Medical chart creation assistance device and medical chart creation assistance method

Abstract

A server according to one aspect of the present invention comprises: a text division unit that divides a free text including information obtained from a patient by splitting the free text into individual clauses; and a clause classification unit that classifies each of the clauses divided by the text division unit into any of a plurality of classifications based on the entry format of a medical chart in accordance with the content indicated by the clause and outputs a classification result.

Description

Medical record creation support device and medical record creation support method

[0001] The present invention relates to a medical record creation support device and a medical record creation support method.

[0002] Many dentists learn about medical science, such as diagnosis and treatment methods, when they attend dental school, but they are often not given the opportunity to learn in depth how to write medical records. As a result, many dentists graduate from dental school without learning how to write medical records and go straight into clinical practice.

[0003] However, providing high-quality dental care is not achieved by simply improving treatment techniques; recording patient information in medical records without omissions and without delay is also an important element in providing high-quality dental care. Since dentists are often replaced at dental clinics, it is ideal to record medical records in a way that allows any dentist to understand the patient's condition equally well as the attending dentist.

[0004] The "SOAP format," which is used to fill out medical records at medical institutions, is a well-known method for maintaining a consistent level of quality in the created medical records. The SOAP format is a method in which information obtained from patients through medical examinations and other procedures is divided into four sections: S (subjective), O (objective), A (assessment), and P (plan).

[0005] "S" indicates subjective information, and S classifies the content of the patient's chief complaint, etc. "O" indicates objective information, and O classifies objective information obtained from examinations and tests, etc. "A" indicates evaluation, and A classifies the doctor's diagnosis and the overall evaluation as a result of analysis or interpretation based on the content of S and O, etc. "P" indicates plan, and P classifies the treatment policy and plan decided based on "A."

[0006] For example, Patent Document 1 describes an electronic medical record system in which diseases selected from a list of diseases displayed on an examination screen, the doctor's findings, and subsequent medical guidelines are copied into SOAP text, and the copied content is recorded in the medical record.

[0007] Japanese Patent Application Laid-Open No. 2002-215794

[0008] However, as mentioned above, it is also true that many dentists have not thoroughly studied how to write medical records. Therefore, when recording the information obtained during a medical examination, inexperienced dentists may struggle with how to create the medical record. Furthermore, such dentists often incorrectly classify and record the information obtained during the examination in SOAP. This issue is not limited to dentists, but is common to all medical professionals who record medical records.

[0009] The present invention has been made in consideration of the above circumstances, and an object of the present invention is to enable appropriate creation of a medical record based on a medical record entry format.

[0010] A medical record creation support device according to one embodiment of the present invention comprises a text division unit that divides free text containing information obtained from a patient into segments, and a segment classification unit that classifies each of the segments divided by the text division unit into one of a plurality of segments based on the medical record entry format in accordance with the content indicated by the segment, and outputs the classification result. The segment classification unit is configured using a learning model that has learned correspondence information between the segmentations contained in the free text and the segments into which the segmentations should be classified. The segment classification unit classifies the segmentations divided by the text division unit into all segments that can be considered as classification targets, and outputs the correspondence information between the segmentations and the segments and information on the degree of certainty of the classification into the segments as the classification result.

[0011] According to at least one aspect of the present invention, it is possible to appropriately create a medical record based on a medical record entry format. Problems, configurations, and effects other than those described above will become apparent from the following description of the embodiments.

[0012] FIG. 1 is a diagram showing an example of a schematic configuration of a medical record creation support system according to one embodiment of the present invention. FIG. 2 is a block diagram showing an example of a configuration of a control system of a server and a terminal device according to one embodiment of the present invention. FIG. 3 is a diagram showing an example of text input to a terminal device according to one embodiment of the present invention. FIG. 4 is a diagram showing an example of training data of a phrase classification unit according to one embodiment of the present invention. FIG. 5 is a diagram showing an example of correspondence between phrases after classification by a phrase classification unit according to one embodiment of the present invention and classification certainty. FIG. 6 is a diagram showing an example of sentences shaped using phrases by a sentence shaping unit according to one embodiment of the present invention. FIG. 7 is a diagram showing an example of a configuration of a medical record creation support screen according to one embodiment of the present invention. FIG. 8 is a diagram showing an example of a configuration of a medical record creation screen according to one embodiment of the present invention. FIG. 9 is a diagram showing an example of a display of classification results on a medical record creation screen according to one embodiment of the present invention. FIG. 10 is a diagram showing an example of a display of a medical record in a case where sentences according to one embodiment of the present invention are displayed in correspondence with each SOAP category. FIG. 11 is a diagram showing an example of a display of a medical record in a case where only sentences according to one embodiment of the present invention are displayed. FIG. 12 is a flowchart showing an example of the procedure of a medical record creation support process by a medical record creation support system 100 according to one embodiment of the present invention.

[0013] Hereinafter, examples of modes for carrying out the present invention (hereinafter referred to as "embodiments") will be described with reference to the accompanying drawings. Various numerical values ​​in the embodiments of the present invention are merely examples, and the present invention is not limited to the embodiments described below. Furthermore, in this specification and drawings, identical components or components having substantially the same functions will be assigned the same reference numerals, and redundant explanations will be omitted.

[0014] <General Configuration of Medical Record Creation Support System> First, the configuration of a medical record creation support system 100 according to an embodiment of the present invention will be described with reference to Fig. 1. Fig. 1 is a diagram showing an example of the general configuration of the medical record creation support system 100.

[0015] 1, the medical record creation support system includes a server 1 (an example of a medical record creation support device) and terminal devices 2-1 to 2-n (n is a natural number equal to or greater than 2). The server 1 and each of the terminal devices 2-1 to 2-n are communicably connected via a network N.

[0016] The server 1 is placed in, for example, a cloud environment, generates sentences to be transcribed into medical records from free text (hereinafter simply referred to as "text") sent from multiple terminal devices 2, and sends the generated sentences to the terminal devices 2. More specifically, the server 1 divides the text sent from the multiple terminal devices 2 into phrases and classifies the phrases into any of the SOAP categories. The server 1 then selects phrases to be transcribed into medical records from the classified phrases, forms the selected phrases into sentences, and sends them to the terminal devices 2.

[0017] The terminal device 2 is a device configured from a PC (Personal Computer) installed in a dental clinic or a mobile terminal such as a smartphone. Note that while only two terminal devices 2, terminal devices 2-1 and 2-n, are shown in FIG. 1, it is assumed that multiple other terminal devices 2 are also connected to the server 1. In the following description, when it is not necessary to distinguish between the terminal devices 2-1 to 2-n, they will be collectively referred to as terminal device 2.

[0018] <Configuration of Control System of Medical Record Creation Support System> Next, the configuration of the control system of the server 1 and the terminal device 2 that constitute the medical record creation support system 100 will be described with reference to Fig. 2. Fig. 2 is a block diagram showing an example of the configuration of the control system of the server 1 and the terminal device 2.

[0019] 2, the server 1 includes a control unit 10, a memory unit 11, an input unit 12, a display unit 13, a communication I / F (Interface) unit 14, a text segmentation unit 15, a phrase classification unit 16, a phrase selection unit 17, and a sentence formatting unit 18, which are all connected to a bus B1. The control unit 10 is a computing device that includes a CPU (Central Processing Unit) 101, a ROM (Read Only Memory) 102, and a RAM (Random Access Memory) 103.

[0020] The CPU 101 reads out from the ROM 102 the program code of the software that realizes each function of the server 1 according to this embodiment, and loads it into the RAM 103 for execution. The server 1 may include a processing device such as an MPU (Micro-Processing Unit) instead of the CPU 101. Variables, parameters, etc. that are generated during the calculation process by the CPU 101 are temporarily written to the RAM 103.

[0021] The storage unit 11 may be, for example, a hard disk drive (HDD), a solid state drive (SSD), a flexible disk, an optical disk, a magneto-optical disk, a CD-ROM, a CD-R, a non-volatile memory card, etc. The storage unit 11 stores an operating system (OS), various parameters, and programs for operating the server 1.

[0022] The program for causing the server 1 to function may be stored in the ROM 102. The program is stored in the form of computer-readable program code. The CPU 101 sequentially executes operations in accordance with the program code. In other words, the ROM 102 or the storage unit 11 is an example of a computer-readable non-transitory recording medium that stores a program to be executed by a computer.

[0023] The input unit 12 is configured with, for example, a mouse and a keyboard, and generates operation signals in response to user operations and supplies the operation signals to the CPU 101. The display unit 13 is, for example, a monitor configured with an LCD (Liquid Crystal Display) or the like. The input unit 12 and the display unit 13 may be integrated into a touch panel. The server 1 may also be configured without the input unit 12 and the display unit 13.

[0024] The communication I / F 14 is configured by a communication device that controls communication with the terminal device 2, etc.

[0025] The text segmentation unit 15 segments the text transmitted from the terminal device 2 into multiple phrases based on information such as punctuation marks, line breaks, spaces, etc. in the text. The text segmentation unit 15 is configured by, for example, a program for segmenting phrases, an AI (Artificial Intelligence) model that has undergone machine learning, and a language model such as LLM (Large Language Models).

[0026] The AI ​​model that has undergone machine learning can be a learning model that has learned correspondence information between various sentences and each of the phrases that make up the sentences as learning data. Furthermore, when the text segmentation unit 15 is configured using a language model such as LLM, the input text can be segmented into phrases by inputting a prompt such as "Please segment the following sentence into phrases" into the language model.

[0027] The phrase classification unit 16 is composed of an AI model or the like that has been fine-tuned by learning from medical records created in SOAP format by experienced dentists or the like. It is believed that medical records created by experienced dentists or the like appropriately associate phrases contained in the text with the SOAP categories into which these phrases should be classified. In other words, the phrase classification unit 16 learns correspondence information between phrases contained in the text and the appropriate SOAP categories into which these phrases should be classified. The phrase classification unit 16 then classifies each segment divided by the text division unit 15 into each SOAP category based on the learned content.

[0028] Note that the phrase classifier 16 does not classify each phrase segmented by the text segmenter 15 into only one of the SOAP categories, but instead classifies it into multiple categories according to the degree of certainty of the classification. The phrase classifier 16 then outputs to the terminal device 2 the category into which the phrase has been classified with the highest degree of certainty and information about the degree of certainty of classification into this category. The phrase classifier 16 also outputs to the terminal device 2 other categories into which the same phrase has been classified and information about the degree of certainty of classification into each category. Note that if there is a phrase that could not be classified into any of the SOAP categories, the phrase classifier 16 also outputs information indicating that the phrase is unclassified.

[0029] The phrase selection unit 17 selects phrases to be transcribed into the medical record from among the phrases classified by the phrase classification unit 16. Specifically, the phrase selection unit 17 deletes phrases that are inappropriate for transcription into the medical record from among the phrases classified by the phrase classification unit 16. Phrases that are inappropriate for transcription into the medical record include, for example, phrases where the difference between the minimum and maximum values ​​of the confidence levels associated with the categories into which the phrase is classified is equal to or less than a predetermined threshold, and phrases that are not classified into any of the SOAP categories.

[0030] For example, a phrase that is weakly related to the contents of the medical examination has a low degree of certainty as to which SOAP category it belongs to, and therefore the difference between the minimum and maximum certainty values ​​is small. Therefore, the phrase selection unit 17 excludes such phrases from being transcribed into the medical record.

[0031] The sentence formatting unit 18 formats each phrase classified into one of the SOAP categories into natural-sounding sentences suitable for recording in a medical record. The sentences formatted by the sentence formatting unit 18 are sentences from which unnecessary phrases have been deleted by the phrase selection unit 17. When the sentence formatting unit 18 is configured using an LLM, the sentence formatting unit 18 can format the sentences into natural-sounding sentences suitable for transcription into a medical record by inputting phrases, SOAP classification results, and prompts into the LLM. For example, the prompt can be configured with text such as, "You are a dentist. You are about to create a SOAP-based medical record. Please format the sentences so that they sound natural Japanese based on the input phrases and SOAP classification information."

[0032] The results of the classification of phrases into each SOAP category by the phrase classification unit 16 and the sentences formatted by the sentence formatting unit 18 are sent to the terminal device 2 and displayed on the screen (see FIG. 9 ) of the terminal device 2. A user such as a dentist can check the contents displayed on the screen of the terminal device 2 and correct the correspondence between the phrases and the classification results into each SOAP category, or correct the formatted sentences.

[0033] [Terminal Device] Next, the configuration of the control system of the terminal device 2 will be described with reference to Fig. 2. The terminal device 2 includes a control unit 20, a storage unit 21, an input unit 22, a display unit 23, a communication I / F unit 24, a display control unit 25, and a medical record transcription unit 26, which are all connected to a bus B2. The control unit 20 is a calculation device including a CPU 201, a ROM 202, and a RAM 203.

[0034] The CPU 201 reads out program code of software that realizes each function according to this embodiment from the ROM 202, expands it in the RAM 203, and executes it. Alternatively, the CPU 201 reads out the program code directly from the ROM 202 and executes it as is. The terminal device 2 may include a processing device such as an MPU instead of the CPU 201. Variables, parameters, and the like that are generated during the calculation process by the CPU 201 are temporarily written to the RAM 203.

[0035] The storage unit 21 may be, for example, an HDD, SSD, flexible disk, optical disk, magneto-optical disk, CD-ROM, CD-R, non-volatile memory card, etc. In addition to the OS and various parameters, the storage unit 21 also stores programs for operating the terminal device 2, medical records, etc.

[0036] The program for causing the terminal device 2 to function may be stored in the ROM 202. The program is stored in the form of computer-readable program code. The CPU 201 sequentially executes operations in accordance with the program code. In other words, the ROM 202 or the storage unit 21 is used as an example of a computer-readable non-transitory recording medium that stores a program to be executed by a computer.

[0037] The input unit 22 is configured with, for example, a mouse, keyboard, microphone, etc., and generates operation signals in response to user input and supplies them to the CPU 201. Information obtained from the patient, such as the patient's chief complaint obtained through a medical interview, the results of an intraoral examination, and the results of an X-ray examination, is input by the user to the input unit 22 in text form. Note that the information obtained from the patient may also be input by voice via a microphone.

[0038] The display unit 23 is, for example, a monitor configured with an LCD or the like. For example, a medical record creation support screen Sc (see FIG. 8 etc.) that supports the creation of a medical record is displayed on the display unit 23. The input unit 22 and the display unit 23 may be integrated into a touch panel.

[0039] The communication I / F 24 is composed of a communication device or the like that controls communication with the server 1. The display control unit 25 controls the display of the medical record creation support screen Sc or the like on the display unit 23. The medical record transcription unit 26 transcribes the formatted text sent from the server 1 into the medical record.

[0040] <Example of Text Input to Terminal Device> Next, a description will be given of an example of text input to the terminal device 2 by a user such as a dentist, i.e., an example of input of information obtained from a patient through a medical examination, etc. Fig. 3 is a diagram showing an example of text input to the terminal device 2.

[0041] Figure 3 shows three examples of text input, #1 to #3. In Figure 3, phrases that should be classified as S (subjective information) are underlined with a solid line, and phrases that should be classified as O (objective information) are underlined with a dashed line. Furthermore, phrases that should be classified as A (evaluation) are underlined with a double-dashed line, and phrases that should be classified as P (plan) are underlined with a dashed line. Note that the classification information for phrases into each SOAP category may be displayed by color coding or the like.

[0042] For example, in Example #1, the phrase "I would like my teeth cleaned" should be classified as "S", and the phrase up to "Gum redness (+) (omitted) Plaque: little" should be classified as "O".

[0043] In Example #2, the phrase "tartar deposits are observed" should be classified as "O." "Avoid brushing too hard from side to side when brushing" is a phrase that is difficult to classify into any of the SOAP categories, so it is not underlined to indicate the type of category it is classified into. "The condition of the oral cavity is improving" is a phrase that should be classified as "A," and "Check brushing" is a phrase that should be classified as "P." "I would like my cavities treated" is a phrase that should be classified as "S."

[0044] <Example of Learning Data for Phrase Classification Unit> Next, the learning data for the phrase classification unit 16 will be described. Fig. 4 is a diagram showing an example of learning data for the phrase classification unit 16. In the example shown in Fig. 4, the learning data includes the phrase "I would like my teeth cleaned" associated with the category "S", the phrase "periodontal examination + scaling" associated with the category "P", and the phrase "tartar deposits were found" associated with the category "O".

[0045] By creating learning data based on medical records prepared by experienced dentists, the phrase classification unit 16, trained using the learning data, can appropriately classify input phrases into the correct (to be classified) category of SOAP.

[0046] <Example of Correspondence Between Phrase Classified by Phrase Classification Unit and Certainty of Classification> Next, an example of phrases classified by the phrase classification unit 16 and the certainty of classification into sections will be described. Fig. 5 is a diagram showing an example of correspondence between phrases classified by the phrase classification unit 16 and the certainty of classification.

[0047] The text column at the top of Figure 5 shows the phrase "There is a tendency for improvement due to the effects of brushing," and the classification result column shows that each SOAP category is classified with a certainty of S = "0.2," O = "0.05," A = "0.7," and P = "0.05." In the example shown in Figure 5, the certainty of the classification is shown by the Softmax function, and according to the Softmax function, the sum of the certainty is always 1.0. Note that the certainty of the classification into categories may also be shown by a method other than the Softmax function.

[0048] 5, the confidence level for classification into category "A" is the highest at "0.7," so the phrase classifying unit 16 classifies the phrase "There is a tendency for improvement due to the effects of brushing" into category "A." The phrase classifying unit 16 then outputs the phrase classification results and information on the confidence level for each category into which the phrase has been classified.

[0049] In the second-to-bottom example in Figure 5, the classification results for the phrase "Okinawa travel ni ikireta sou desu ka" (It seems they went on a trip to Okinawa) are "S: 0.3, O: 0.2, A: 0.25, P: 0.25." Because the phrase "Okinawa travel ni ikireta sou desu ka" (It seems they went on a trip to Okinawa) is difficult to classify into any SOAP category (it has little relevance), the confidence values ​​are similar across all SOAP categories. Therefore, the difference between the maximum confidence value of "0.3" and the minimum confidence value of "0.2" is also small. Note that the text (phrases) in lines 2 to 5 and line 7 in Figure 5 can also be considered in the same way.

[0050] In this embodiment, the phrase selection unit 17 deletes phrases where the difference between the maximum and minimum confidence values ​​is equal to or less than a predetermined value. This prevents phrases that are difficult to classify into each SOAP category from being transcribed into the medical record. The "predetermined value" can be set to a value such as 0.1 as an initial value, and it is preferable that this value be adjustable by the user while viewing the classification results by the phrase selection unit 17. The user's adjustment range can be set to, for example, a range from 0.0 to 1.0. The medical record creation support system 100 may transcribe information indicated by phrases that are difficult to classify into each SOAP category into the medical record as accompanying information of the patient, etc.

[0051] <Examples of Sentence Formatting by Sentence Formatting Unit> Next, examples of sentence formatting by the sentence formatting unit 18 will be described. Fig. 6 is a diagram showing examples of sentences formatted by the sentence formatting unit 18 using phrases. Example #1 in Fig. 6 shows the following sentence: "The patient visited the clinic complaining of wanting a teeth cleaning. The findings were gingival redness (+), tartar (+), no pus discharge, and pocket depths of 90.9% mild, 8.0% moderate, and 1.1% severe, with little plaque."

[0052] This sentence was generated based on the text of Example #1 in Figure 3. The types of underlining added to the sentence shown in Figure 6 indicate the types of SOAP sections into which the phrases are classified, similar to those shown in Figures 3 to 5. Sentences that are not underlined are portions added by the sentence formatting unit 18. Using the phrases classified into each SOAP section, the sentence formatting unit 18 can generate sentences that are natural in Japanese and suitable for entry in a medical record, as shown in Examples #1 to #3 in Figure 6.

[0053] <Configuration of Medical Record Creation Support Screen and Example of Screen Transition> Next, a description will be given of the configuration of the medical record creation support screen Sc displayed on the display unit 23 of the terminal device 2, and an example of screen transitions on the medical record creation support screen Sc. First, the configuration of the medical record creation support screen Sc will be described with reference to Fig. 7. Fig. 7 is a diagram showing an example of the configuration of the medical record creation support screen Sc.

[0054] The medical record creation support screen Sc shown in Fig. 7 is a screen displayed when creating a medical record for patient X. Menus such as "File," "Edit," and "View" are displayed at the top of the medical record creation support screen Sc. Below the menu display area, three areas are provided, from left to right on the screen: an oral cavity information display area Ar1, a medical record creation area Ar2, and a patient information display area Ar3.

[0055] The oral cavity information display area Ar1 displays information about the patient's oral cavity. The medical record creation area Ar2 displays information related to the creation of a medical record. The patient information display area Ar3 displays information about the patient, such as the patient's age, gender, and medical history. Menus such as "Specify Date" and "Search" are displayed at the top of the medical record creation area Ar2, and an AI menu Ma is provided at the right end of the menu. This AI menu Ma is a menu for executing the medical record creation support process according to this embodiment. When this AI menu Ma is selected by the user, the medical record creation screen Sc1 shown in the following Figure 8 is popped up on the medical record creation support screen Sc.

[0056] 8 is a diagram showing an example of the configuration of the medical record creation screen Sc1. The medical record creation screen Sc1 shown in FIG. 8 has a text input area Ar11, and a classification button Bn1 is provided to the right of the text input area Ar11. In the example shown in FIG. 8, information obtained from the patient through a medical examination or the like is entered in the text input area Ar11. The text entered in the text input area Ar11 is the same as the text shown in example #3 in FIG. 3.

[0057] The classification button Bn1 is a button for instructing classification of the text entered in the text input area Ar11 into each SOAP category. When the classification button Bn1 is pressed, the text entered in the text input area Ar11 is divided into phrases by the text dividing unit 15, and then classified into one of the SOAP categories by the phrase classifying unit 16.

[0058] 9 is a diagram showing an example of display of the classification results on the medical record creation screen Sc1. The medical record creation screen Sc1 shown in FIG. 9 has a classification result display area Ar12 and a sentence display area Ar13 below a text input area Ar11.

[0059] The classification result display area Ar12 displays SOAP categories, phrases associated with the categories, and information on the confidence level of the classification of the phrases into each SOAP category. By checking the content displayed in the classification result display area Ar12, the user can confirm the classification results of the text he or she entered into each SOAP category.

[0060] When the SOAP category display portion is selected by tapping, clicking, or other operations, a classification change instruction section Sc11 is displayed in the form of a speech bubble, as shown in FIG. 9. The classification change instruction section Sc11 displays buttons corresponding to each SOAP category and an "x" button. The user can change the category to which the phrase is classified by pressing one of the SOAP buttons. The "x" button is a button for inputting the instruction "cannot be classified."

[0061] A shaping button Bn2 and a transcription button Bn3 are located to the right of the text input area Ar11. The shaping button Bn2 is a button that instructs shaping of a sentence using the classified phrases displayed in the classification result display area Ar12. When the shaping button Bn2 is pressed, the sentence shaping unit 18 performs shaping of the sentence, and the shaped sentence is displayed in the sentence display area Ar13. By checking the sentence displayed in the sentence display area Ar13, the user can determine whether the content of the text they input is appropriately reflected in the sentence, or whether the generated sentence is suitable for transcription into a medical record.

[0062] The TRANSCRIPTION button Bn3 is a button for instructing that the results of classification displayed in the classification result display area Ar12 that have been classified into the SOAP category be directly transcribed to the medical record. An example of transcription to the medical record when the TRANSCRIPTION button Bn3 is pressed will be described in detail with reference to the following FIG.

[0063] The text displayed in the text display area Ar13 can be edited by the user. The transcribe button Bn4 located on the right side of the text display area Ar13 is a button for instructing transcription of the text displayed in the text display area Ar13 to the medical record. Therefore, the user can edit the text displayed in the text display area Ar13 and press the transcribe button Bn4 to transcribe the edited text formatted by the text formatting unit 18 to the medical record.

[0064] Next, an example of the display of a medical record after the transcribe button Bn3 or the transcribe button Bn4 is pressed on the medical record creation screen Sc1 will be described. Figure 10 is a diagram showing an example of the display of a medical record when text is displayed in association with each SOAP category. The medical record creation support screen Sc shown in Figure 10 is displayed when the transcribe button Bn3 is pressed on the medical record creation screen Sc1 shown in Figure 9. In the area surrounded by a bold frame in the medical record creation area Ar2 of the medical record creation support screen Sc, the classification results shown in the classification result display area Ar12 of the medical record creation screen Sc1 shown in Figure 9 are displayed in association with each SOAP category.

[0065] Therefore, according to the display example shown in Fig. 10, a user checking the contents of the medical record can properly understand the contents of the medical record while checking their correspondence with each SOAP category. If the medical record system (not shown) has input fields corresponding to each SOAP category, it is also possible to assign and transcribe the classification results shown in the classification result display area Ar12 of the medical record creation screen Sc1 shown in Fig. 9 to the input fields for each category. If the medical record system does not have such input fields, the classification results for each SOAP category can be transcribed as itemized lists, each on a single line.

[0066] 11 is a diagram showing an example of a display of a medical record when only text is displayed. According to the display example shown in FIG. 11, a user checking the contents of the medical record can easily understand the contents of the medical record by using natural text.

[0067] <Medical Record Creation Support Processing by Medical Record Creation Support System> Next, a description will be given of the medical record creation support processing by the medical record creation support system 100 according to this embodiment. FIG. 12 is a flowchart showing an example of the procedure of the medical record creation support processing by the medical record creation support system 100.

[0068] First, the input unit 22 of the terminal device 2 accepts text input from the user (step SA1). This text is information obtained from the patient through a medical examination or the like, and is input by a dentist or other such person without considering classification into SOAP or the like. The text input to the terminal device 2 in step S1 is transmitted to the server 1 via the communication I / F unit 24 (see FIG. 2).

[0069] Next, the text segmentation unit 15 of the server 1 segments the text transmitted from the terminal device 2 into phrase units (step SB1). The text segmentation unit 15 then inputs each segmented phrase to the phrase classification unit 16 (step SB2). The phrase classification unit 16 classifies each input segment into one of the SOAP categories (step SB3). After processing step SB3, the phrase classification unit 16 outputs the phrases and information on the confidence level of the phrase classification into a category. At this time, the phrase classification unit 16 outputs not only information on the category with the highest confidence level into which the phrase was classified, but also information on the categories with the second highest confidence level and below, and information on the confidence level of the classification into each category.

[0070] Next, the phrase classification unit 16 determines whether the phrase to be classified is the last phrase (step SB4). If it is determined in step SB4 that the phrase is not the last phrase (NO in step SB4), the phrase classification unit 16 changes the phrase to be classified into each SOAP category to the next phrase and returns to step SB2.

[0071] On the other hand, if it is determined in step SB4 that the phrase is the last phrase (YES in step SB4), the phrase classification unit 16 returns correspondence information between the phrase classification result and the classification certainty to the terminal device 2 via the communication I / F unit 14 (see FIG. 2) (step SB5).Then, the phrase selection unit 17 excludes phrases for which the difference between the minimum and maximum classification certainty values ​​is equal to or less than a predetermined value from the phrases to be shaped into text for transcription into the medical record (step SB6).

[0072] Meanwhile, the correspondence information between the phrase classification results and the classification certainty returned from the server 1 in step SB5 is received by the terminal device 2. Then, the display control unit 25 (see FIG. 2) of the terminal device 2 displays the correspondence information between the phrase classification results and the classification certainty on the medical record creation screen Sc1 (see FIG. 8) (step SA2). By performing the processing of step SA2, a user such as a dentist can confirm, via the display content of the medical record creation screen Sc1, into which SOAP category and with what degree of certainty each segment included in the text he or she created has been classified.

[0073] Next, the user determines whether the classification result is appropriate (step SA3). If the user determines in step SA3 that the classification result is inappropriate (NO in step SA3), the user corrects the classification. The user then transmits the corrected classification and phrase to the server 1 via the communication I / F unit 24 of the terminal device 2 (step SA4).

[0074] The phrase classification unit 16 of the server 1 inputs the phrases excluding the phrases that were not targeted for shaping into sentences in step SB6, the category into which the phrases were classified, and the prompt to the sentence shaping unit 18 (step SB7). Also, in step SB7, if the information on the corrected classification was transmitted from the terminal device 2 in step SA4, the phrase classification unit 16 inputs the information on the corrected classification to the sentence shaping unit 18.

[0075] Next, the sentence formatting unit 18 formats the sentence based on the phrase input in step SB7, the category into which the phrase was classified, and the prompt, and returns the formatted sentence to the terminal device 2 via the communication I / F unit 14 (step SB8).Then, the medical record transcription unit 26 of the terminal device 2 transcribes the formatted sentence returned from the server 1 into the medical record (step SA5).After processing of step SA5, the medical record creation support process by the medical record creation support system 100 ends.

[0076] In the medical record creation support system 100 according to the embodiment described above, text entered by a user such as a dentist is divided into phrases, and the divided phrases are appropriately classified into SOAP categories. Therefore, according to this embodiment, even a dentist who is unfamiliar with creating medical records and has little experience can appropriately create a medical record in accordance with the medical record entry format (SOAP format) based on the classification results.

[0077] Furthermore, in the medical record creation support system 100 according to the embodiment described above, text entered into the terminal device 2 is automatically classified into the SOAP category. Therefore, according to this embodiment, it is possible to prevent a dentist or other professional from being confused about how to classify text into SOAP, which would result in a prolonged time required to create a medical record.

[0078] In the above-described embodiment, the phrase classification unit 16 outputs, as the classification result of the phrases into SOAP categories, not only the category with the highest classification certainty, but also information on the category with the lowest classification certainty, along with the certainty. This information is then displayed on the medical record creation support screen Sc. Therefore, the user can check, via the medical record creation support screen Sc, whether the text they created properly contains phrases corresponding to each SOAP category, whether the SOAP category they assumed was correct, etc.

[0079] In other words, the medical record creation support system 100 according to this embodiment provides users with insight into how to create sentences that will improve the accuracy of classification into each SOAP category. Therefore, according to this embodiment, the medical record creation skills of users such as dentists who use the medical record creation support system 100 can be improved.

[0080] Furthermore, in the medical record creation support system 100 according to the embodiment described above, the classification results of the phrase classifier 16 into each SOAP category are first presented to the user via the medical record creation support screen Sc (see FIG. 9), and the user can change the category. Therefore, according to this embodiment, even if the classification by the phrase classifier 16 is incorrect, the user can appropriately correct the correspondence between the phrase and the SOAP category.

[0081] In the above-described embodiment, the phrase selection unit 17 excludes phrases for which the difference between the minimum and maximum values ​​of the confidence level for classification into each SOAP category is equal to or less than a predetermined value from the classification results to be output. Therefore, according to this embodiment, phrases that have little relevance to any SOAP category and are not relevant to the diagnosis can be prevented from being transcribed into the medical record.

[0082] In the above-described embodiment, the sentence formatting unit 18, which is configured using a large-scale language model, formats the sentences to be transcribed into the medical record. Therefore, according to this embodiment, a medical record written in natural sentences is generated.

[0083] In the above-described embodiment, an example has been given in which the sentence is shaped by the sentence shaping unit 18 after classification by the phrase classification unit 16, but the present invention is not limited to this. For example, a sentence shaped by the sentence shaping unit 18 may be input to the phrase classification unit 16.

[0084] The phrase classification unit 16 may also be configured using a large-scale language model. In this case, for example, by inputting a prompt such as "Please answer in one of the SOAP categories as the category that can be read from this phrase" to the phrase classification unit configured using the large-scale language model, the phrase can be classified into a SOAP category. Then, the sentence formatting unit 18 formats a sentence corresponding to the category after classification using the phrase classified into one of the SOAP categories.

[0085] By configuring the phrase classification unit 16 and the sentence formatting unit 18 in this manner, even if there is a plurality of pieces of information corresponding to each SOAP category in the text input by a user such as a dentist, or even if the text is not input in the order of S, O, A, P, it is possible to obtain as output each sentence corresponding to each SOAP category.

[0086] Furthermore, the medical record creation support system 100 according to the present invention may be configured not to include the sentence shaping unit 18. Alternatively, it may be configured so that the user can decide whether or not to perform sentence shaping by the sentence shaping unit 18.

[0087] In the above-described embodiment, the medical record is written in SOAP format, but the present invention is not limited to this. The medical record may be written in another format such as SOAPIE (Subject Object Assessment Plan Intervention Evaluation).

[0088] In the above embodiment, an example was given in which the text segmentation unit 15 and the phrase classification unit 16 were provided separately in the server 1, but the present invention is not limited to this. The phrase classification unit 16 may also perform text segmentation.

[0089] In the above-described embodiment, the content learned by the phrase classification unit 16 is used only for creating medical records, but the present invention is not limited to this. For example, the medical record creation support system 100 according to the present invention may generate an educational screen for creating medical records based on the content learned by the phrase classification unit 16 and display the screen on the display unit 23.

[0090] In addition, the above-mentioned embodiment provides a detailed and specific description of the configuration of the device (server, terminal device) and system (medical record creation support system) in order to clearly explain the present invention, and is not necessarily limited to having all of the configurations described.

[0091] 1 and 2, the control lines or information lines shown by solid lines are those considered necessary for explanation, and do not necessarily show all the control lines or information lines in the product. In reality, it can be considered that almost all components are interconnected.

[0092] In addition, in this specification, processing steps describing chronological processing include not only processing that is performed chronologically in the order described, but also processing that is not necessarily performed chronologically but is performed in parallel or individually (for example, parallel processing or processing by objects).

[0093] 1...server, 2...terminal device, 2-1...terminal device, 2-n...terminal device, 15...text segmentation unit, 16...phrase classification unit, 17...phrase selection unit, 18...sentence formatting unit, 25...display control unit, 26...medical record transcription unit, 100...medical record creation support system, Sc1...medical record creation screen

Claims

1. A medical record creation support device comprising: a text segmentation unit which divides free text including information obtained from a patient into phrases; and a phrase classification unit which classifies each of the phrases divided by the text segmentation unit into one of a plurality of categories based on the format of the medical record in accordance with the content indicated by the phrase, and outputs the classification result, wherein the phrase classification unit is configured by a learning model which has learned correspondence information between phrases included in the free text and the categories into which the phrases should be classified, and the phrase classification unit classifies the phrases divided by the text segmentation unit into all of the categories which may be expected as classification targets, and outputs the correspondence information between the phrases and the categories and information on the confidence of the classification into the category as the classification result.

2. The medical record creation support device according to claim 1, further comprising a phrase selection unit that excludes phrases that are difficult to classify into a specific category from the classification results.

3. The medical record creation support device of claim 2, wherein the phrase that is difficult to classify into the specific category is a phrase for which the difference between the minimum and maximum classification certainty values ​​when the phrase classification unit classifies into a plurality of categories is less than a predetermined value, or a phrase that cannot be classified into the specific category.

4. The medical record creation support device according to claim 3, further comprising a text formatting unit that formats text to be recorded in the medical record based on the classification results, and outputs the formatted text and information on the category into which each segment contained in the text is classified.

5. The medical record creation support device according to claim 4, wherein the sentence formatting unit is configured using a large-scale language model.

6. The medical record creation support device according to claim 5, further comprising a display control unit that causes a display device to display a screen that displays the classification results and enables correction of the correspondence between the phrases and the categories in the classification results.

7. The medical record creation support device according to claim 6, further comprising a medical record transcription unit which transcribes the classification results output from the phrase classification unit or the sentences output from the sentence formatting unit into a medical record.

8. A medical record creation support method performed by a medical record creation support device having a text division unit and a phrase classification unit, comprising: a text division step in which the text division unit divides free text including information obtained from a patient into phrases; and a phrase classification step in which the phrase classification unit classifies each of the phrases divided by the text division step into one of a plurality of categories based on the medical record entry format according to the content indicated by the phrase, and outputs a classification result, wherein the phrase classification unit is configured by a learning model that has learned correspondence information between phrases included in the free text and the categories into which the phrases should be classified, and in the phrase classification step, the phrase classification unit classifies the phrases divided by the text division unit into all of the categories that may be expected as classification targets, and outputs the correspondence information between the phrases and the categories and information on the confidence of the classification into the categories as the classification result.

Images