Medical list correction method, device and equipment and computer readable storage medium

Abstract

The present application relates to the field of medical technology data processing, and discloses a medical list error correction method, comprising: performing row-column segmentation on a to-be-corrected medical list to obtain a first medical list, constructing form structure features of the first medical list and cell structure features of each cell, calculating structure feature differences between the cell structure features and the form structure features, identifying a to-be-corrected cell according to the structure feature differences, performing structure correction on the to-be-corrected cell to obtain a second medical list, constructing context contexts of each cell in the second medical list, generating predicted texts of the corresponding cells according to the context contexts, calculating text differences between the predicted texts of each cell and real texts, identifying a to-be-corrected cell according to the text differences, and performing text correction on the to-be-corrected cell. The present application also provides a medical list error correction device, an electronic device and a computer readable storage medium. The present application can improve the accuracy of medical list error correction.

Description

Technical Field

[0001] This invention relates to the field of medical technology data processing, and in particular to a method, apparatus, electronic device, and computer-readable storage medium for correcting errors in medical lists. Background Technology

[0002] Medical bills contain a wealth of information, such as patient personal details, medical history, physical examination data, symptoms, and test results. Consequently, their table structures are often complex, including merged cells, nested tables, and multi-level headers.

[0003] The commonly used OCR (Optical Character Recognition) technology cannot handle the complex table structure well when recognizing the above medical list, which easily leads to problems such as misalignment or misaligned rows in the recognition results. At the same time, the existing OCR technology has limited ability in context understanding and error correction. In complex tables, especially when there are misaligned rows, the existing OCR technology cannot understand the overall semantics and structure of the table well, and the accuracy of error correction for medical lists is not high. Summary of the Invention

[0004] This invention provides a medical billing error correction method, device, electronic device, and computer-readable storage medium, the main purpose of which is to improve the accuracy of medical billing error correction.

[0005] To achieve the above objectives, the present invention provides a medical inventory error correction method, comprising:

[0006] Perform a row and column splitting operation on the medical list to obtain the first medical list;

[0007] Construct the form structure features of the first medical list and the cell structure features of each cell in the first medical list;

[0008] Calculate the structural feature difference between the cell structural features of each cell and the form structural features;

[0009] The cells to be corrected are identified based on the structural feature differences, and structural correction is performed on the cells to be corrected to obtain the second medical list.

[0010] Construct the contextual information for each cell in the second medical list, and generate the predicted text for the corresponding cell based on the contextual information;

[0011] Calculate the text difference between the predicted text and the actual text in each cell of the second medical list;

[0012] The cells to be corrected are identified based on the text difference, and the text in the cells to be corrected is corrected to obtain the corrected medical list.

[0013] Optionally, the medical list to be corrected is subjected to row and column splitting to obtain a first medical list, including:

[0014] Edge detection is performed on the medical list to be corrected to obtain a grid structure corresponding to the medical list to be corrected.

[0015] Feature extraction is performed on each grid in the grid structure to obtain grid features.

[0016] Using machine learning algorithms, the row and column position information of the corresponding grid is identified based on the grid features;

[0017] Based on the row and column position information of each grid, each grid is split into cells to obtain the first medical list.

[0018] Optionally, the form structure features for constructing the first medical list include:

[0019] Based on the preset table dimensions, obtain the value of each preset table dimension in the first medical list;

[0020] The values ​​of each of the preset table dimensions are normalized and encoded to obtain the table dimension codes;

[0021] The form structure features of the first medical list are obtained by matrix-concatenating the dimensional codes of each table.

[0022] Optionally, calculating the structural feature difference between the cell structural feature of each cell and the form structural feature includes:

[0023] The cell structure features of each cell are transformed into a vector to obtain a cell vector;

[0024] The form structure features are then transformed into a vector to obtain a form vector;

[0025] Calculate the distance between the cell vector of each cell and the form vector, and use the calculated distance as the structural feature difference.

[0026] Optionally, the step of constructing the contextual information of each cell in the second medical list and generating the predicted text for the corresponding cell based on the contextual information includes:

[0027] Get the text content and position information of each cell;

[0028] Using a pre-trained language model, the context of each cell is generated based on the text content and position information of each cell.

[0029] Select one cell from the second medical list as the target cell;

[0030] Based on the context of the target cell, the predicted text of the target cell is generated using the pre-trained language model.

[0031] Optionally, before performing row and column splitting operations on the medical list to be corrected, the method further includes:

[0032] Image processing techniques are used to perform image enhancement and noise reduction on the medical list to be corrected.

[0033] The distorted regions in the medical list to be corrected are identified using a table detection algorithm.

[0034] The distorted region is corrected using a table correction algorithm.

[0035] To address the above problems, the present invention also provides a medical billing error correction device, the device comprising:

[0036] The row and column splitting module is used to perform row and column splitting operations on the medical list to be corrected, so as to obtain the first medical list;

[0037] The structural feature extraction module is used to construct the form structural features of the first medical list and the cell structural features of each cell in the first medical list.

[0038] The structural error correction module is used to calculate the structural feature difference between the cell structural features of each cell and the form structural features, identify the cells to be corrected based on the structural feature difference, and perform structural error correction on the cells to be corrected to obtain the second medical list.

[0039] The context generation module is used to construct the context of each cell in the second medical list and generate the predicted text for the corresponding cell based on the context.

[0040] The context correction module is used to calculate the text difference between the predicted text and the actual text of each cell in the second medical list, identify the cells to be corrected based on the text difference, and perform text correction on the cells to be corrected to obtain the corrected medical list.

[0041] Optionally, the row and column splitting module performs row and column splitting operations from the medical list to be corrected using the following method:

[0042] Edge detection is performed on the medical list to be corrected to obtain a grid structure corresponding to the medical list to be corrected.

[0043] Feature extraction is performed on each grid in the grid structure to obtain grid features.

[0044] Using machine learning algorithms, the row and column position information of the corresponding grid is identified based on the grid features;

[0045] Based on the row and column position information of each grid, each grid is split into cells to obtain the first medical list.

[0046] To address the above problems, the present invention also provides an electronic device, the electronic device comprising:

[0047] Memory, storing at least one computer program; and

[0048] The processor executes the program stored in the memory to implement the medical inventory error correction method described above.

[0049] To address the aforementioned problems, the present invention also provides a computer-readable storage medium storing at least one computer program, which is executed by a processor in an electronic device to implement the aforementioned medical list error correction method.

[0050] This invention applies to construct the table structure features and cell structure features of a medical list to be corrected. By utilizing the structural feature difference between the table structure features and cell structure features, cells to be corrected are identified, and structural correction is performed on these cells. Furthermore, by constructing the context of the cells, predicted text for each cell is generated, and then the text difference between the predicted text and the actual text is used to perform text correction on the cells to be corrected. Through the above method, structural and text correction can be performed on the medical list to be corrected from both structural and contextual perspectives, thereby improving the accuracy of medical list correction. Attached Figure Description

[0051] Figure 1 A flowchart illustrating a medical billing error correction method according to an embodiment of the present invention;

[0052] Figure 2 A detailed implementation flowchart of one step of the medical list error correction method provided in an embodiment of the present invention is shown below;

[0053] Figure 3 A detailed implementation flowchart of one step of the medical list error correction method provided in an embodiment of the present invention is shown below;

[0054] Figure 4A detailed implementation flowchart of one step of the medical list error correction method provided in an embodiment of the present invention is shown below;

[0055] Figure 5 A functional block diagram of a medical billing error correction device provided in an embodiment of the present invention;

[0056] Figure 6 This is a schematic diagram of the structure of an electronic device for implementing the medical list error correction method according to an embodiment of the present invention.

[0057] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0058] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0059] This application provides a method for correcting errors in a medical bill of contents. The executing entity of this method includes, but is not limited to, at least one of the following: a server, a terminal, or other electronic devices configured to execute the method provided in this application. In other words, the method can be executed by software or hardware installed on a terminal device or a server device; the software may be a blockchain platform. The server may be an independent server or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0060] Reference Figure 1 The diagram shown is a flowchart illustrating a medical bill error correction method according to an embodiment of the present invention. In this embodiment, the medical bill error correction method includes:

[0061] S1. Perform row and column splitting operations on the medical list to be corrected to obtain the first medical list;

[0062] In this embodiment of the invention, the medical list to be corrected refers to an electronic medical list that has been identified by technologies such as OCR, including but not limited to patient hospitalization lists, patient examination lists, and patient treatment lists.

[0063] Understandably, due to the often complex table structure of medical lists, electronic medical lists recognized using OCR technology may exhibit table tilting, bending, or distortion. Therefore, before performing row and column segmentation on the medical list to be corrected, the following preprocessing operations can be performed to denoise or enhance the list, thereby improving the clarity of the table lines and correcting tilted or distorted areas to reduce the occurrence of misaligned rows. These operations include:

[0064] Image processing techniques are used to perform image enhancement and noise reduction on the medical list to be corrected.

[0065] The distorted regions in the medical list to be corrected are identified using a table detection algorithm.

[0066] The distorted region is corrected using a table correction algorithm.

[0067] In this embodiment of the invention, the image processing techniques include, but are not limited to, image enhancement techniques, image binarization techniques, and image distortion removal techniques.

[0068] In this embodiment of the invention, table detection algorithms and table correction algorithms based on machine learning and vision technologies can be used to correct the distorted area, making the table lines in the distorted area horizontal or vertical, thereby reducing the occurrence of misaligned lines.

[0069] For details, please refer to Figure 2 As shown, S1 includes:

[0070] S11. Perform edge detection on the medical list to be corrected to obtain a grid structure corresponding to the medical list to be corrected;

[0071] S12. Extract features from each grid in the grid structure to obtain grid features.

[0072] S13. Using machine learning algorithms, the row and column position information of the corresponding grid is identified based on the grid features;

[0073] S14. Based on the row and column position information of each grid, split each grid into cells to obtain the first medical list.

[0074] In this embodiment of the invention, an edge detection algorithm is used to scan the image of the medical list to be corrected, identify the start and end positions of each row or column, and obtain the grid structure corresponding to the medical list to be corrected.

[0075] In this embodiment of the invention, the grid features include, but are not limited to, features such as grid size and position.

[0076] In this embodiment of the invention, machine learning algorithms such as support vector machines, decision trees, or neural networks can be used to identify the row and column position information of the corresponding grid based on the grid features.

[0077] Understandably, each grid may contain merged cells, such as merging several rows or columns. Therefore, each grid can be split based on its row and column position information, adjusting a grid into one or more cells.

[0078] In this embodiment of the invention, by dividing the medical list to be corrected into rows and columns, the object to be corrected can be decomposed from the medical list into multiple cells composed of row and column information, which is beneficial for correcting errors based on each cell, thereby improving the accuracy of error correction.

[0079] S2. Construct the form structure features of the first medical list and the cell structure features of each cell in the first medical list;

[0080] In this embodiment of the invention, the form structure features refer to the overall distribution features of the first medical list, including but not limited to header distribution area features, form content distribution area features, built-in form features, number of rows and columns of the table, etc. The header distribution area features also include features such as row headers, list headers, and cascading headers.

[0081] In this embodiment of the invention, the cell structure features refer to the position of each cell and the positional and textual data relationship features between adjacent cells, such as the number of rows and columns of the cell and the header features corresponding to the cell.

[0082] For details, please refer to Figure 3 The form structure features for constructing the first medical list include:

[0083] S21. Based on the preset table dimensions, obtain the value of each preset table dimension in the first medical list;

[0084] S22. Normalize the values ​​of each of the preset table dimensions to obtain the table dimension codes;

[0085] S23. Perform matrix concatenation on the dimensional codes of each table to obtain the form structure features of the first medical list.

[0086] In this embodiment of the invention, the preset table dimensions include, but are not limited to, the overall size of the table area, the size of the header area, the size of the text content area, and dimensions such as the data type, data format, or data relationship in the table.

[0087] In this embodiment of the invention, normalized encoding of the values ​​of each preset table dimension can be achieved by performing hotspot encoding with the same number of bits on the values ​​of each preset table dimension.

[0088] In this embodiment of the invention, the cell structure features of each cell in the first medical list can be constructed using preset cell dimensions. The preset cell dimensions can be set according to the content features of the actual medical list. For example, the preset cell dimensions include, but are not limited to, cell position, the header position corresponding to the cell, and the data type in the cell.

[0089] It should be noted that the method for constructing the cell structure features of each cell in the first medical list is the same as the method for constructing the form structure features of the first medical list, and will not be repeated here.

[0090] The purpose of constructing the form structure features of the first medical list and the cell structure features of each cell in the first medical list in this embodiment of the invention is to use the form structure features and cell structure features as error correction dimensions to correct the medical list to be corrected from the perspective of structural features.

[0091] S3. Calculate the structural feature difference between the cell structural feature of each cell and the form structural feature, identify the cells to be corrected based on the structural feature difference, and perform structural correction on the cells to be corrected to obtain the second medical list.

[0092] Under normal circumstances, the cell structure features of each cell in the first medical list basically conform to the form structure features of the first medical list. For example, the header features of each cell are consistent with the header features of the corresponding form structure features, and the positional features of each cell, consisting of the number of rows and columns, conform to the form distribution features in the corresponding form structure features. Therefore, when the structural feature difference between the cell structure features of a certain cell and the corresponding form structure features is relatively large, it indicates that the cell may contain errors and needs to be corrected.

[0093] Specifically, calculating the structural feature difference between the cell structural features of each cell and the form structural features includes:

[0094] The cell structure features of each cell are transformed into a vector to obtain a cell vector;

[0095] The form structure features are then transformed into a vector to obtain a form vector;

[0096] Calculate the distance between the cell vector of each cell and the form vector, and use the calculated distance as the structural feature difference.

[0097] In this embodiment of the invention, the distance between the cell vector and the form vector can be calculated using the Mahalanobis distance algorithm or the Euclidean distance algorithm.

[0098] Furthermore, when the structural feature difference is greater than a preset structural feature difference threshold, the corresponding cell is designated as a cell to be corrected, and a table correction algorithm can be used to correct the cell to be corrected.

[0099] In this embodiment of the invention, the preset structural feature difference threshold can be set according to the actual business situation.

[0100] This invention utilizes the structural feature difference between cell structure features and form structure features to identify cells that need correction. By correcting these cells, the accuracy of the second medical list is further improved compared to the first medical list.

[0101] S4. Construct the contextual information of each cell in the second medical list, and generate the predicted text of the corresponding cell based on the contextual information;

[0102] In this embodiment of the invention, an artificial intelligence-based language model, such as a neural network model, can be used to construct the context of each cell and generate the predicted text for each cell.

[0103] For details, please refer to Figure 4 As shown, S4 includes:

[0104] S41. Obtain the text content and position information of each cell;

[0105] S42. Using a pre-trained language model, generate the contextual information corresponding to each cell based on the text content and position information of each cell.

[0106] S43. Select one cell from the second medical list as the target cell;

[0107] S44. Based on the context of the target cell, generate the predicted text of the target cell using the pre-trained language model.

[0108] In this embodiment of the invention, the pre-trained language model can be either a deep learning-based convolutional neural network model or a ChatGPT model.

[0109] In this embodiment of the invention, predicted text for a cell is generated based on the context of the cell, and the accuracy of the text content of the corresponding cell can be judged using the predicted text.

[0110] S5. Calculate the text difference between the predicted text and the actual text of each cell in the second medical list, identify the cells to be corrected based on the text difference, and perform text correction on the cells to be corrected to obtain the corrected medical list.

[0111] In this embodiment of the invention, the actual text of each cell in the second medical list can be obtained. The text difference is obtained by comparing the predicted text with the actual text and quantifying the result of the comparison. The text difference is then used to identify cells requiring correction.

[0112] In another optional embodiment of the present invention, the text difference can be obtained by performing vector transformations on the predicted text and the real text respectively, and calculating the vector distance between the two.

[0113] In another optional embodiment of the present invention, the semantics of the predicted text and the semantics of the real text can be analyzed using the pre-trained language model, and the text difference can be obtained by comparing the semantic similarity between the two.

[0114] In this embodiment of the invention, when the text difference is greater than a preset text difference threshold, the corresponding cell is taken as the cell to be corrected, and the predicted text can be used to replace the real text of the corresponding cell, thereby realizing the text correction of the cell to be corrected.

[0115] In this embodiment of the invention, the preset text difference threshold can be set according to the actual business situation.

[0116] This invention applies to construct the table structure features and cell structure features of a medical list to be corrected. By utilizing the structural feature difference between the table structure features and cell structure features, cells to be corrected are identified, and structural correction is performed on these cells. Furthermore, by constructing the context of the cells, predicted text for each cell is generated, and then the text difference between the predicted text and the actual text is used to perform text correction on the cells to be corrected. Through the above method, structural and text correction can be performed on the medical list to be corrected from both structural and contextual perspectives, thereby improving the accuracy of medical list correction.

[0117] like Figure 5 The diagram shown is a functional block diagram of a medical billing error correction device provided in an embodiment of the present invention.

[0118] The medical list error correction device 100 of the present invention can be installed in an electronic device. Depending on the functions implemented, the medical list error correction device 100 includes a row and column segmentation module 101, a structural feature extraction module 102, a structural error correction module 103, a context generation module 104, and a context correction module 105. The modules described in this invention can also be called units, referring to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, stored in the memory of the electronic device.

[0119] In this embodiment, the functions of each module / unit are as follows:

[0120] The row and column segmentation module 101 is used to perform row and column segmentation operations on the medical list to be corrected, so as to obtain the first medical list.

[0121] The structural feature extraction module 102 is used to construct the form structural features of the first medical list and the cell structural features of each cell in the first medical list.

[0122] The structure correction module 103 is used to calculate the structural feature difference between the cell structure feature of each cell and the form structure feature, identify the cell to be corrected based on the structural feature difference, and perform structure correction on the cell to be corrected to obtain the second medical list.

[0123] The context generation module 104 is used to construct the context of each cell in the second medical list and generate the predicted text of the corresponding cell based on the context.

[0124] The context correction module 105 is used to calculate the text difference between the predicted text and the real text of each cell in the second medical list, identify the cell to be corrected based on the text difference, and perform text correction on the cell to be corrected to obtain the corrected medical list.

[0125] In detail, the specific implementation methods of each module of the medical billing error correction device 100 are as follows:

[0126] Step 1: Perform a row and column splitting operation on the medical list to be corrected to obtain the first medical list;

[0127] In this embodiment of the invention, the medical list to be corrected refers to an electronic medical list that has been identified by technologies such as OCR, including but not limited to patient hospitalization lists, patient examination lists, and patient treatment lists.

[0128] Understandably, due to the often complex table structure of medical lists, electronic medical lists recognized using OCR technology may exhibit table tilting, bending, or distortion. Therefore, before performing row and column segmentation on the medical list to be corrected, the following preprocessing operations can be performed to denoise or enhance the list, thereby improving the clarity of the table lines and correcting tilted or distorted areas to reduce the occurrence of misaligned rows. These operations include:

[0129] Image processing techniques are used to perform image enhancement and noise reduction on the medical list to be corrected.

[0130] The distorted regions in the medical list to be corrected are identified using a table detection algorithm.

[0131] The distorted region is corrected using a table correction algorithm.

[0132] In this embodiment of the invention, the image processing techniques include, but are not limited to, image enhancement techniques, image binarization techniques, and image distortion removal techniques.

[0133] In this embodiment of the invention, table detection algorithms and table correction algorithms based on machine learning and vision technologies can be used to correct the distorted area, making the table lines in the distorted area horizontal or vertical, thereby reducing the occurrence of misaligned lines.

[0134] Specifically, the process of performing row and column splitting on the medical list to be corrected to obtain a first medical list includes:

[0135] Edge detection is performed on the medical list to be corrected to obtain a grid structure corresponding to the medical list to be corrected.

[0136] Feature extraction is performed on each grid in the grid structure to obtain grid features.

[0137] Using machine learning algorithms, the row and column position information of the corresponding grid is identified based on the grid features;

[0138] Based on the row and column position information of each grid, each grid is split into cells to obtain the first medical list.

[0139] In this embodiment of the invention, an edge detection algorithm is used to scan the image of the medical list to be corrected, identify the start and end positions of each row or column, and obtain the grid structure corresponding to the medical list to be corrected.

[0140] In this embodiment of the invention, the grid features include, but are not limited to, features such as grid size and position.

[0141] In this embodiment of the invention, machine learning algorithms such as support vector machines, decision trees, or neural networks can be used to identify the row and column position information of the corresponding grid based on the grid features.

[0142] Understandably, each grid may contain merged cells, such as merging several rows or columns. Therefore, each grid can be split based on its row and column position information, adjusting a grid into one or more cells.

[0143] In this embodiment of the invention, by dividing the medical list to be corrected into rows and columns, the object to be corrected can be decomposed from the medical list into multiple cells composed of row and column information, which is beneficial for correcting errors based on each cell, thereby improving the accuracy of error correction.

[0144] Step 2: Construct the form structure features of the first medical list and the cell structure features of each cell in the first medical list;

[0145] In this embodiment of the invention, the form structure features refer to the overall distribution features of the first medical list, including but not limited to header distribution area features, form content distribution area features, built-in form features, number of rows and columns of the table, etc. The header distribution area features also include features such as row headers, list headers, and cascading headers.

[0146] In this embodiment of the invention, the cell structure features refer to the position of each cell and the positional and textual data relationship features between adjacent cells, such as the number of rows and columns of the cell and the header features corresponding to the cell.

[0147] Specifically, the form structure features for constructing the first medical list include:

[0148] Based on the preset table dimensions, obtain the value of each preset table dimension in the first medical list;

[0149] The values ​​of each of the preset table dimensions are normalized and encoded to obtain the table dimension codes;

[0150] The form structure features of the first medical list are obtained by matrix-concatenating the dimensional codes of each table.

[0151] In this embodiment of the invention, the preset table dimensions include, but are not limited to, the overall size of the table area, the size of the header area, the size of the text content area, and dimensions such as the data type, data format, or data relationship in the table.

[0152] In this embodiment of the invention, normalized encoding of the values ​​of each preset table dimension can be achieved by performing hotspot encoding with the same number of bits on the values ​​of each preset table dimension.

[0153] In this embodiment of the invention, the cell structure features of each cell in the first medical list can be constructed using preset cell dimensions. The preset cell dimensions can be set according to the content features of the actual medical list. For example, the preset cell dimensions include, but are not limited to, cell position, the header position corresponding to the cell, and the data type in the cell.

[0154] It should be noted that the method for constructing the cell structure features of each cell in the first medical list is the same as the method for constructing the form structure features of the first medical list, and will not be repeated here.

[0155] The purpose of constructing the form structure features of the first medical list and the cell structure features of each cell in the first medical list in this embodiment of the invention is to use the form structure features and cell structure features as error correction dimensions to correct the medical list to be corrected from the perspective of structural features.

[0156] Step 3: Calculate the structural feature difference between the cell structural feature of each cell and the form structural feature, identify the cells to be corrected based on the structural feature difference, and perform structural correction on the cells to be corrected to obtain the second medical list;

[0157] Under normal circumstances, the cell structure features of each cell in the first medical list basically conform to the form structure features of the first medical list. For example, the header features of each cell are consistent with the header features of the corresponding form structure features, and the positional features of each cell, consisting of the number of rows and columns, conform to the form distribution features in the corresponding form structure features. Therefore, when the structural feature difference between the cell structure features of a certain cell and the corresponding form structure features is relatively large, it indicates that the cell may contain errors and needs to be corrected.

[0158] Specifically, calculating the structural feature difference between the cell structural features of each cell and the form structural features includes:

[0159] The cell structure features of each cell are transformed into a vector to obtain a cell vector;

[0160] The form structure features are then transformed into a vector to obtain a form vector;

[0161] Calculate the distance between the cell vector of each cell and the form vector, and use the calculated distance as the structural feature difference.

[0162] In this embodiment of the invention, the distance between the cell vector and the form vector can be calculated using the Mahalanobis distance algorithm or the Euclidean distance algorithm.

[0163] Furthermore, when the structural feature difference is greater than a preset structural feature difference threshold, the corresponding cell is designated as a cell to be corrected, and a table correction algorithm can be used to correct the cell to be corrected.

[0164] In this embodiment of the invention, the preset structural feature difference threshold can be set according to the actual business situation.

[0165] This invention utilizes the structural feature difference between cell structure features and form structure features to identify cells that need correction. By correcting these cells, the accuracy of the second medical list is further improved compared to the first medical list.

[0166] Step 4: Construct the contextual information of each cell in the second medical list, and generate the predicted text of the corresponding cell based on the contextual information;

[0167] In this embodiment of the invention, an artificial intelligence-based language model, such as a neural network model, can be used to construct the context of each cell and generate the predicted text for each cell.

[0168] Specifically, the step of constructing the contextual information of each cell in the second medical list and generating the predicted text for the corresponding cell based on the contextual information includes:

[0169] Get the text content and position information of each cell;

[0170] Using a pre-trained language model, the context of each cell is generated based on the text content and position information of each cell.

[0171] Select one cell from the second medical list as the target cell;

[0172] Based on the context of the target cell, the predicted text of the target cell is generated using the pre-trained language model.

[0173] In this embodiment of the invention, the pre-trained language model can be either a deep learning-based convolutional neural network model or a ChatGPT model.

[0174] In this embodiment of the invention, predicted text for a cell is generated based on the context of the cell, and the accuracy of the text content of the corresponding cell can be judged using the predicted text.

[0175] Step 5: Calculate the text difference between the predicted text and the actual text of each cell in the second medical list, identify the cells to be corrected based on the text difference, and perform text correction on the cells to be corrected to obtain the corrected medical list.

[0176] In this embodiment of the invention, the actual text of each cell in the second medical list can be obtained. The text difference is obtained by comparing the predicted text with the actual text and quantifying the result of the comparison. The text difference is then used to identify cells requiring correction.

[0177] In another optional embodiment of the present invention, the text difference can be obtained by performing vector transformations on the predicted text and the real text respectively, and calculating the vector distance between the two.

[0178] In another optional embodiment of the present invention, the semantics of the predicted text and the semantics of the real text can be analyzed using the pre-trained language model, and the text difference can be obtained by comparing the semantic similarity between the two.

[0179] In this embodiment of the invention, when the text difference is greater than a preset text difference threshold, the corresponding cell is taken as the cell to be corrected, and the predicted text can be used to replace the real text of the corresponding cell, thereby realizing the text correction of the cell to be corrected.

[0180] In this embodiment of the invention, the preset text difference threshold can be set according to the actual business situation.

[0181] This invention applies to construct the table structure features and cell structure features of a medical list to be corrected. By utilizing the structural feature difference between the table structure features and cell structure features, cells to be corrected are identified, and structural correction is performed on these cells. Furthermore, by constructing the context of the cells, predicted text for each cell is generated, and then the text difference between the predicted text and the actual text is used to perform text correction on the cells to be corrected. Through the above method, structural and text correction can be performed on the medical list to be corrected from both structural and contextual perspectives, thereby improving the accuracy of medical list correction.

[0182] like Figure 6 The diagram shown is a structural schematic of an electronic device for implementing a medical list error correction method according to an embodiment of the present invention.

[0183] The electronic device 1 may include a processor 10, a memory 11 and a bus, and may also include a computer program stored in the memory 11 and executable on the processor 10, such as a medical list error correction program.

[0184] The memory 11 includes at least one type of readable storage medium, such as flash memory, portable hard drive, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 11 can be an internal storage unit of the electronic device 1, such as a portable hard drive. In other embodiments, the memory 11 can be an external storage device of the electronic device 1, such as a plug-in portable hard drive, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the electronic device 1. Furthermore, the memory 11 can include both internal and external storage units of the electronic device 1. The memory 11 can be used not only to store application software and various types of data installed on the electronic device 1, such as medical list error correction code, but also to temporarily store data that has been output or will be output.

[0185] In some embodiments, the processor 10 may be composed of integrated circuits, such as a single packaged integrated circuit or multiple integrated circuits with the same or different functions, including combinations of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips. The processor 10 is the control unit of the electronic device, connecting various components of the entire electronic device through various interfaces and lines. It executes programs or modules stored in the memory 11 (e.g., medical list error correction) and calls data stored in the memory 11 to perform various functions of the electronic device 1 and process data.

[0186] The bus can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This bus can be divided into an address bus, a data bus, a control bus, etc. The bus is configured to enable communication between the memory 11 and at least one processor 10, etc.

[0187] Figure 6Only electronic devices with components are shown; it will be understood by those skilled in the art that... Figure 6 The structure shown does not constitute a limitation on the electronic device 1, and may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0188] For example, although not shown, the electronic device 1 may also include a power supply (such as a battery) to power various components. Preferably, the power supply can be logically connected to the at least one processor 10 through a power management device, thereby enabling functions such as charging management, discharging management, and power consumption management. The power supply may also include one or more DC or AC power supplies, recharging devices, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components. The electronic device 1 may also include various sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be described in detail here.

[0189] Furthermore, the electronic device 1 may also include a network interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a Wi-Fi interface, a Bluetooth interface, etc.), which is typically used to establish communication connections between the electronic device 1 and other electronic devices.

[0190] Optionally, the electronic device 1 may further include a user interface, which may be a display, an input unit (such as a keyboard), or a standard wired or wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen. The display may also be appropriately referred to as a screen or display unit, used to display information processed in the electronic device 1 and to display a visual user interface.

[0191] It should be understood that the embodiments described are for illustrative purposes only and are not limited to this structure in the scope of the patent application.

[0192] The medical list error correction stored in the memory 11 of the electronic device 1 is a combination of multiple instructions, which, when executed in the processor 10, can achieve the following:

[0193] Perform a row and column splitting operation on the medical list to obtain the first medical list;

[0194] Construct the form structure features of the first medical list and the cell structure features of each cell in the first medical list;

[0195] Calculate the structural feature difference between the cell structural features of each cell and the form structural features;

[0196] The cells to be corrected are identified based on the structural feature differences, and structural correction is performed on the cells to be corrected to obtain the second medical list.

[0197] Construct the contextual information for each cell in the second medical list, and generate the predicted text for the corresponding cell based on the contextual information;

[0198] Calculate the text difference between the predicted text and the actual text in each cell of the second medical list;

[0199] The cells to be corrected are identified based on the text difference, and the text in the cells to be corrected is corrected to obtain the corrected medical list.

[0200] Furthermore, if the modules / units integrated in the electronic device 1 are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. The computer-readable storage medium can be volatile or non-volatile. For example, the computer-readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, or a read-only memory (ROM).

[0201] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor of an electronic device, can perform the following:

[0202] Perform a row and column splitting operation on the medical list to obtain the first medical list;

[0203] Construct the form structure features of the first medical list and the cell structure features of each cell in the first medical list;

[0204] Calculate the structural feature difference between the cell structural features of each cell and the form structural features;

[0205] The cells to be corrected are identified based on the structural feature differences, and structural correction is performed on the cells to be corrected to obtain the second medical list.

[0206] Construct the contextual information for each cell in the second medical list, and generate the predicted text for the corresponding cell based on the contextual information;

[0207] Calculate the text difference between the predicted text and the actual text in each cell of the second medical list;

[0208] The cells to be corrected are identified based on the text difference, and the text in the cells to be corrected is corrected to obtain the corrected medical list.

[0209] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0210] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0211] Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be embraced within the invention. No appended diagram markings in the claims should be construed as limiting the scope of the claims.

[0212] The blockchain referred to in this invention is a novel application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanisms, and encryption algorithms. Essentially, a blockchain is a decentralized database, a chain of data blocks linked together using cryptographic methods. Each data block contains information about a batch of network transactions, used to verify the validity of the information (anti-counterfeiting) and generate the next block. A blockchain can include an underlying blockchain platform, a platform product service layer, and an application service layer.

[0213] The embodiments of this application can acquire and process relevant data based on holographic projection technology. Artificial Intelligence (AI) is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0214] Furthermore, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices recited in a system claim may also be implemented by a single unit or device through software or hardware. The term "second class" is used to indicate names and does not indicate any specific order.

[0215] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for correcting errors in a medical bill, characterized in that, The method includes: Perform a row and column splitting operation on the medical list to obtain the first medical list; The form structure features of the first medical list and the cell structure features of each cell in the first medical list are constructed. The form structure features refer to the overall distribution features of the first medical list, including header distribution area features, form content distribution area features, built-in form features, and the number of rows and columns of the table. The header distribution area features include row headers, column headers, and cascading headers. The cell structure features refer to the position of each cell in the first medical list and the positional relationship and text data relationship features between adjacent cells, including the number of rows and columns of each cell and the header features corresponding to that cell. Calculate the structural feature difference between the cell structural features of each cell and the form structural features; The cells to be corrected are identified based on the structural feature differences, and structural correction is performed on the cells to be corrected to obtain the second medical list. Construct the contextual information for each cell in the second medical list, and generate the predicted text for the corresponding cell based on the contextual information; Calculate the text difference between the predicted text and the actual text in each cell of the second medical list; The cells to be corrected are identified based on the text difference, and the text in the cells to be corrected is corrected to obtain the corrected medical list.

2. The medical billing error correction method as described in claim 1, characterized in that, The medical list to be corrected is subjected to row and column splitting operation to obtain a first medical list, including: Edge detection is performed on the medical list to be corrected to obtain a grid structure corresponding to the medical list to be corrected. Feature extraction is performed on each grid in the grid structure to obtain grid features; Using machine learning algorithms, the row and column position information of the corresponding grid is identified based on the grid features; Based on the row and column position information of each grid, each grid is split into cells to obtain the first medical list.

3. The medical billing error correction method as described in claim 1, characterized in that, The form structure features for constructing the first medical list include: Based on the preset table dimensions, obtain the value of each preset table dimension in the first medical list; The values ​​of each of the preset table dimensions are normalized and encoded to obtain the table dimension codes; The form structure features of the first medical list are obtained by matrix-concatenating the dimensional codes of each table.

4. The medical billing error correction method as described in claim 1, characterized in that, The calculation of the structural feature difference between the cell structural features of each cell and the form structural features includes: The cell structure features of each cell are transformed into a vector to obtain a cell vector; The form structure features are then transformed into a vector to obtain a form vector; Calculate the distance between the cell vector of each cell and the form vector, and use the calculated distance as the structural feature difference.

5. The medical billing error correction method as described in claim 1, characterized in that, The step of constructing the contextual information for each cell in the second medical list and generating the predicted text for the corresponding cell based on the contextual information includes: Get the text content and position information of each cell; Using a pre-trained language model, the context of each cell is generated based on the text content and position information of each cell. Select one cell from the second medical list as the target cell; Based on the context of the target cell, the predicted text of the target cell is generated using the pre-trained language model.

6. The medical billing error correction method as described in claim 1, characterized in that, Before performing row and column splitting operations on the medical list to be corrected, the method further includes: Image processing techniques are used to perform image enhancement and noise reduction on the medical list to be corrected. The distorted regions in the medical list to be corrected are identified using a table detection algorithm. The distorted region is corrected using a table correction algorithm.

7. A medical billing error correction device, characterized in that, The device includes: The row and column splitting module is used to perform row and column splitting operations on the medical list to be corrected, so as to obtain the first medical list; The structural feature extraction module is used to construct the form structure features of the first medical list and the cell structure features of each cell in the first medical list. The form structure features refer to the overall distribution features of the first medical list, including header distribution area features, form content distribution area features, built-in form features, and the number of rows and columns of the table. The header distribution area features include row headers, column headers, and cascading headers. The cell structure features refer to the position of each cell in the first medical list and the positional relationship and text data relationship features between adjacent cells, including the number of rows and columns of each cell and the header features corresponding to that cell. The structural error correction module is used to calculate the structural feature difference between the cell structural features of each cell and the form structural features, identify the cells to be corrected based on the structural feature difference, and perform structural error correction on the cells to be corrected to obtain the second medical list. The context generation module is used to construct the context of each cell in the second medical list and generate the predicted text of the corresponding cell based on the context. The context correction module is used to calculate the text difference between the predicted text and the actual text of each cell in the second medical list, identify the cells to be corrected based on the text difference, and perform text correction on the cells to be corrected to obtain the corrected medical list.

8. The medical billing error correction device as described in claim 7, characterized in that, The row and column splitting module performs row and column splitting operations from the medical list to be corrected using the following method: Edge detection is performed on the medical list to be corrected to obtain a grid structure corresponding to the medical list to be corrected. Feature extraction is performed on each grid in the grid structure to obtain grid features; Using machine learning algorithms, the row and column position information of the corresponding grid is identified based on the grid features; Based on the row and column position information of each grid, each grid is split into cells to obtain the first medical list.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor to enable the at least one processor to perform the medical inventory correction method as described in any one of claims 1 to 6.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the medical inventory error correction method as described in any one of claims 1 to 6.

Images