Method and device for identifying content of medical record image
By multi-layer processing of paper medical record images, structured medical record data is generated and encrypted to store, the problem of lengthy and professional difficulty in understanding paper medical records is solved, the electronicization and security of medical records is realized, and the generation of health suggestions is supported. and medical research.
Patent Information
- Application Number
- CN202411090893.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-09
- Publication Date
- 2024-11-29
AI Technical Summary
Paper medical records are still used in existing medical systems, which makes it difficult for users to save and understand the content of the medical records, and it is difficult to quickly count and study pathological data.
By pre-processing, optical character recognition, semantic recognition and AI model processing of paper medical record images, structured medical record data is generated, encrypted and stored, and health suggestions are generated, which solves the lengthy and professional difficult-to-understand medical record content. Problem, while improving data security.
It realizes efficient electronicization and structure of medical records, simplifies user medical record management, generates health suggestions, improves the availability and security of medical record data, and supports the progress of medical research.
Smart Images

Figure CN119049625A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of electronic medical records, and in particular to a method and device for identifying the content of a medical record image. Background Art
[0002] With the emergence and development of big model technology, artificial intelligence has played a role in replacing intensive human labor in various industries such as medical care and education. However, the existing medical system still uses paper medical records as the carrier of user medical information. In the scenarios of users' medical treatment, during and after diagnosis, there are still problems such as the trouble of saving paper medical records after users obtain medical records and test reports, the lengthy content of medical records is easy to forget, and the overly professional content of medical records makes it difficult for users to understand. At the same time, it is difficult to count the pathological data resources in the large number of medical records generated, and it is difficult for researchers to quickly extract research points from a large number of medical records. Summary of the invention
[0003] In view of this, the present invention proposes a method and device for identifying the content of a medical record image, and the specific scheme is as follows:
[0004] In a first aspect, the present application proposes a method for identifying the content of a medical record image, comprising:
[0005] Obtaining a paper medical record image of a user, performing image preprocessing on the medical record image, and obtaining a processed image;
[0006] Perform optical character recognition on the processed image to obtain initial text information;
[0007] Perform semantic recognition related processing on the initial text information to obtain text data;
[0008] The trained AI model is used to extract information and structure the text data in turn to obtain medical record data;
[0009] Encrypt the medical record data in a preset encryption method and store it in a database corresponding to the user's account and / or a database corresponding to the user's doctor's account;
[0010] For each user, the medical record data in the database is summarized in the order of the consultation time, and the medical record data within the current preset time period is analyzed through the preset medical data model to generate health advice; and the health advice is sent to the user.
[0011] In some specific embodiments, the process of performing image preprocessing on a paper medical record image includes:
[0012] Using a denoising algorithm to denoise the paper medical record image so that the image quality of the paper medical record image reaches a preset value; and / or
[0013] Adjust the contrast, brightness, and / or sharpness of the paper medical record image; and / or
[0014] Binarize the paper medical record image.
[0015] In some specific embodiments, the initial text information is processed with semantic recognition to obtain text data including:
[0016] Perform noise removal on the initial text information to remove information irrelevant to the medical record;
[0017] Based on semantic recognition, the text in the initial text information after noise removal is segmented to obtain text data.
[0018] In some specific embodiments, it also includes:
[0019] If a medical record modification instruction fed back by a preset first person is received, the medical record data corresponding to the medical record modification instruction and located in the database is determined;
[0020] Send the determined medical record data and the corresponding correction content in the medical record correction instruction to the preset second person;
[0021] If a confirmation instruction from a preset second person is received, the determined medical record data is corrected based on the correction content; and the comparison before and after the correction is added to the determined medical record data;
[0022] The medical record data in the database is updated based on the corrected medical record data.
[0023] In some specific embodiments, it also includes:
[0024] If the health advice meets the preset serious conditions, the health advice is sent to other users related to the user;
[0025] The process of sending health advice to users includes:
[0026] Health advice that meets preset serious conditions is sent to the user in multiple ways and multiple times.
[0027] In some specific embodiments, by configuring preset severity conditions, the current health advice can be evaluated for importance. In actual applications, the preset severity conditions can have multiple levels, wherein the higher the level of the preset severity conditions reached by the health advice, the more frequently and the more times the health advice is sent to the user, making it more likely that the user will view the health advice.
[0028] In some specific embodiments, it also includes:
[0029] Retrieving medical record data of users in a preset area and / or preset age group and / or preset disease from a database;
[0030] Determine the common medical record data and individual medical record data from the retrieved medical record data through the preset AI medical analysis model;
[0031] And the determined common medical record data and individual medical record data are displayed to the preset objects.
[0032] In actual applications, the preset objects can be doctors or researchers in the medical system. By using AI medical models to determine the common medical record data and individual medical record data from the retrieved medical records, doctors or researchers in the medical system can quickly judge the overall characteristics and development trends of the patient group, and assist in further research.
[0033] In some specific embodiments, it also includes:
[0034] If a medical record query request is received, searching the database for medical record data corresponding to the medical record query request;
[0035] Determine whether the requester of the medical record query request is the user or the doctor corresponding to the user;
[0036] If the judgment result is no, the found medical record data will be anonymized and sent to the requester.
[0037] In some specific embodiments, it also includes:
[0038] Extract pathological data from text data based on semantic recognition, and detect whether there is personal information data related to the user in the pathological data;
[0039] Desensitize the pathology data and hide or replace the user's personal information in the pathology data;
[0040] The desensitized pathological data are stored in a preset pathology database.
[0041] In practical applications, this step may be performed after the step of performing semantic recognition-related processing on the initial text information to obtain text data.
[0042] By desensitizing text data and anonymizing medical record data, pathology-related content in medical records can be shared with doctors or relevant researchers while ensuring information security. This can form a pathology database and anonymous medical records for medical research, which helps advance medical-related research while minimizing the risk of leakage of users' personal information.
[0043] In actual applications, users can also perform secondary encryption on medical record data, such as setting personal keys.
[0044] In some specific embodiments, medical record data, health advice and paper medical record images can also be associated and stored, and search keywords can be assigned according to preset rules to facilitate users to query medical record data, and / or health advice, and / or paper medical record images based on keywords.
[0045] By storing the original paper medical record images in association with the medical record data and health advice, users can quickly find and compare the original paper medical record images when they have questions about the content of the medical record data and / or health advice. When seeing a doctor, users can also provide the original paper medical record images to professional medical staff for further manual verification of the contents of the structured medical records.
[0046] In a second aspect, the present application proposes a device for identifying the content of a medical record image, which is used to implement any of the methods in the aforementioned technical solutions, including:
[0047] An image preprocessing module is used to obtain a paper medical record image of a user, perform image preprocessing on the medical record image, and obtain a processed image;
[0048] A text processing module is used to perform optical character recognition on the processed image to obtain initial text information;
[0049] A semantic recognition module is used to perform semantic recognition related processing on the initial text information to obtain text data;
[0050] The structuring module is used to extract and structure information from text data in turn through the trained AI model to obtain medical record data;
[0051] Database module, used to store medical record data;
[0052] The personalized service and recommendation module is used to summarize the medical record data in the database for each user in the order of the consultation time, and analyze the medical record data within the current preset time period through the preset medical data model to generate health recommendations; and send the health recommendations to the user.
[0053] In some specific embodiments, an encryption module is also included; the encryption module is used to encrypt the medical record data in a preset encryption method and store it in a database corresponding to the corresponding user's account and / or a database corresponding to the user's related doctor's account.
[0054] Beneficial effects: The present application proposes a method and device for identifying the content of a medical record image. After multiple and multi-layer processing of a paper medical record image, structured medical record data can be obtained. This medical record data can be quickly called from a database by the user and / or a doctor related to the user. Furthermore, the medical record data in the database can be summarized in the order of the time of the consultation, and the medical record data in the preset time period before the current one can be analyzed by a preset medical data model to generate health advice and send the health advice to the user. The electronic digitization and structuring of medical records are efficiently realized. It can solve the problems of the trouble of preserving paper medical records and the lengthy and easy-to-forget medical record content; further, the corresponding health advice can be generated according to the user's medical record data, which can solve the problem that the medical record content is too professional and difficult for the user to understand. And by encrypting the medical record data, the user's relevant personal privacy can be effectively protected from being leaked, and the storage security of the medical record data is improved. BRIEF DESCRIPTION OF THE DRAWINGS
[0055] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.
[0056] Figure 1 A schematic diagram of the simplified steps of the method in this application;
[0057] Figure 2 A schematic diagram of the steps included in step S100 of the method in this application;
[0058] Figure 3 A schematic diagram of one of the steps of the method in this application;
[0059] Figure 4 It is a schematic diagram of some steps of the method in this application;
[0060] Figure 5 It is a schematic diagram of some steps of the method in this application;
[0061] Figure 6 A schematic diagram of one of the steps of the method in this application;
[0062] Figure 7 Schematic diagram of the connection relationship between the modules of the device in this application.
[0063] Figure numerals: 1-image preprocessing module; 2-text processing module; 3-semantic recognition module; 4-structuring module; 5-encryption module; 6-database module; 7-personalized service and recommendation module. DETAILED DESCRIPTION
[0064] In the following, various embodiments of the present disclosure will be described more fully. The present disclosure may have various embodiments, and adjustments and changes may be made therein. However, it should be understood that there is no intention to limit the various embodiments of the present disclosure to the specific embodiments disclosed herein, but rather the present disclosure should be understood to cover all adjustments, equivalents and / or alternatives that fall within the spirit and scope of the various embodiments of the present disclosure.
[0065] The terms used in the various embodiments disclosed in the present invention are only used for the purpose of describing specific embodiments and are not intended to limit the various embodiments disclosed in the present invention. As used herein, the singular form is intended to also include the plural form, unless the context clearly indicates otherwise. Unless otherwise specified, all terms used herein (including technical terms and scientific terms) have the same meaning as the meanings commonly understood by ordinary technicians in the field of the various embodiments disclosed in the present invention. Terms (such as terms defined in generally used dictionaries) will be interpreted as having the same meaning as the contextual meaning in the relevant technical field and will not be interpreted as having an idealized meaning or an overly formal meaning, unless clearly defined in the various embodiments disclosed in the present invention.
[0066] Example 1
[0067] This embodiment discloses a method for identifying the content of a medical record image. The specific steps are as follows: Figure 1 As shown, including:
[0068] S100, obtaining a paper medical record image of a user, performing image preprocessing on the medical record image, and obtaining a processed image;
[0069] S200, performing optical character recognition on the processed image to obtain initial text information;
[0070] S300, performing semantic recognition related processing on the initial text information to obtain text data;
[0071] S400, extracting information and structuring information from the text data in sequence through the trained AI model to obtain medical record data;
[0072] S500, encrypting the medical record data in a preset encryption method and storing it in a database corresponding to the account of the corresponding user and / or a database corresponding to the account of the doctor associated with the user;
[0073] S600. For each user, the medical record data in the database is summarized in the order of the consultation time, and the medical record data within the preset time period before the current one is analyzed through the preset medical data model to generate health advice; and the health advice is sent to the user.
[0074] In some specific embodiments, Figure 2 As shown, the process of performing image preprocessing on the paper medical record image in step S100 includes:
[0075] S101, using a denoising algorithm to perform denoising processing on the paper medical record image, so that the image quality of the paper medical record image reaches a preset value; and / or
[0076] S102, adjusting the contrast, and / or brightness, and / or sharpness of the paper medical record image; and / or
[0077] S103, performing binarization processing on the paper medical record image.
[0078] In actual use, by preprocessing the paper medical record images, the image quality of the paper medical record images can be effectively improved, and the accuracy of the initial text information subsequently obtained can be further improved.
[0079] In some specific embodiments, Figure 3 As shown, step S300 specifically includes:
[0080] S301, performing noise removal processing on the initial text information to remove information irrelevant to the medical record in the initial text information;
[0081] S302 : performing word segmentation processing on the text in the initial text information after the noise removal processing based on semantic recognition to obtain text data.
[0082] In some specific embodiments, Figure 4 As shown, it also includes:
[0083] S701, if a medical record modification instruction fed back by a preset first person is received, determining the medical record data located in the database corresponding to the medical record modification instruction;
[0084] S702, sending the determined medical record data and the corresponding correction content in the medical record correction instruction to a preset second person;
[0085] S703: if a confirmation instruction from the preset second person is received, the determined medical record data is corrected based on the correction content; and a comparison before and after the correction is added to the determined medical record data;
[0086] S704. Update the medical record data in the database based on the corrected medical record data.
[0087] In actual applications, the preset first person and the preset second person can be any two of the three, including the reviewer, the doctor related to the user, and the user himself. This ensures that when there are errors or omissions in the medical record data, the medical record data will be modified and updated through at least two levels of confirmation, and at least one level of verification will be for medical-related professionals, such as reviewers and doctors. By setting up multiple layers of verification, the accuracy and authority of medical record data modification can be effectively improved.
[0088] In some specific embodiments, it also includes:
[0089] S810, if the health advice meets the preset serious condition, the health advice is sent to other users related to the user;
[0090] The process of sending health advice to users includes:
[0091] S811. Send health advice that meets the preset serious conditions to the user in multiple ways and multiple times.
[0092] In some specific embodiments, by configuring preset severity conditions, the current health advice can be evaluated for importance. In actual applications, the preset severity conditions can have multiple levels, wherein the higher the level of the preset severity conditions reached by the health advice, the more frequently and the more times the health advice is sent to the user, making it more likely that the user will view the health advice.
[0093] In some specific embodiments, Figure 5 As shown, it also includes:
[0094] S901, retrieve medical records of users in a preset region and / or preset age group and / or preset disease from a database;
[0095] S902. Determine the common medical record data and individual medical record data from the retrieved medical record data through a preset AI medical analysis model, and display the determined common medical record data and individual medical record data to a preset object.
[0096] In actual applications, the preset objects can be doctors or researchers in the medical system. By using AI medical models to determine the common medical record data and individual medical record data from the retrieved medical records, doctors or researchers in the medical system can quickly judge the overall characteristics and development trends of the patient group, and assist in further research.
[0097] In some specific embodiments, it also includes:
[0098] S1001. If a medical record query request is received, search the database for medical record data corresponding to the medical record query request;
[0099] S1002, determining whether the requester of the medical record query request is the user or the doctor corresponding to the user;
[0100] S1003. If the judgment result is no, the found medical record data is anonymized and sent to the requester.
[0101] In some specific embodiments, Figure 6 As shown, it also includes:
[0102] S311, extracting pathological data from the text data based on semantic recognition, and detecting whether there is personal information data related to the user in the pathological data;
[0103] S312, desensitizing the pathology data, hiding or replacing the user's personal information data in the pathology data;
[0104] S313, storing the desensitized pathological data into a preset pathological database.
[0105] By desensitizing text data and anonymizing medical record data, pathology-related content in medical records can be shared with doctors or relevant researchers while ensuring information security. This can form a pathology database and anonymous medical records for medical research, which helps advance medical-related research without risking leakage of users' personal information.
[0106] In some specific embodiments, it also includes:
[0107] Medical record data, health advice and paper medical record images are associated and stored in the database corresponding to the corresponding user's account, and search keywords are assigned according to preset rules to facilitate users to query medical record data, and / or health advice, and / or paper medical record images based on keywords.
[0108] By storing the original paper medical record images in association with the medical record data and health advice, users can quickly find and compare the original paper medical record images when they have questions about the content of the medical record data and / or health advice. When seeing a doctor, users can also provide the original paper medical record images to professional medical staff for further manual verification of the contents of the structured medical records.
[0109] In practical applications, in the process of generating health advice for the user in step S600, the keyword frequency statistics of the keywords in the user's medical record data can also be used to assist in the generation. The user's medical record data includes keywords such as disease types, and / or disease names, and / or symptom descriptions. By ranking these keywords by frequency, some keywords that are ranked high mean that the word appears more times in the user's medical record, which can better describe the user's health status. Therefore, these keywords can also be used as user feature tags. Therefore, health information including user feature tags or health information with a correlation with user feature tags higher than a preset value can be further subscribed in the information flow, and pushed to the user as part of the health advice. Among them, health information can be information including health detection and screening project recommendations, diseases and disease prevention, healthy lifestyles, drugs and drug use, public health information, etc. By performing targeted push, the corresponding health information can be displayed in the field of vision of more audiences with corresponding needs, which can improve the utilization rate of medical related information.
[0110] In a specific embodiment, if the user has a history of influenza, after public health information about the influenza season released by the Health Bureau appears in the health information flow, the health information can be captured based on the feature tag of the user's influenza history and pushed to the user.
[0111] This embodiment proposes a method for identifying the content of a medical record image. After multiple and multi-layer processing of paper medical record images, structured medical record data can be obtained. This medical record data can be quickly called from a database by the user and / or doctors related to the user. It can solve the problem of the trouble of saving paper medical records and the lengthy and easy-to-forget medical record content; further, generating corresponding health advice based on the user's medical record data can solve the problem of the medical record content being too professional and difficult for users to understand. And by encrypting the medical record data, it can effectively protect the user's relevant personal privacy from being leaked, and improve the storage security of medical record data. It can quickly digitize and standardize the storage of a large number of paper medical records, and can effectively promote the data digitization process of the medical system.
[0112] Example 2
[0113] This embodiment discloses a device for identifying the content of a medical record image, which can be used to implement any method in the above technical solution, including:
[0114] The image preprocessing module 1 is used to obtain the paper medical record image of the user, perform image preprocessing on the medical record image, and obtain a processed image;
[0115] The text processing module 2 is used to perform optical character recognition on the processed image to obtain initial text information;
[0116] Semantic recognition module 3, used to perform semantic recognition related processing on the initial text information to obtain text data;
[0117] Structuring module 4 is used to extract information and perform information structuring processing on text data in sequence through the trained AI model to obtain medical record data;
[0118] Database module 6, used to store medical record data;
[0119] The personalized service and recommendation module 7 is used to summarize the medical record data in the database for each user in the order of the consultation time, and analyze the medical record data within the current preset time period through a preset medical data model to generate health advice; and send the health advice to the user.
[0120] In some specific embodiments, an encryption module 5 is also included; the encryption module 5 is used to encrypt the medical record data in a preset encryption method and store it in a database corresponding to the corresponding user's account and / or a database corresponding to the user's related doctor's account.
[0121] The connection relationship between the modules is as follows: Figure 7 As shown, the image preprocessing module 1 is connected to the text processing module 2, the text processing module 2 is connected to the semantic recognition module 3, the structuring module 4 is respectively connected to the encryption module 5 and the personalized service and recommendation module 7, wherein the database module 6 is connected to the encryption module 5.
[0122] In actual application, the database module 6 includes a database for user accounts and a database for doctor accounts.
[0123] In a specific embodiment, medical record data, health advice and paper medical record images are associated and stored in a database corresponding to the account of the corresponding user or doctor, and search keywords are assigned according to preset rules to facilitate users or doctors to query medical record data, and / or health advice, and / or paper medical record images based on keywords.
[0124] By storing the original paper medical record images in association with the medical record data and health advice, users can quickly find and compare the original paper medical record images when they have questions about the content of the medical record data and / or health advice. When seeing a doctor, users can also provide the original paper medical record images to professional medical staff for further manual verification of the contents of the structured medical records.
[0125] Furthermore, a pathology database may be included to store the desensitized pathology data. The semantic recognition module 3 is also used to extract the pathology data in the text data based on semantic recognition, and detect whether there is personal information data related to the user in the pathology data; desensitize the pathology data, hide or replace the personal information data of the user in the pathology data; and store the desensitized pathology data in the pathology database.
[0126] In practical applications, this device can be a device with interactive pages, including a smart phone and a computer. This device can be portable, so that users can quickly call up their medical records anytime and anywhere. In a specific embodiment, the database corresponding to the user account can be deployed locally on the device or in the cloud, and the user can choose the deployment of the database according to their own needs.
[0127] In a specific use process, the user can obtain the user's paper medical record image by taking a photo or scanning and upload it to the image preprocessing module 1. In practical applications, this device can also be used in a medical system, and the staff can use this device to quickly digitize a large number of paper medical record images, which can enable the medical system and patients to quickly synchronize medical record data information, and can also effectively promote the process of electronic medical record data in the medical system.
[0128] This embodiment provides a device for identifying the content of medical record images, which can quickly process paper medical record images multiple times and multiple layers, and further obtain structured medical record data, and this medical record data can be quickly called from the database by the user and / or the doctor related to the user. Furthermore, the medical record data in the database can be summarized according to the chronological order of the consultation time, and the medical record data within the preset time period before the current one can be analyzed through a preset medical data model to generate health advice. The electronic digitization and structuring of medical records are achieved efficiently. It can solve the problems of the trouble of preserving paper medical records and the lengthy and easy-to-forget medical record content, and by encrypting the medical record data, it can effectively protect the user's relevant personal privacy from being leaked, thereby improving the storage security of medical record data.
[0129] Those skilled in the art will appreciate that the accompanying drawings are only schematic diagrams of a preferred implementation scenario, and the modules or processes in the accompanying drawings are not necessarily necessary for the implementation of the present invention. Those skilled in the art will appreciate that the modules in the devices in the implementation scenario can be distributed in the devices of the implementation scenario according to the description of the implementation scenario, or can be changed accordingly and located in one or more devices different from the present implementation scenario. The modules of the above-mentioned implementation scenarios can be combined into one module, or can be further divided into multiple sub-modules. The above-mentioned serial numbers of the present invention are only for description and do not represent the advantages and disadvantages of the implementation scenarios. Disclosed above are only a few specific implementation scenarios of the present invention, but the present invention is not limited thereto, and any changes that can be thought of by those skilled in the art should fall within the scope of protection of the present invention.
Claims
1. A method for identifying the content of a medical record image, characterized in that: include: Acquire a paper medical record image of a user, and perform image preprocessing on the medical record image to obtain a processed image; Performing optical character recognition on the processed image to obtain initial text information; Performing semantic recognition related processing on the initial text information to obtain text data; The trained AI model is used to extract information and structure the information of the text data in turn to obtain medical record data; Encrypting the medical record data in a preset encryption method and storing it in a database corresponding to the user's account and / or a database corresponding to the user's doctor's account; For each of the users, the medical record data in the database are summarized in the order of the consultation time, and the medical record data within the preset time period before the current time period are analyzed by a preset medical data model to generate health advice; And send the health advice to the user.
2. A method for identifying the content of a medical record image according to claim 1, characterized in that: The process of performing image preprocessing on the paper medical record image includes: Using a denoising algorithm to perform denoising on the paper medical record image so that the image quality of the paper medical record image reaches a preset value; and / or Adjusting the contrast, and / or brightness, and / or sharpness of the paper medical record image; and / or The paper medical record image is binarized.
3. The method for identifying the content of a medical record image according to claim 1, characterized in that: The initial text information is processed with semantic recognition to obtain text data including: Performing noise removal processing on the initial text information to remove information irrelevant to the medical record in the initial text information; Based on semantic recognition, the text in the initial text information after the noise removal process is performed is segmented to obtain text data.
4. The method for identifying the content of a medical record image according to claim 1, characterized in that: Also includes: If a medical record modification instruction fed back by a preset first person is received, determining the medical record data located in the database corresponding to the medical record modification instruction; Sending the determined medical record data and the corresponding correction content in the medical record correction instruction to a preset second person; If a confirmation instruction fed back by the preset second person is received, the determined medical record data is corrected based on the correction content; and adding the comparison before and after the correction to the determined medical record data; The medical record data in the database is updated based on the corrected medical record data.
5. The method for identifying the content of a medical record image according to claim 1, characterized in that: Also includes: If the health advice meets the preset serious condition, the health advice is sent to other users related to the user; The process of sending the health advice to the user comprises: The health advice that meets the preset serious conditions is sent to the user in multiple ways and multiple times.
6. The method for identifying the content of a medical record image according to claim 1, characterized in that: Also includes: Retrieving medical record data of the user in a preset region and / or preset age group and / or preset disease from the database; Determine the medical record common data and medical record individual data from the retrieved medical record data through a preset AI medical analysis model; And the determined common medical record data and individual medical record data are displayed to the preset objects.
7. The method for identifying the content of a medical record image according to claim 1, characterized in that: Also includes: If a medical record query request is received, searching the database for medical record data corresponding to the medical record query request; Determine whether the requester of the medical record query request is the user or the doctor corresponding to the user; If the judgment result is no, the found medical record data will be anonymized and then sent to the requester.
8. The method for identifying the content of a medical record image according to claim 1, characterized in that: Also includes: Extracting pathological data from the text data based on semantic recognition, and detecting whether personal information data related to the user exists in the pathological data; Desensitizing the pathological data to hide or replace the user's personal information data in the pathological data; The pathological data after desensitization treatment is stored in a preset pathology database.
9. A device for identifying the content of a medical record image, characterized in that: The method for implementing any one of claims 1 to 8 comprises: An image preprocessing module is used to obtain a paper medical record image of a user, and perform image preprocessing on the medical record image to obtain a processed image; A text processing module, used for performing optical character recognition on the processed image to obtain initial text information; A semantic recognition module, used to perform semantic recognition related processing on the initial text information to obtain text data; A structuring module is used to extract information and perform information structuring processing on the text data in sequence through a trained AI model to obtain medical record data; A database module, used for storing the medical record data; The personalized service and recommendation module is used to summarize the medical record data in the database for each user in the order of the consultation time, and analyze the medical record data within the current preset time period through a preset medical data model to generate health advice; and send the health advice to the user.
10. The device for identifying the content of a medical record image according to claim 9, characterized in that: It also includes an encryption module; the encryption module is used to encrypt the medical record data in a preset encryption method and store it in a database corresponding to the user's account and / or a database corresponding to the user's related doctor's account.
Citation Information
Cited By
Paper entity medical record recognition system and method based on image extraction fusion learning
CN121708612A
