Medical examination slip data digitization system and method

Abstract

To provide a system and method for efficiently and highly accurately converting medical examination slips that have been handwritten by patients before their examination into electronic information. [Solution] The medical examination ticket digitization system of the present invention is a system that converts medical questionnaires filled out by hand by patients into electronic information, and includes an optical scanner that converts the medical examination ticket into image data, a rule-based OCR that generates electronic information such as text information based on the image data generated by the optical scanner, an AI-OCR that generates electronic information such as text information based on the image data generated by the optical scanner, and a matching unit that inputs the electronic information generated by the rule-based OCR and the AI-OCR and determines whether the two information match or do not match.

Description

[Technical Field]

[0001] The present invention relates to a system and method for efficiently and accurately converting into electronic information medical examination forms that examinees handwrite before (or after) their medical examinations, such as mobile health checkups, or examinations at medical institutions. [Background technology]

[0002] In the past, in health checkups such as mobile health checkups, patients would handwrite their medical records before their visit, and administrative staff would then visually input the data. This process required a lot of time and manpower, and there was also the problem of detecting and correcting input errors, which required a lot of effort. In recent years, attempts have been made to digitize handwritten medical examination slips using OCR technology (for example, Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2003-058642 Summary of the Invention [Problem to be solved by the invention]

[0004] However, conventional OCR has low recognition accuracy for unclearly written characters and special symbols, and manual confirmation is still required. An object of the present invention is to provide a system and method for efficiently and accurately converting a medical examination slip that a patient has handwritten before visiting a doctor into electronic information. [Means for solving the problem]

[0005] In order to achieve the above object, the present invention is summarized as follows. [1] A system that converts handwritten medical records into electronic information. An optical scanner that converts the medical examination slip into image data, Rule-based OCR generates electronic information such as text information based on image data generated by an optical scanner; AI-OCR generates electronic information such as text information based on image data generated by an optical scanner; A medical examination ticket data digitization system that has a matching unit that inputs the electronic information generated by rule-based OCR and AI-OCR and determines whether the two match or not. [2] For character information that is determined to be inconsistent by the matching unit, The electronic medical examination ticket data system described in [1] is provided with a correction unit that receives a user selection operation to select between electronic data generated by rule-based OCR and electronic data generated by AI-OCR, and / or an input operation from the user to input data equivalent to character information determined to be a mismatch, and generates corrected data based on the selection operation and / or the input operation. [3] A method for converting a medical examination form handwritten by a patient into electronic information, comprising: The medical examination slip is converted into image data using an optical scanner, Rule-based OCR processing generates electronic information such as text information based on image data generated by an optical scanner. AI-OCR processing generates electronic information such as text information based on image data generated by an optical scanner. A method for digitizing medical examination ticket data by inputting the electronic information generated by rule-based OCR and AI-OCR and determining whether the two match or not. [4] Regarding the text information that is determined to be inconsistent, A method for digitizing medical examination ticket data described in [3], which accepts a user selection operation to select between electronic data generated by rule-based OCR and electronic data generated by AI-OCR, and / or an input operation to input data equivalent to character information determined to be a mismatch, and generates corrected data based on the selection operation and / or input operation. [Effects of the Invention]

[0006] According to the system and method for digitizing medical examination slip data of the present invention, medical examination slips that have been handwritten by examinees before their examination can be efficiently and highly accurately digitized. [Brief explanation of the drawings]

[0007] [Figure 1] 1 is a diagram showing the configuration of an examination slip data digitization system according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0008] The present invention will be described in detail below with reference to the accompanying drawings. The medical examination slip data digitization system and method according to the present invention can be used to digitize medical examination slips handwritten by examinees and examiners in health checkups such as mobile health checkups.

[0009] 1 is a diagram showing the configuration of an examination slip data digitization system according to an embodiment of the present invention. The examination slip data digitization system (100) according to an embodiment of the present invention includes a scanner (102) with an OCR function that reads an examination slip (101) that has been handwritten by a patient before an examination and outputs image data, and an AI-OCR (machine learning / deep learning OCR) (103).

[0010] The scanner with OCR function (102) includes an optical scanner (1021) and a rule-based OCR (1022), which is a conventional OCR. The optical scanner (1021) converts the medical examination slip (101) into image data and can output the image data in various formats (JPEG, GIF, PDF, etc.). The optical scanner (1021) converts the medical examination slip (101) into any image data and outputs it to the rule-based OCR (1022). The optical scanner (1021) also converts the medical examination slip (101) into image data in PDF format and outputs it to the AI-OCR (103). The data input to the rule-based OCR (1022) and the AI-OCR (103) may be in any data format, as long as it can be processed by the rule-based OCR (1022) and the AI-OCR (103). The data input to the two OCRs may be in the same data format or different data formats.

[0011] The rule-based OCR (1022) recognizes characters by comparing the character shapes (patterns) in image data with templates and dictionaries. The rule-based OCR (1022) generates electronic information such as text information based on image data generated by the optical scanner (1021). The AI-OCR (103) improves OCR accuracy by using AI such as neural networks. Similarly, the AI-OCR (103) generates electronic information such as text information based on image data (PDF data in the example of Figure 1) generated by the optical scanner (1021). The AI-OCR (103) may perform machine learning on a computer or may use an AI engine and learning data on the cloud.

[0012] As described above, the examination slip (101) is converted into image data by a single optical scanner (1021), thereby creating image data to be processed by the rule-based OCR (1022) and AI-OCR (103). The rule-based OCR (1022) and AI-OCR (103) use their respective discrimination methods to convert the digitized image data into character information (including characters, symbols, numbers, etc.), and output it to the collation unit (105). The collation unit (105) compares the character information input from the rule-based OCR (1022) and AI-OCR (103), respectively, and determines whether the two processing results match or do not match.

[0013] The collation unit (105) displays the discrepancies on an external display unit or the like based on the collation results. The administrative staff (user) checks the discrepancies and corrects the data by referring to the PDF of the medical examination slip (101) or the like. The correction unit (106) accepts the results of the check and correction by the administrative staff, reflects the accepted correction results in the data, and generates corrected data (107). For example, the correction unit (106) can be configured to display both the character information determined by the rule-based OCR (1022) and the AI-OCR (103) on an external display unit or the like, and allow the administrative staff (user) to select one of them. In addition, upon the user's selection operation, the correction unit (106) may also display the corresponding portion of the image data related to the discrepancy on the display unit or the like. In addition, if the character information from both the rule-based OCR (1022) and the AI-OCR (103) is inappropriate, the correction unit (106) can be configured to accept a user input operation to manually input correct information by keyboard operation or the like.

[0014] The corrected data (107) is registered in a test result management database (108) and used for statistical processing and report creation. Each component of the examination slip data digitization system (100) is realized by a computer executing a predetermined program. The program may be stored in the computer itself, or in an external memory or cloud storage connectable to the computer. [Example]

[0015] As an example of using the examination slip data digitization system (100) according to the embodiment of the present invention, a method of using it in a mobile medical checkup will be described.

[0016] 1) Preparation The health checkup (test) provider obtains from the client's business a list of employees scheduled to undergo health checkups (subject information). Based on the subject information, the test provider creates a medical examination ticket (questionnaire) for each employee and sends it to the client. The business distributes the medical examination tickets to employees. The employees undergoing the examination handwrite the answers to the questions on the medical examination ticket.

[0017] 2) Inspection The employee who is to receive the medical examination visits the testing site with the medical examination form on which they have completed the medical questionnaire. The employee (the examinee) undergoes each test at the site (health checkup). Height, weight, eyesight, etc. are measured at the venue, and the results are available on the spot, and for examinations by a doctor, the results are entered on the examination slip (101) at the venue. After the examination is completed, the examination slip (101) is collected from the patient. For blood tests, blood is drawn at the testing site, and the sample is sent to a testing company for testing.

[0018] 3) Result processing The examination slip (101) collected from the examinee at the examination site is read by the optical scanner (1021). The examination slip (101) is converted into image data by the optical scanner (1021), and is then digitized into text information by the rule-based OCR (1022) and the AI-OCR (103). The collation unit (105) inputs the results of the digitization by the rule-based OCR (1022) and the AI-OCR (103) and determines whether they match or mismatch. Based on the collation results of the collation unit (105), any mismatches are displayed on an external display unit or the like.

[0019] The administrative staff (user) checks the discrepancies and corrects the data by referring to the PDF of the examination ticket (101) or the like. The correction unit (106) accepts the data input operation by the administrative staff (user), reflects the correction results in the data, and generates corrected data (107). Specifically, the correction unit (106) accepts a selection operation by the user to select electronic data converted by rule-based OCR or electronic data converted by AI-OCR for the character information determined to be a mismatch by the comparison unit (105). Alternatively, the correction unit (106) accepts an input operation by the user to input data corresponding to the character information determined to be a mismatch. The correction unit (106) generates corrected data based on the selection operation and / or input operation by the user. The corrected data (107) is imported into the health checkup result management database (108), which is a core system. Blood test result data is received from the testing contractor and imported into the test result management database (108). The results of the doctor's assessment of X-rays, ultrasound examinations, electrocardiograms, etc. are entered into a medical examination result management database (108).

[0020] Based on the data imported into the test result management database (108) as described above, test determination result data for each examinee is created, and a medical examination result sheet for each examinee is prepared. The medical examination result sheets for each site are compiled and delivered to the customer in paper or digital form. In the above embodiment, the medical examination slip data digitization system (100) is applied to the process of digitizing medical examination slips, but the present invention is not limited to this and can be applied to the digitization of medical examination slips in general. For example, the medical examination slip data digitization system (100) according to this embodiment can be used at the reception desk of a medical institution. During the reception work at a medical institution, the medical examination slip can be digitized each time it is received, or multiple medical examination slips can be collectively digitized.

[0021] According to the present invention, by converting the medical examination slip (101) into image data using a single optical scanner (1021) and using both rule-based OCR processing and AI-OCR processing, it is possible to improve the accuracy of digitizing handwritten information into electronic information such as character information while keeping hardware costs low. The rule-based OCR (1022) can perform character recognition with high accuracy if the character font and size are limited, but it has difficulty recognizing broken handwritten characters. On the other hand, the AI-OCR (103) can improve recognition accuracy by learning context and patterns, and can handle broken handwritten characters, but its accuracy may be lower than that of the rule-based OCR (1022) for neat characters. Therefore, in the examination slip data digitization system (100) according to the present invention, OCR processing is performed by both the rule-based OCR (1022) and the AI-OCR (103), and the character information from the rule-based OCR (1022) and the AI-OCR (103) is compared by the comparison unit (105), and only the mismatched parts are notified to the user, thereby making use of the characteristics of both systems to digitize the examination slip (101). Furthermore, by accepting user operations, manual confirmation points can be minimized, and the number of work steps can be significantly reduced.

[0022] In this way, by combining the rule-based OCR (1022), which has a high character recognition rate for well-formed characters, with the AI-OCR (103), which has a high character recognition rate for disorganized characters, it is possible to recognize characters of various styles with high accuracy. This property can also be utilized to achieve further automation. For example, if the rule-based OCR (1022) and the AI-OCR (103) make different judgments, i.e., if the discrimination unit (105) determines that they do not match, the image data of the target handwritten character can be analyzed, and if the target character is well-formed, the result of the rule-based OCR (1022) is adopted, and if the target character is not well-formed, the result of the AI-OCR (103) is adopted. The degree of neatness of handwritten characters can be determined using a similarity index, which compares the character in the input image with a pattern of a font of well-formed characters and scores the similarity between them, or indices such as the feature values ​​of the character shape, the smoothness of the outline, the continuity of the strokes, and the distribution of the curvature. Adding such processing can achieve further automation. The automatically processed determined portion may be displayed to inform the user.

[0023] The present invention is not limited to the above-described embodiments, and various design modifications can be made without departing from the spirit of the present invention. [Explanation of symbols]

[0024] 101 Medical Examination Ticket 102 Scanner with OCR function 1021 Optical Scanner 1022 Rule-based OCR 103 AI-OCR 105 Matching unit 106 Correction section 107 revised data 108 Test result management database

Claims

1. A system that converts a medical examination slip that a patient has handwritten into electronic information, An optical scanner that converts the medical examination slip into image data, A rule-based OCR that generates electronic information such as character information based on image data generated by an optical scanner; An AI-OCR that generates electronic information such as character information based on image data generated by an optical scanner; A medical examination ticket data digitization system that has a matching unit that inputs the electronic information generated by the rule-based OCR and the AI-OCR and determines whether the two match or not.

2. Regarding the character information that is determined to be inconsistent by the collating unit, 2. The electronic examination ticket data system of claim 1, further comprising: a correction unit that receives a user selection operation to select between electronic data generated by rule-based OCR and electronic data generated by AI-OCR, and / or an input operation from the user to input data corresponding to character information determined to be a mismatch, and generates corrected data based on the selection operation and / or the input operation.

3. A method for converting a medical examination slip that has been handwritten by a patient into electronic information, comprising: The medical examination slip is converted into image data using an optical scanner, Rule-based OCR processing generates electronic information such as text information based on image data generated by an optical scanner, AI-OCR processing generates electronic information such as text information based on image data generated by an optical scanner, A method for digitizing medical examination ticket data by inputting electronic information generated by rule-based OCR and AI-OCR and determining whether the two match or not.

4. Regarding the character information that is determined to be inconsistent, The method for digitizing medical examination ticket data described in claim 3 accepts a user selection operation to select between electronic data generated by rule-based OCR and electronic data generated by AI-OCR, and / or an input operation to input data corresponding to character information determined to be a mismatch from the user, and generates corrected data based on the selection operation and / or the input operation.

Images