Method, device and equipment for generating orthopedic electronic health records based on orthopedic clinical medical record documents

Abstract

The application discloses a method, device and equipment for generating an orthopedic electronic health record based on an orthopedic clinical medical record document, and relates to the field of medical data arrangement. The application respectively identifies picture format documents and documents containing both text information and table information in the orthopedic clinical medical record document. In view of the shortcomings of an OCR algorithm in the identification process of table information, the application first enhances the table area in the document, and then identifies by using the OCR algorithm, thereby improving the accuracy of the identification of table data in the orthopedic clinical medical record document. In addition, the application performs de-identification processing on the first identification information, the second identification information and medical image data before constructing the orthopedic electronic health record, thereby ensuring the privacy of patient data.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of medical data processing, in particular to a method, device and equipment for generating orthopedic electronic health records based on orthopedic clinical medical record documents. BACKGROUND

[0002] A large amount of electronic health records (EHR) data has been collected from medical institutions, including medical images, measurement data, doctor's notes, treatment records, demographic data, etc. These data contain valuable information for assessing the clinical condition of patients and their response to treatment, providing a promising opportunity for clinical research. Usually, these data are stored in document format without structured processing for interpretation and analysis, and do not support research purposes. A benchmark dataset for orthopedic medical treatment is constructed from unstructured archives. Two major challenges are faced: (1) the conversion from documents to EHR datasets usually requires high-quality optical character recognition (OCR) technology to extract important information; (2) it is challenging to share patient data using EHR systems in a medical environment. Although OCR technology is widely used to extract information from hospital archives in the form of unstructured digital documents (such as portable document format and images), it is difficult to parse special format data (such as table data) into structured machine-readable format due to the complexity of its structure and style. Figure 1 An example of recognition results using the advanced OCR tool PP-Structure of Baidu is shown. Obviously, this tool fails to accurately parse the hospital document, with multiple errors in words and formats. SUMMARY

[0003] The purpose of the present application is to provide a method, device and equipment for generating orthopedic electronic health records based on orthopedic clinical medical record documents, which can improve the accuracy of table data recognition in orthopedic clinical medical record documents and ensure the privacy and security of patient data.

[0004] To achieve the above purpose, the present application provides the following solutions:

[0005] In a first aspect, the present application provides a method for generating orthopedic electronic health records based on orthopedic clinical medical record documents, comprising:

[0006] obtaining an orthopedic clinical medical record document; the orthopedic clinical medical record document comprises a first medical record document picture and a second medical record document picture, both of which are medical record documents in picture format, the first medical record document picture contains only text information, and the second medical record document picture contains both text information and table information;

[0007] extract the text information in the first medical record document picture using the PaddleOCR algorithm as first recognition information;

[0008] divide the text area and the table area in the second medical record document picture using the OpenCV algorithm;

[0009] enhance the table area using a line segment detection algorithm to obtain an enhanced table area;

[0010] extract the text information in the text area using the PaddleOCR algorithm, and extract the text information in the enhanced table area using the PP-Structure algorithm to obtain second recognition information;

[0011] obtain medical image data corresponding to the orthopedic clinical medical record document;

[0012] perform de-identification processing on the first recognition information, the second recognition information, and the medical image data;

[0013] generate an orthopedic electronic health record based on the first recognition information, the second recognition information, and the medical image data after de-identification processing.

[0014] In a second aspect, the application provides a device for generating an orthopedic electronic health record based on an orthopedic clinical medical record document, which applies the method for generating an orthopedic electronic health record based on an orthopedic clinical medical record document described above. The device for generating an orthopedic electronic health record based on an orthopedic clinical medical record document includes:

[0015] a medical record document acquisition module for acquiring an orthopedic clinical medical record document; the orthopedic clinical medical record document includes a first medical record document picture and a second medical record document picture, both of which are medical record documents in picture format, the first medical record document picture contains only text information, and the second medical record document picture contains both text information and table information;

[0016] a first text information extraction module for extracting the text information in the first medical record document picture using the PaddleOCR algorithm as first recognition information;

[0017] a region division module for dividing the text area and the table area in the second medical record document picture using the OpenCV algorithm;

[0018] a table enhancement module for enhancing the table area using a line segment detection algorithm to obtain an enhanced table area;

[0019] The second character information extraction module is configured to extract character information in a character region using a PaddleOCR algorithm and extract character information in an enhanced table region using a PP-Structure algorithm, to obtain second recognition information.

[0020] The medical image data acquisition module is configured to acquire medical image data corresponding to the orthopedic clinical medical record document.

[0021] The de-identification module is configured to perform de-identification processing on the first recognition information, the second recognition information, and the medical image data.

[0022] The orthopedic electronic health record generation module is configured to generate an orthopedic electronic health record based on the de-identified first recognition information, the de-identified second recognition information, and the medical image data.

[0023] In a third aspect, the present application provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor executes the computer program to implement the method for generating an orthopedic electronic health record based on an orthopedic clinical medical record document.

[0024] According to the embodiments provided in the present application, the following technical effects are disclosed.

[0025] The present application provides a method, device, and equipment for generating an orthopedic electronic health record based on an orthopedic clinical medical record document. The present application identifies first medical record document pictures containing only character information and second medical record document pictures containing both character information and table information in the orthopedic clinical medical record document. In view of the shortcomings of the OCR algorithm in the identification process of table information, the present application first enhances the table region in the document, and then uses the OCR algorithm for identification, thereby improving the accuracy of the identification of table data in the orthopedic clinical medical record document. In addition, the present application performs de-identification processing on the first recognition information, the second recognition information, and the medical image data before constructing the orthopedic electronic health record, thereby ensuring the privacy of the patient data. BRIEF DESCRIPTION OF DRAWINGS

[0026] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0027] Figure 1 An example diagram of the identification result of the table data by the OCR algorithm in the background art of the present application;

[0028] Figure 2A flowchart of a method for generating orthopedic electronic health records based on orthopedic clinical medical records provided in one embodiment of the present application;

[0029] Figure 3 A schematic diagram of a method for generating orthopedic electronic health records based on orthopedic clinical medical records provided in one embodiment of the present application;

[0030] Figure 4 A schematic diagram of the recognition results of the OCR algorithm before enhancement provided in one embodiment of the present application;

[0031] Figure 5 A schematic diagram of the recognition results of an enhanced OCR algorithm provided in one embodiment of the present application;

[0032] Figure 6 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0033] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0034] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0035] In an exemplary embodiment, Figure 2 and Figure 3 As shown, a method for generating an orthopedic electronic health record based on an orthopedic clinical medical record document is provided, comprising the following steps 201 to 208. In which:

[0036] Step 201, obtain an orthopedic clinical medical record document; the orthopedic clinical medical record document includes: a first medical record document image and a second medical record document image, the first medical record document image and the second medical record document image are both medical record documents in image format, the first medical record document image only contains text information, and the second medical record document image contains both text information and table information.

[0037] Step 202: Use the PaddleOCR algorithm to extract text information from the first medical record document image as first recognition information.

[0038] Step 203: Use the OpenCV algorithm to divide the text area and table area in the second medical record document image.

[0039] Step 204: Use a line segment detection algorithm to enhance the table area to obtain an enhanced table area.

[0040] Step 205 : Use the PaddleOCR algorithm to extract text information in the text area, and use the PP-Structure algorithm to extract text information in the enhanced table area to obtain second recognition information.

[0041] Step 206: Acquire medical imaging data corresponding to the orthopedic clinical medical record document.

[0042] Step 207: De-identify the first identification information, the second identification information, and the medical image data.

[0043] Step 208 : Generate an orthopedic electronic health record based on the de-identified first identification information, the second identification information, and the medical imaging data.

[0044] In steps 201-208 of this application, patient examinations, diagnoses, treatment records, medical images, and clinical notes are first extracted from the documents and personal identifying information is removed. Automatic OCR-assisted tools are then used for text recognition and data structuring. Researchers can use this tool to generate self-collected EHR datasets for use in medical practice.

[0045] In this embodiment, medical imaging data refers to CT images taken by orthopedics, and the data only needs to be de-identified. Medical medical records refer to clinical medical records obtained from the hospital. The records are all saved in the form of pictures. The table information or text information in the pictures needs to be distinguished and then identified separately.

[0046] In another exemplary embodiment, in the above step 201 , orthopedic clinical medical records are collected from the hospital.

[0047] In another exemplary embodiment, in steps 202-205, an automatic OCR-assisted tool is used to identify and extract information from the document. Five types of information are extracted: patient examination records, diagnosis records, treatment records, clinical notes, and medical images, specifically including:

[0048] First, the regions of interest in the document are recognized using OpenCV, and a preliminary recognition is performed using PP-Structure to extract five types of information: patient examination, diagnosis, treatment, clinical notes, and medical images, and each type of information is processed separately. The specific implementation process is as follows: for a document in image format, the text information in the document is extracted using PaddleOCR to obtain first recognition information. For a document containing both text and table information, OpenCV is used to segment it into different regions containing various types of data, and then PaddleOCR is used for text recognition. For the table information, a Line Segment Detector (LSD) algorithm is applied to enhance the recognition effect. After enhancement, PP-Structure is used to extract the table information to obtain second recognition information.

[0049] When using the automatic OCR assistant tool, the embodiments of the present application observed that using different colors for line enhancement can result in different OCR recognition performance. Therefore, this experiment was conducted to investigate the effect of different color line enhancement on OCR recognition. The recognition performance of 10 significantly different colors was tested on one document to evaluate their effectiveness in capturing document information.

[0050] The formula for calculating the cell-level recognition accuracy of the document is:

[0051]

[0052] where Cell-level Accuracy is the cell-level recognition accuracy, δ is the number of cells in the recognition result that are completely matched with the true values of each cell in the third type of document, and ε is the total number of cells in the third type of document. To ensure the accuracy of the true values used for experimental comparison, these true values are manually recognized and collected from the original document through a manual process.

[0053] As can be seen in the test results of 10 colors (see Table 1), the pink color with RGB value (255, 125, 255) has the best enhancement effect, while the purple color with RGB value (255, 0, 255) has the worst enhancement effect.

[0054] Table 1 Effect of different colors on OCR recognition

[0055] Color RGB Cell-level accuracy Pink (255，125，255) 0.96 Yellow (255，255，0) 0.90 Orange-red (255，165，0) 0.90 Blue-green (0，255，255) 0.89 Gray (128，128，128) 0.81 Green (0，255，0) 0.78 Blue (0，0，255) 0.76 Red (255，0，0) 0.75 Black (0，0，0) 0.73 Purple (255，0，255) 0.62

[0056] According to the experimental results on the effect of color on OCR recognition performance, the embodiments of the present application select the best pink color for the automatic OCR assistant tool.

[0057] To demonstrate the effectiveness of the above method, this example sampled a total of 30 documents, 10 each from treatment records, patient examination records, and diagnostic records, for OCR recognition. The recognition results before and after enhancement were compared and analyzed, and a comprehensive evaluation was conducted. The evaluation metric used cell-level accuracy. The ground truth values ​​were obtained through manual sampling and comparison.

[0058] We observed improvements of 7%, 34%, and 18% in cell-level recognition accuracy for treatment records, patient examination records, and diagnosis records after augmentation, respectively (see Table 2). This demonstrates a significant improvement in OCR recognition performance when augmented using the LSD algorithm. This directly demonstrates the superiority of OCR-assisted tools over current unenhanced recognition tools.

[0059] Table 2 Cell-level recognition accuracy after enhancement of treatment records, patient examination records, and diagnosis records

[0060] Picture type Accuracy before enhancement Accuracy after enhancement Treatment record 0.96 0.89 Patient examination record 0.88 0.54 Diagnosis record 0.87 0.69

[0061] By using automatic OCR auxiliary tools, compare the effects before and after use, such as Figure 4 and Figure 5 As shown, Figure 4 (a) is the second medical record document image before enhancement. Figure 4 (b) is for Figure 4 The recognition result obtained by performing the recognition on (a) in the figure is: Figure 5 (a) is the enhanced second medical record document image. Figure 5 (b) in the Figure 5 The method of the present application can extract high-quality key information for constructing an electronic health record (EHR) dataset. This tool can significantly reduce the manual work required for data extraction.

[0062] Secondly, the format of the extracted information is standardized using regular expressions, and data is organized using Pandas. In order to distinguish the OCR recognition results, a keyword-based method is used to classify the extracted information by identifying whether the information contains predefined keywords related to each type of data.

[0063] For example, when the information first recognized in the PP-Structure contains keywords such as "item category" and "result", the information is classified as a patient examination.

[0064] In addition, considering the specific properties of the data: (1) the size of each image is fixed, and (2) the locations to be recognized are consistent, the embodiment of the present application extracts images from predefined areas for recognition. However, during the recognition process, it can be observed that the lines separating tables in the document are usually very blurred. To solve this problem, the LSD algorithm is applied to automatically detect straight lines and enhance the table by making the boundaries clearer. During the table recognition process, the original data without enhanced lines often misidentifies the format information of certain areas, while after enhancement, this information is correctly recognized. This result shows that OCR auxiliary tools can effectively enhance the recognition of table information.

[0065] In an exemplary embodiment, to ensure the accuracy of data used to generate orthopedic electronic health records, all data extracted from the automatic OCR-assisted tool is manually compared and cleaned to ensure that the extracted electronic health record (EHR) is consistent with the original data, specifically including:

[0066] Medical imaging data was collected from hospital patient records to construct the Orthopedic Electronic Health Record (OEHR). Medical image annotation in the OEHR was performed by orthopedic specialists. To further improve data quality, this application manually compared and cleaned all data extracted from the automated OCR-assisted tool to ensure that the extracted electronic health record (EHR) was consistent with the original data. This meticulous approach helped reduce potential inaccuracies that could be introduced by automated OCR-assisted tools and ensured the high quality and reliability of the OEHR.

[0067] In an exemplary embodiment, step 207 primarily de-identifies sensitive patient data, dates, and patient IDs. This embodiment of the present application strictly complies with the Health Insurance Portability and Accountability Act. A series of pre-processing steps are performed to protect data privacy.

[0068] First, we extracted essential key information from electronic health records (EHRs) and filtered out sensitive data containing personally identifiable information (such as name, address, and contact information). We also de-identified the patient IDs and admission IDs obtained from the documents so that they were no longer associated with specific individuals in the dataset by removing the associated information.

[0069] Then, for the date information, we used a time randomization algorithm to de-identify it.

[0070] Specifically, we use the Random Time Drift algorithm (see Table 3) to encrypt date information in electronic health records (EHRs). This algorithm receives a date θ and outputs an encrypted date γ. The specific process is as follows: First, we convert the input date θ into a string. Next, we apply the one-way encryption algorithm SHA256 to the string θ. Specifically, we use the SHA256 hash function to generate a hash value string α. We then randomly sample α. The sampling process is as follows: 3 random numbers are taken from α and added together to convert it to the year; 1 random number is taken from α and converted to the month; 2 random numbers are taken from α and added together to convert it to the day; The year, month, and day are then processed separately using the modulo addition operation to strictly adhere to the real-world year, month, and day mappings, ensuring the authenticity of the date information. Furthermore, we only encrypt the starting date information in the module containing date information. This process ensures that the date displayed in the EHR is consistent with the virtual time point while maintaining the interval between related dates.

[0071] Table 3 Virtual date generation process

[0072]

[0073] Then, in order to manage patient identification in the study, the embodiment of the present application uses a randomization method to assign a virtual ID to each patient and establish a mapping relationship between the actual patient ID and the virtual ID. This method ensures that the patient identity is de-identified by using the virtual ID when the data set is released.

[0074] Based on the same inventive concept, the embodiment of the present application also provides a device for generating orthopedic electronic health records based on orthopedic clinical medical records documents, which is used to implement the above-mentioned method for generating orthopedic electronic health records based on orthopedic clinical medical records documents. The implementation solution provided by the device is similar to the implementation solution described in the above-mentioned method. Therefore, the specific limitations of the embodiments of one or more devices for generating orthopedic electronic health records based on orthopedic clinical medical records documents provided below can be found in the above-mentioned limitations of the method for generating orthopedic electronic health records based on orthopedic clinical medical records documents, and will not be repeated here.

[0075] In an exemplary embodiment, a device for generating an orthopedic electronic health record based on an orthopedic clinical medical record document is provided, comprising:

[0076] A medical record document acquisition module is used to acquire orthopedic clinical medical record documents; the orthopedic clinical medical record documents include: a first medical record document image and a second medical record document image, the first medical record document image and the second medical record document image are both medical record documents in image format, the first medical record document image only contains text information, and the second medical record document image contains both text information and table information.

[0077] The first text information extraction module is used to extract text information from the first medical record document image using the PaddleOCR algorithm as the first recognition information.

[0078] The region segmentation module is used to segment the text region and the table region in the second medical record document image using the OpenCV algorithm.

[0079] The table enhancement module is used to enhance the table area using a line segment detection algorithm to obtain an enhanced table area.

[0080] The second text information extraction module is used to extract text information in the text area using the PaddleOCR algorithm, and to extract text information in the enhanced table area using the PP-Structure algorithm to obtain second recognition information.

[0081] The medical image data acquisition module is used to acquire the medical image data corresponding to the orthopedic clinical medical record document.

[0082] The de-identification module is used to perform de-identification processing on the first identification information, the second identification information and the medical image data.

[0083] The orthopedic electronic health record generation module is used to generate an orthopedic electronic health record based on the de-identified first identification information, the second identification information and the medical imaging data.

[0084] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 6As shown in the figure. The computer device includes a processor, a memory, an input / output interface (I / O for short) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through the system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capability. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the data determined by complementarity. The input / output interface of the computer device is used to exchange information between the processor and the external device. The communication interface of the computer device is used to communicate with the terminal outside through the network connection. The computer program is executed by the processor to realize a method for generating an orthopedic electronic health record based on an orthopedic clinical medical record document.

[0085] Those skilled in the art can understand that, Figure 6 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different component arrangement.

[0086] In one exemplary embodiment, a computer device is also provided, including a memory and a processor, the memory storing a computer program, and the processor executing the computer program to realize the steps in the above-mentioned method embodiments for generating an orthopedic electronic health record based on an orthopedic clinical medical record document.

[0087] The database involved in each of the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a blockchain, etc., without being limited thereto. The processor involved in each of the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, etc., without being limited thereto.

[0088] Each of the technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above embodiments are described, however, as long as the combination of the technical features does not exist contradictory, it should be considered as the scope of the present disclosure.

[0089] This document uses specific examples to illustrate the principles and implementation methods of this application. The description of the above examples is only intended to help understand the method and core concept of this application. At the same time, for those skilled in the art, based on the concept of this application, there may be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for generating orthopedic electronic health records based on orthopedic clinical medical records, characterized in that: include: Obtain orthopedic clinical medical records; The orthopedic clinical medical record document includes: a first medical record document image and a second medical record document image, wherein the first medical record document image and the second medical record document image are both medical record documents in image format, the first medical record document image only contains text information, and the second medical record document image contains both text information and table information; Use the PaddleOCR algorithm to extract text information from the first medical record document image as the first recognition information; Use OpenCV algorithm to divide the text area and table area in the second medical record document image; Use the line segment detection algorithm to enhance the table area to obtain the enhanced table area; Use the PaddleOCR algorithm to extract text information in the text area, and use the PP-Structure algorithm to extract text information in the enhanced table area to obtain second recognition information; Acquiring medical imaging data corresponding to the orthopedic clinical medical record document; performing de-identification processing on the first identification information, the second identification information, and the medical image data; generating an orthopedic electronic health record based on the de-identified first identification information, the second identification information, and the medical imaging data; Use the line segment detection algorithm to enhance the table area to obtain the enhanced table area, including: Determine the color with the highest unit-level recognition accuracy in the color set as the target enhancement color; Detecting a table border in the table area using a line segment detection algorithm, and setting the color of the table border to the target enhancement color to obtain an enhanced table area; Determine the color with the highest unit-level recognition accuracy in the color set as the target enhancement color, specifically including: Select any document containing table information as the third type of document; respectively setting the color of the table borders in the third type of documents to each color in the color set to obtain a plurality of enhanced documents; Recognize each enhanced document based on the PaddleOCR algorithm and obtain the recognition results of each enhanced document; According to the recognition results of each enhanced document, the unit-level recognition accuracy of each enhanced document is calculated; The color corresponding to the enhanced document with the highest unit-level recognition accuracy is determined as the target enhanced color.

2. The method for generating orthopedic electronic health records based on orthopedic clinical medical records according to claim 1, characterized in that: The calculation formula for unit-level recognition accuracy is: Among them, Cell-levelAccuracy is the cell-level recognition accuracy, δ is the number of cells in the recognition result that completely match the true value of each cell in the third type document, and ε is the total number of cells in the third type document.

3. The method for generating orthopedic electronic health records based on orthopedic clinical medical records according to claim 1, characterized in that: De-identifying the first identification information, the second identification information, and the medical image data, which also includes: Standardizing the formats of the first identification information and the second identification information using a regular expression to obtain standardized first identification information and standardized second identification information; Use the Pandas tool to perform data sorting on the standardized first identification information and the standardized second identification information to obtain the sorted first identification information and the sorted second identification information; The keyword-based classification method is used to determine the types of the sorted first identification information and the sorted second identification information.

4. The method for generating orthopedic electronic health records based on orthopedic clinical medical records according to claim 1, characterized in that: De-identifying the first identification information, the second identification information, and the medical imaging data, specifically including: Filtering out sensitive data in the first identification information, the second identification information, and the medical imaging data; the sensitive data includes name, address, and contact information; Using a time randomization algorithm to replace actual dates in the first identification information, the second identification information, and the medical imaging data with virtual dates generated based on the time randomization algorithm; The actual IDs in the first identification information, the second identification information and the medical image data are replaced with virtual IDs generated by a randomization method.

5. The method for generating orthopedic electronic health records based on orthopedic clinical medical records according to claim 4, characterized in that: The time randomization algorithm specifically includes: Convert said actual date to a string; Encrypt the string using the SHA256 hash function to obtain a hash value string; Randomly select three numbers from the hash value string and add them together, using the result of the addition as the first number D1; Randomly select a number from the hash value string as the second number D2; Randomly select two numbers from the hash value string and add them together, using the result of the addition as the third number D3; Calculate the year of the virtual date using the following formula based on the first number: year=3000+D1%1000+1; Calculate the month of the virtual date using the following formula according to the second number; month=D2%12+1; Based on the third digit and the year and month of the virtual date, use the following formula to calculate the day of the virtual date; 6. The method for generating orthopedic electronic health records based on orthopedic clinical medical records according to claim 5, characterized in that: The leap year condition is: year%4=0 and year%100!=0, or year%400=0.

7. A device for generating orthopedic electronic health records based on orthopedic clinical medical records, characterized in that: The device for generating orthopedic electronic health records based on orthopedic clinical medical records documents applies the method for generating orthopedic electronic health records based on orthopedic clinical medical records documents according to any one of claims 1 to 6, and the device for generating orthopedic electronic health records based on orthopedic clinical medical records documents comprises: A medical record document acquisition module is used to acquire orthopedic clinical medical record documents; the orthopedic clinical medical record documents include: a first medical record document image and a second medical record document image, the first medical record document image and the second medical record document image are both medical record documents in image format, the first medical record document image contains only text information, and the second medical record document image contains both text information and table information; A first text information extraction module is used to extract text information from the first medical record document image using a PaddleOCR algorithm as first recognition information; A region segmentation module, used to segment the text region and the table region in the second medical record document image using an OpenCV algorithm; A table enhancement module is used to enhance the table area using a line segment detection algorithm to obtain an enhanced table area; A second text information extraction module is used to extract text information in the text area using the PaddleOCR algorithm and to extract text information in the enhanced table area using the PP-Structure algorithm to obtain second recognition information; A medical imaging data acquisition module, configured to acquire medical imaging data corresponding to the orthopedic clinical medical record document; a de-identification module, configured to perform de-identification processing on the first identification information, the second identification information, and the medical image data; The orthopedic electronic health record generation module is used to generate an orthopedic electronic health record based on the de-identified first identification information, the second identification information and the medical imaging data.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method for generating an orthopedic electronic health record based on an orthopedic clinical medical record document as described in any one of claims 1 to 6.

Images