bill is improved.

In an embodiment, the client A submits a claim application request to the claim system through the client, and after the claim system obtains the claim application request, the case ID of the case for obtaining the claim according to the claim application request is 3432, and the claim system is based on the case. The ID queries the claim history. When the case where the case ID is 3432 is found to have the claim history, the historical bill information corresponding to the case ID is further obtained, and the historical bill information includes the bill number, the admission date, and the date of admission. Billing amount, type of treatment, and at least one historical item information in the hospital.

S30: Determine current bill information and calendar based on at least one current item information and at least one historical item information Whether the history bill information corresponds to the same bill.

The current billing information is formed as a whole by at least one current item information, and the historical billing information is formed integrally by at least one historical item information, and the current billing information is judged based on the at least one current item information and the at least one historical item information in step S30. And whether the historical billing information corresponds to the same bill, specifically comparing whether at least one current item information completely matches at least one historical item information, and if the two match exactly, the current bill and historical bill corresponding to the current billing information are determined The historical bill corresponding to the information is the same bill (ie, repeated billing). If you continue to pay based on the current billing information, it may cause property damage to the insurance institution. Among them, the duplicate bill refers to at least two bills with the same item information on the bill.

In a specific implementation, as shown in FIG. 5, step S30 specifically includes the following steps:

S31: Perform matching processing on the at least one current item information and the at least one historical item information by using the BF algorithm to determine whether the current item information and the historical item information are completely matched.

The BF algorithm is a pattern matching algorithm. The idea of the algorithm is to match the first character of the target string S with the first character of the pattern string T. If they are equal, continue to compare the second character of the target string S. And the second character of the pattern string T; if not equal, comparing the second character of the target string S with the first character of the pattern string T, and then comparing them until the final matching result is obtained, the algorithm is simple to implement , low complexity. After obtaining the current billing information in step S13 and obtaining the historical billing information in step S20, the BF algorithm is used to match any of the obtained current item information and the corresponding historical item information, according to whether any current item information and corresponding The historical item information is completely matched to determine whether the current item information and the historical item information are completely matched.

S32: If the current item information and the historical item information completely match, the current billing information and the historical billing information correspond to the same bill.

If any item information in the current item information completely matches the corresponding item information in the historical item information obtained based on the case ID, it indicates that the current bill has already had a claim payment before the current time, that is, the current bill information and the historical bill information correspond to the same The bill, the claim application request is a claim application request based on the same bill repeated application.

S33: If the current item information does not completely match the historical item information, the current billing information and the historical billing information do not correspond to the same bill.

If any item information in the current item information does not completely match at least one item information in the historical item information obtained based on the case ID of the current bill, it indicates that the current bill has not occurred before the current time, that is, the current bill information and history. The billing information does not correspond to the same bill, and the claim application request is not a claim application request based on the same bill repeated application.

S40: If the current billing information and the historical billing information correspond to the same bill, the rejecting payment information is output.

All current item information in the current billing information is completely consistent with all historical item information in the historical billing information obtained based on the case ID of the current bill, that is, the current bill and the historical bill correspond to the same bill, indicating that the current bill has already been settled before the current time. Therefore, the claim application request belongs to the duplicate claim request, and the claim system outputs the refusal payment information to avoid the loss of the property of the insurance institution caused by the repeated payment.

In a specific embodiment, the claim application request further includes current case information. Among them, the current case information and all the information related to the case corresponding to the case ID, including but not limited to the time of the case, the place where the case occurred, the cause of the case and the nature of the case. The medical claim refusal method further includes the following steps:

S50: If the current billing information and the historical billing information do not correspond to the same bill, obtain at least one chargeback rule based on the case ID, and determine whether the current case information meets at least one chargeback rule.

Among them, the chargeback rule is a rule for rejecting the payment determined by the insurance contract corresponding to the case ID. If all the current item information in the current billing information and all the historical item information in the historical billing information corresponding to the case ID are not completely consistent, the current bill and the historical bill do not correspond to the same bill, indicating that the current bill has not been claimed before the current time. At this time, the corresponding at least one chargeback rule is searched based on the case ID in the claim application request, and the chargeback rule includes, but is not limited to, the claim application request provided by the embodiment exceeds the insurance term, and the claim application request content does not conform to the contract. Regulations, etc.

When the claim system knows that the current bill information and the historical bill case information corresponding to the case ID do not correspond to the same order, it is necessary to further determine whether the current case information meets at least one chargeback rule to further determine whether the claim application request needs to be paid. To avoid false claims, the property of the insurance institution is lost.

S60: If the current case information meets at least one chargeback rule, the rejected claim information is output.

After the claim system determines that the current billing information meets at least one chargeback rule, the user's claim application request is rejected, and the rejected claim information is outputted to the claim description of the current case. The refusal payment information includes the reason for the refusal, the number of refusal rules for the current case information, and the number of refusal claims that the claims system will output when the claim for the current case is rejected. The multiple reasons for refusal to pay in the refusal information are presented in a spliced form to facilitate the staff or users within the insurance organization to understand the reason for the claim being rejected.

For example, when customer A proposes a claim application request, and the claim system determines that the current bill information in the claim application request does not correspond to the same bill as the historical bill information corresponding to the case ID, the claim system further detects the current case information corresponding to the case ID. Whether the claim for claim settlement exceeds the contract period and the content of the claims application does not satisfy the contract payment rules, the claim system outputs the rejection information, and the reason for the refusal in the rejection information is: “Bill ID3234 exceeds the contract The term; the bill ID3234 request content does not meet the contract requirements, and is not paid."

S70: If the current case information does not meet all the chargeback rules, the consent payment information is output.

If the current case information does not meet all the chargeback rules, it means that the same bill has not been paid for the same bill before the current time, and the current case information corresponding to the claim application request does not meet all the chargeback rules, and the claim system audits the claims through the current case. The application request, output consent payment information, so that the insurance institution staff can handle the payment based on the consent payment information.

In a specific embodiment, in order to ensure the security of the insurance claims, the claims processing process needs to be monitored. Therefore, the claim application request in the specific embodiment may further include a monitoring mailbox, which is a mailbox of the claim reviewer. The medical claims refusal method also includes:

S80: Sending the refusal payment information to the monitoring mailbox, and refusing the payment information includes at least one reason for refusal, and the reason for refusing corresponds to the refusal rule.

If the current case information meets at least one chargeback rule, it indicates that although the current case has not been paid claims based on the same bill, the current case information corresponding to the claim application request meets at least one chargeback claim and should be refused. Therefore, The output rejects the payment information to avoid the loss of the property of the insurance institution caused by the wrong refusal. The claim system sends the rejected refusal payment information to the monitoring mailbox, so that the claim reviewer of the monitoring mailbox understands the case refusal, wherein the refusal payment information includes at least one reason for refusal, and each refusal rule corresponds to a refusal reason. How many chargeback rules are met in the current case information, and the number of chargeback reasons that the claims system will include in the splicing form when the claim for the current case information is automatically rejected.

In the medical claim refusal method provided in this embodiment, the current billing information and the historical billing information are obtained through the claim application request, and the historical billing information is obtained according to the case ID of the claim application request, thereby ensuring the correspondence between the historical billing information and the current billing information. . Based on the current billing information and the historical billing information, it is more accurate to determine whether the current bill and the historical bill correspond to the same bill, so that the current bill has been judged whether the claim has occurred. When the current bill and the historical bill correspond to the same bill, the claim system rejects the claim for the claim, and outputs the chargeback information to avoid the duplicate claim, and informs the user that the claim request is a duplicate claim request. If the current bill and the historical bill do not correspond to the same bill, the current case ID is queried according to the current case ID to meet at least one chargeback rule in the claim system. If at least one chargeback rule is met, the claim system rejects the current claims application. And output the rejection information; if it does not meet any of the chargeback rules, enter the consent payment information. For the chargeback case, the rejection email is sent to the monitoring mailbox to monitor the claim application request. The medical claim refusal method realizes the automatic refusal function under the automatic refusal and refusal rules of the same bill, and improves the processing efficiency of the insurance institution for the claim application request.

It should be understood that the size of the sequence of the steps in the above embodiments does not mean that the order of execution is performed. The order of execution of each process should be determined by its function and internal logic, and should not be construed as limiting the implementation process of the embodiments of the present application.

Example 2

Fig. 6 is a block diagram showing the principle of the medical claim refusal device corresponding to the medical claim refusal method in the first embodiment. As shown in FIG. 6, the medical claim refusal device includes a claim application request acquisition module 10, a historical bill information acquisition module 20, a same bill determination module 30, and a first refusal information output module 40. The claim application request acquisition module 10, the historical bill information acquisition module 20, the same bill judgment module 30, the first reject payout information output module 40, the chargeback rule acquisition and judgment module 50, and the second reject payout information output module 60 The implementation functions of the consent payment information output module 70 and the rejection compensation information transmission module 80 correspond one-to-one with the steps corresponding to the medical claims rejection method in the embodiment.

The claim application requesting module 10 is configured to obtain a claim application request, where the claim application request includes a case ID and current bill information, and the current bill information includes at least one current item information.

The historical billing information obtaining module 20 is configured to obtain historical billing information corresponding to the case ID based on the case ID, and the historical billing information includes at least one historical item information.

The same billing determining module 30 is configured to determine whether the current billing information and the historical billing information correspond to the same bill based on the at least one current item information and the at least one historical item information.

The first reject payment information outputting module 40 is configured to output the reject payout information if the current billing information and the historical billing information correspond to the same bill.

Preferably, before the claim application request acquisition module 10, the medical claim rejection device further includes an original bill image acquisition unit 11, a current bill image acquisition unit 12, and a current bill information acquisition unit 13.

The original bill image obtaining unit 11 is configured to acquire an original bill image.

The current bill image obtaining unit 12 is configured to extract the original bill image by using a single detector, a first convolution layer, and a non-maximum suppression criterion to obtain a current bill image.

The current bill information obtaining unit 13 is configured to identify the current bill image by using the bidirectional long-term and short-term memory model, the second convolution layer, and the translation layer, and obtain current bill information.

Preferably, the current bill image acquisition unit 12 includes a normalization processing sub-unit 121, a feature extraction sub-unit 122, and a result selection sub-unit 123.

The normalization processing sub-unit 121 is configured to normalize the original bill image by using the trained single-shot detector to obtain an initial bill image.

The feature extraction sub-unit 122 is configured to perform multi-scale feature extraction on the initial bill image by using a convolution layer, obtain a plurality of layer feature maps, and extract a plurality of layer feature maps by using a plurality of default boxes with different ratios to obtain each default box. Classification results.

The result selecting sub-unit 123 is configured to select a classification result of the default box by using a non-maximum suppression criterion, Get the current billing image.

Preferably, the current bill image acquisition unit 13 includes a strip feature map acquisition sub-unit 131, a feature sequence acquisition sub-unit 132, a character feature acquisition sub-unit 133, and a current bill information acquisition sub-unit 134.

The strip feature map obtaining sub-unit 131 is configured to cut the current bill image to obtain a plurality of strip feature maps.

The feature sequence obtaining sub-unit 132 is configured to perform feature extraction on the plurality of strip-shaped feature maps by using the convolution layer, and obtain a feature sequence formed by splicing the plurality of strip-shaped feature maps from left to right.

The character feature acquisition sub-unit 133 is configured to perform character recognition on the feature sequence by using the bidirectional long-term and short-term memory model to acquire character features.

The current bill information obtaining sub-unit 134 is configured to process the character feature by using the translation layer to obtain current billing information.

Preferably, the same billing determination module 30 includes an information matching judging unit 31, the same billing determining unit 32, and a non-identical billing determining unit 33.

The information matching judging unit 31 is configured to perform matching processing on the at least one current item information and the at least one historical item information by using the BF algorithm to determine whether the current item information and the historical item information are completely matched.

The same billing determining unit 32 is configured to determine that the current billing information and the historical billing information correspond to the same bill when the current item information and the historical item information are completely matched.

The non-same billing determining unit 33 is configured to determine that the current billing information and the historical billing information do not correspond to the same bill when the current item information does not completely match the historical item information.

Preferably, the claim application request further includes current case information;

The medical claim refusal device further includes a chargeback rule acquisition and judgment module 50, a second refusal payout information output module 60, and a consent payout information output module 70.

The chargeback rule obtaining and determining module 50 is configured to: when the current billing information and the historical billing information do not correspond to the same bill, obtain at least one chargeback rule based on the case ID, and determine whether the current case information meets at least one chargeback rule.

The second rejection payment information output module 60 is configured to output the rejection payment information when the current case information meets at least one chargeback rule.

The consent payment information output module 70 is configured to output the consent payment information if the current case information does not meet all the chargeback rules.

Preferably, the claim application request further includes monitoring the mailbox;

The medical claim refusal device further includes a refusal payment information sending module 80, configured to send the refusal payment information to the monitoring mailbox, and the refusal payment information includes at least one reason for refusal, and the refusal reason corresponds to the refusal rule.

Example 3

The embodiment provides a computer readable storage medium, where the computer readable storage medium is stored by the processor, and the medical claim compensation method in Embodiment 1 is implemented. I won't go into details here. Alternatively, when the computer readable instructions are executed by the processor, the functions of the modules/units in the medical claims refusal device in Embodiment 2 are implemented. To avoid repetition, details are not described herein again.

Example 4

FIG. 7 is a schematic diagram of a terminal device according to an embodiment of the present application. As shown in FIG. 7, the terminal device 90 of this embodiment includes a processor 91, a memory 92, and computer readable instructions 93 stored in the memory 92 and operable on the processor 91. The processor 91 implements the various steps of the medical claims rejection method of the embodiment 1 when the computer readable instructions 93 are executed, such as steps S10, S20, S30, S40, S50, S60, S70, and S80 shown in FIG. Alternatively, when the processor 91 executes the computer readable instructions 93, the functions of the modules/units in the embodiment 2 are implemented, for example, the claim application request acquisition module 10, the historical bill information acquisition module 20, and the same bill determination module 30 shown in FIG. The functions of the first reject payout information output module 40, the chargeback rule acquisition and judgment module 50, the second reject payout information output module 60, the consent payout information output module 70, and the reject payout information sending module 80.

Illustratively, computer readable instructions 93 may be partitioned into one or more modules/units, one or more modules/units being stored in memory 92 and executed by processor 91 to complete the application. The one or more modules/units can be a series of computer readable instruction segments capable of performing a particular function for describing the execution of computer readable instructions 93 in the terminal device 90. For example, the computer readable instructions 93 may be divided into a claims application request acquisition module 10, a historical bill information acquisition module 20, a same bill determination module 30, and a first reject payout information output module 40.

The terminal device 90 can be a computing device such as a desktop computer, a notebook, a palmtop computer, and a cloud server. The terminal device may include, but is not limited to, a processor 91, a memory 92. It will be understood by those skilled in the art that FIG. 7 is merely an example of the terminal device 90 and does not constitute a limitation of the terminal device 90, and may include more or less components than those illustrated, or may combine certain components or different components. For example, the terminal device may further include an input/output device, a network access device, a bus, and the like.

The processor 91 may be a central processing unit (CPU), or may be another general-purpose processor, a digital signal processor (DSP), or an application specific integrated circuit (ASIC). Field-Programmable Gate Array (FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware components, etc. The general purpose processor may be a microprocessor or the processor or any conventional processor or the like.

The memory 92 may be an internal storage unit of the terminal device 90, such as a hard disk or a memory of the terminal device 90. The memory 92 may also be an external storage device of the terminal device 90, such as a plug-in hard disk equipped with the terminal device 90, a smart memory card (SMC), a Secure Digital (SD) card, and a flash memory card (Flash). Card) and so on. Further, the memory 92 may also include both an internal storage unit of the terminal device 90 and an external storage device. Memory 92 is used to store computer readable instructions as well as other programs and data required by the terminal device. The memory 92 can also be used to temporarily store data that has been output or is about to be output.

It will be apparent to those skilled in the art that, for convenience and brevity of description, only the division of each functional unit and module described above is exemplified. In practical applications, the above functions may be assigned to different functional units as needed. The module is completed by dividing the internal structure of the device into different functional units or modules to perform all or part of the functions described above.

In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

The integrated modules/units, if implemented in the form of software functional units and sold or used as separate products, may be stored in a computer readable storage medium. Based on such understanding, the present application implements all or part of the processes in the foregoing embodiments, and may also be implemented by computer readable instructions, which may be stored in a computer readable storage medium. The computer readable instructions, when executed by a processor, may implement the steps of the various method embodiments described above. Wherein, the computer readable instructions comprise computer readable instruction code, which may be in the form of source code, an object code form, an executable file or some intermediate form or the like. The computer readable medium can include any entity or device capable of carrying the computer readable instruction code, a recording medium, a USB flash drive, a removable hard drive, a magnetic disk, an optical disk, a computer memory, a read only memory (ROM, Read-Only) Memory), random access memory (RAM), electrical carrier signals, telecommunications signals, and software distribution media. It should be noted that the content contained in the computer readable medium may be appropriately increased or decreased according to the requirements of legislation and patent practice in a jurisdiction, for example, in some jurisdictions, according to legislation and patent practice, computer readable media It does not include electrical carrier signals and telecommunication signals.

The above-mentioned embodiments are only used to explain the technical solutions of the present application, and are not limited thereto; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still implement the foregoing embodiments. The technical solutions described in the examples are modified or equivalently replaced with some of the technical features; and the modifications or substitutions do not deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in Within the scope of protection of this application.

Claims

A medical claim refusal method, characterized in that it comprises:

Obtaining a claim application request, the claim application request includes a case ID and current bill information, and the current bill information includes at least one current item information;

Obtaining historical billing information corresponding to the case ID based on the case ID, the historical billing information including at least one historical item information;

Determining, according to the at least one of the current item information and the at least one historical item information, whether the current billing information and the historical billing information correspond to the same bill;

If the current billing information and the historical billing information correspond to the same bill, the rejecting payment information is output.
The medical claim refusal method according to claim 1, wherein before the step of obtaining the claim application request, the medical claim refusal method further comprises:

Get the original billing image;

Extracting the original bill image by using a single detector, a first convolutional layer, and a non-maximum suppression criterion to obtain a current bill image;

The current billing image is identified by using a bidirectional long-term and short-term memory model, a second convolutional layer, and a translation layer to obtain the current billing information.
The medical claim rejection method according to claim 2, wherein said extracting said original bill image using a single detector, a first convolutional layer, and a non-maximum suppression criterion to obtain a current bill Images, including:

Normalizing the original bill image by using the trained single detector to obtain an initial bill image;

Performing multi-scale feature extraction on the initial billing image by using the first convolution layer to obtain a plurality of layer feature maps, and extracting the feature maps of the plurality of layers by using a plurality of default boxes with different ratios to obtain each of the The classification result of the default box;

The non-maximum suppression criterion is used to select a classification result of the default box to obtain a current billing image.
The medical claim rejection method according to claim 2, wherein said identifying a current bill image by using a two-way long-term memory model, a second convolution layer, and a translation layer, and obtaining the current bill Information, including:

Cutting the current bill image to obtain a plurality of strip feature maps;

Extracting a plurality of the strip feature maps by using the second convolution layer to obtain a plurality of the strip features a sequence of features spliced from left to right;

Performing character recognition on the feature sequence by using the two-way long-term and short-term memory model to obtain character features;

The character feature is processed by the translation layer to obtain the current bill information.
The medical claim refund method according to claim 1, wherein said determining whether said current bill information and said historical bill information correspond to the same account based on at least one of said current item information and at least one history item information Single, including:

Performing matching processing on at least one of the current item information and the at least one historical item information by using a BF algorithm to determine whether the current item information and the historical item information are completely matched;

If the current item information completely matches the historical item information, determining that the current bill information and the historical bill information correspond to the same bill;

If the current item information does not completely match the historical item information, it is determined that the current bill information and the historical bill information do not correspond to the same bill.
The medical claim refusal method according to claim 1, wherein the claim application request further includes current case information, and the medical claim refusal method further comprises:

If the current billing information and the historical billing information do not correspond to the same bill, obtaining at least one chargeback rule based on the case ID, determining whether the current case information meets at least one of the chargeback rules;

If the current case information meets at least one of the chargeback rules, outputting the rejection information;

If the current case information does not meet all the chargeback rules, the consent payment information is output.
The medical claim refund method according to claim 6, wherein the claim application request further comprises monitoring a mailbox;

The medical claim refusal method further includes: transmitting the refusal payment information to the monitoring mailbox, the refusal payment information includes at least one refusal reason, and the refusal reason corresponds to the refusal rule.
A medical claims refusal device, comprising:

a claim application requesting module, configured to obtain a claim application request, where the claim application request includes a case ID and current bill information, where the current bill information includes at least one current item information;

a historical billing information obtaining module, configured to acquire historical billing information corresponding to the case ID based on the case ID, the historical billing information including at least one historical item information;

The same billing determining module is configured to determine, according to the at least one current item information and the at least one historical item information, whether the current billing information and the historical billing information correspond to the same bill;

a first reject payment information output module, configured to: the current bill information and the historical bill information correspond to the same When the bill is posted, the output rejects the payment information.
A terminal device comprising a memory, a processor, and computer readable instructions stored in the memory and operable on the processor, wherein the processor executes the computer readable instructions as follows step:

Obtaining a claim application request, the claim application request includes a case ID and current bill information, and the current bill information includes at least one current item information;

Obtaining historical billing information corresponding to the case ID based on the case ID, the historical billing information including at least one historical item information;

Determining, according to the at least one of the current item information and the at least one historical item information, whether the current billing information and the historical billing information correspond to the same bill;

If the current billing information and the historical billing information correspond to the same bill, the rejecting payment information is output.
The terminal device according to claim 9, wherein said computer readable instructions are further executed by said processor before said step of obtaining a claim application request further comprises the steps of:

Get the original billing image;

Extracting the original bill image by using a single detector, a first convolutional layer, and a non-maximum suppression criterion to obtain a current bill image;

The current billing image is identified by using a bidirectional long-term and short-term memory model, a second convolutional layer, and a translation layer to obtain the current billing information.
The terminal device according to claim 10, wherein said extracting said original bill image by a single detector, a first convolution layer and a non-maximum suppression criterion, obtaining a current bill image, including :

Normalizing the original bill image by using the trained single detector to obtain an initial bill image;

Performing multi-scale feature extraction on the initial billing image by using the first convolution layer to obtain a plurality of layer feature maps, and extracting the feature maps of the plurality of layers by using a plurality of default boxes with different ratios to obtain each of the The classification result of the default box;

The non-maximum suppression criterion is used to select a classification result of the default box to obtain a current billing image.
The terminal device according to claim 10, wherein said identifying a current billing image by using a bidirectional long-term and short-term memory model, a second convolutional layer, and a translation layer, and obtaining the current billing information, including :

Cutting the current bill image to obtain a plurality of strip feature maps;

Feature extraction of a plurality of the strip feature maps by using the second convolution layer, and acquiring a feature sequence formed by splicing a plurality of the strip feature maps from left to right;

Performing character recognition on the feature sequence by using the two-way long-term and short-term memory model to obtain character features;

The character feature is processed by the translation layer to obtain the current bill information.
The terminal device according to claim 9, wherein the determining whether the current bill information and the historical bill information correspond to the same bill based on the at least one of the current item information and the at least one history item information comprises :

Performing matching processing on at least one of the current item information and the at least one historical item information by using a BF algorithm to determine whether the current item information and the historical item information are completely matched;

If the current item information completely matches the historical item information, determining that the current bill information and the historical bill information correspond to the same bill;

If the current item information does not completely match the historical item information, it is determined that the current bill information and the historical bill information do not correspond to the same bill.
The terminal device according to claim 9, wherein the claim application request further includes current case information, and when the computer readable instructions are executed by the processor, the following steps are further implemented:

If the current billing information and the historical billing information do not correspond to the same bill, obtaining at least one chargeback rule based on the case ID, determining whether the current case information meets at least one of the chargeback rules;

If the current case information meets at least one of the chargeback rules, outputting the rejection information;

If the current case information does not meet all the chargeback rules, the consent payment information is output.
A computer readable storage medium storing computer readable instructions, wherein the computer readable instructions, when executed by a processor, implement the following steps:

Obtaining a claim application request, the claim application request includes a case ID and current bill information, and the current bill information includes at least one current item information;

Obtaining historical billing information corresponding to the case ID based on the case ID, the historical billing information including at least one historical item information;

Determining, according to the at least one of the current item information and the at least one historical item information, whether the current billing information and the historical billing information correspond to the same bill;

If the current billing information and the historical billing information correspond to the same bill, the rejecting payment information is output.
A computer readable storage medium as recited in claim 15, wherein said computer readable instructions are further executed by said processor prior to said step of obtaining a claim request request:

Get the original billing image;

Extracting the original bill image by using a single detector, a first convolutional layer, and a non-maximum suppression criterion to obtain a current bill image;

The current billing image is identified by using a bidirectional long-term and short-term memory model, a second convolutional layer, and a translation layer to obtain the current billing information.
The computer readable storage medium of claim 16 wherein said extracting said original billing image using a single detector, first convolutional layer, and non-maximum suppression criteria to obtain current bill Images, including:

Normalizing the original bill image by using the trained single detector to obtain an initial bill image;

Performing multi-scale feature extraction on the initial billing image by using the first convolution layer to obtain a plurality of layer feature maps, and extracting the feature maps of the plurality of layers by using a plurality of default boxes with different ratios to obtain each of the The classification result of the default box;

The non-maximum suppression criterion is used to select a classification result of the default box to obtain a current billing image.
The computer readable storage medium of claim 16 wherein said identifying a current billing image by using a bidirectional long and short term memory model, a second convolutional layer, and a translation layer, obtaining said current bill Information, including:

Cutting the current bill image to obtain a plurality of strip feature maps;

Feature extraction of a plurality of the strip feature maps by using the second convolution layer, and acquiring a feature sequence formed by splicing a plurality of the strip feature maps from left to right;

Performing character recognition on the feature sequence by using the two-way long-term and short-term memory model to obtain character features;

The character feature is processed by the translation layer to obtain the current bill information.
The computer readable storage medium according to claim 15, wherein said determining whether said current bill information and said historical billing information correspond to the same account based on at least one of said current item information and at least one history item information Single, including:

Performing matching processing on at least one of the current item information and the at least one historical item information by using a BF algorithm to determine whether the current item information and the historical item information are completely matched;

If the current item information completely matches the historical item information, determining that the current bill information and the historical bill information correspond to the same bill;

If the current item information does not completely match the historical item information, determining the current bill information and the location The historical billing information does not correspond to the same bill.
The computer readable storage medium of claim 15 wherein said claims request further comprises current case information, said computer readable instructions being further executed by said processor to:

If the current billing information and the historical billing information do not correspond to the same bill, obtaining at least one chargeback rule based on the case ID, determining whether the current case information meets at least one of the chargeback rules;

If the current case information meets at least one of the chargeback rules, outputting the rejection information;

If the current case information does not meet all the chargeback rules, the consent payment information is output.