Medical support device, medical support method, and medical support program

Abstract

To provide a medical support device, a medical support method and a medical support program for reducing the labor of a medical worker who handles medical-related documents, and for automatically converting a medical document including technical terms into an expression which is easy to understand for a patient.SOLUTION: A medical support device includes a storage unit that stores a plurality of dictionaries including a language model for converting a technical term into a simple expression and a correspondence between a word registered in advance and the dictionary, a data acquisition unit that acquires text data of a created medical-related document, an extraction unit that extracts a technical term included in the text data with reference to the storage unit, a term conversion unit that converts the extracted technical term into a simple expression, and a document creation unit that inputs the text data converted into the simple expression to a large-scale language model and generates and outputs an explanatory document for a patient.SELECTED DRAWING: Figure 5

Description

[Technical Field]

[0001] The present invention relates to a medical support device, a medical support method, and a medical support program. [Background technology]

[0002] In recent years, the electronicization of documents such as medical records prepared by medical professionals and diagnostic imaging reports for radiological interpretation has been progressing. Such electronic medical records and reports are typically created by doctors or other professionals who manually input text using a keyboard or by selecting preset phrases. To assist in the creation of medical documents, a medical support device has been proposed that converts words uttered during a medical examination into appropriate words based on the scene and circumstances of the examination or examination during text conversion using voice recognition, and then displays the converted text on an electronic medical record or reporting device, with the aim of reducing the burden on doctors of manually inputting and memorizing medical information (see, for example, Patent Document 1). Medical documents such as diagnostic imaging reports are generally intended to communicate information between medical professionals, such as from one doctor to another, and therefore often contain technical terms and unique writing styles. Therefore, these documents are difficult for patients to understand in their current state, necessitating the creation of new documents to explain the situation to patients. When creating explanatory documents, rewriting them into text that is easier for patients to understand is extremely time-consuming and places a burden on doctors and other professionals. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2008-136646 Summary of the Invention [Problem to be solved by the invention]

[0004] Therefore, the main object of the present invention is to provide a medical support device, a medical support method, and a medical support program that can reduce the burden on medical professionals who handle medical-related documents and automatically convert medical documents containing technical terms into expressions that are easy for patients to understand. [Means for solving the problem]

[0005] The medical support device according to the present invention comprises: a storage means for storing a plurality of dictionaries including a language model for converting technical terms into simple expressions, and correspondences between pre-registered words and the dictionaries; a data acquisition means for acquiring text data of the created medical-related document; an extraction means for extracting technical terms contained in the text data by referring to the storage means; a term conversion means for converting the extracted technical terms into simple expressions; and The system is characterized by comprising a document creation means for inputting the sentence data converted into simplified expressions into a large-scale language model and generating and outputting an explanatory document for a patient.

[0006] The medical support method according to the present invention includes a storage step of storing a plurality of dictionaries including a language model for converting technical terms into simple expressions, and correspondences between pre-registered words and the dictionaries; a data acquisition step of acquiring text data of the created medical-related documents; an extraction step of extracting technical terms contained in the text data by referring to the information stored in the storage step; a term conversion step of converting the extracted technical terms into simple expressions; and The method is characterized by comprising a document creation step of inputting the sentence data converted into simplified expressions into a large-scale language model, and generating and outputting an explanatory document for the patient.

[0007] The medical support program according to the present invention includes a storage step of storing a plurality of dictionaries including a language model for converting technical terms into simple expressions, and a correspondence between pre-registered words and the dictionaries; a data acquisition step of acquiring text data of the created medical-related documents; an extraction step of extracting technical terms contained in the text data by referring to the information stored in the storage step; a term conversion step of converting the extracted technical terms into simple expressions; and The medical support program executes a document creation step in which the text data converted into simplified expressions is input into a large-scale language model, and an explanatory document for a patient is generated and output. [Effects of the Invention]

[0008] According to the present invention, it is possible to provide a medical support device, a medical support method, and a medical support program that can reduce the burden on medical professionals who handle medical-related documents and automatically convert medical documents containing technical terms into expressions that are easy for patients to understand. In particular, it is possible to easily create explanatory documents that are easy for patients to understand from medical documents that are difficult for patients to understand in their original state, thereby reducing the burden on medical professionals.

[0009] The above and other objects, features and advantages of the present invention will become more apparent from the following detailed description of exemplary embodiments, which proceeds with reference to the drawings. [Brief explanation of the drawings]

[0010] [Figure 1] 1 is a block diagram showing the configuration of a medical support system applied to this embodiment. [Figure 2] FIG. 2 is an illustrative view showing an example of the configuration of a table referred to as technical terminology. [Figure 3] FIG. 2 is an illustrative diagram showing an example of the structure of a DICOM file created by a modality. [Figure 4] FIG. 10 is an illustrative view showing an example of a top screen of a diagnostic support screen displayed on a main monitor of the medical support device; [Figure 5] FIG. 10 is an illustrative view showing an example of an image interpretation report screen of the diagnosis support screen displayed on the main monitor of the medical support device; [Figure 6] FIG. 10 is an illustrative view showing another example of the top screen of the diagnostic support screen displayed on the main monitor of the medical support device; [Figure 7] FIG. 4 is a flowchart showing part of the operation of a CPU provided in the medical support device. [Figure 8] FIG. 10 is a flowchart showing another part of the operation of the CPU provided in the medical support device. DETAILED DESCRIPTION OF THE INVENTION

[0011] 1, the medical support system of this embodiment is composed of a medical support device 10, a modality 20, a PACS 30, and a report database 40, all of which are connected via an in-hospital LAN 50. In this embodiment, a case will be described in which an image interpretation doctor (medical professional) performs image diagnosis, creates a report, and then creates an explanatory document for the patient based on the created report.

[0012] In the medical support device 10, a bus BS1 is connected to a communication I / F 12cm, a CPU 12pr, a keyboard / mouse 12km, a DRAM 12mm, an HDD 12hd, a main monitor (first monitor) 12m1, and a sub-monitor (second monitor) 12m2.

[0013] The sub-monitor 12m2 is attached to the main monitor 12m1. The radiologist (diagnostician) faces the main monitor 12m1 and the sub-monitor 12m2 and operates the keyboard / mouse 12km to diagnose images and create reports.

[0014] The report database 40 is configured, for example, by a technical terminology database. FIG. 2 shows an example of the configuration of a table referenced as a predetermined character string. In this table, input technical terms used by medical professionals are stored (registered) as predetermined character strings. Each stored (registered) predetermined character string is then associated with an output term for the patient. While this example shows an example of technical terms configuring a table, a table may be configured for each requesting department. Furthermore, a flag may be set in the technical terminology table for all departments so that the reports are classified by requesting department. The report database 40 is not limited to the configuration of this embodiment and may be configured on the HDD 12hd.

[0015] As shown in Figure 2, input terminology such as "pleural effusion," "right < left," "no obvious infiltrates," "ascites present," "atrophy and cysts in both kidneys," and "same as last time" are stored (registered) in the terminology table. The strings may be words, such as names of diseases or symptoms, or even sentences. The table also stores (registers) output terms for patients corresponding to these input terminology, such as "fluid around the lungs," "more on the left side," "no signs of inflammation or infection," "excess fluid in the abdomen," "kidneys slightly shrunk, and small fluid sacs (cysts) are present in the kidneys," and "this has been a condition for some time."

[0016] When creating a patient explanation report (described later), the specified character strings (technical terms) are extracted from medical documents. The extracted character strings are then converted into patient output terms according to the correspondence in Figure 2. In this way, technical terms such as "No obvious infiltrates were found" can be converted into expressions that are easy for patients to understand, such as "There are no signs of inflammation or infection."

[0017] The modality 20 is, for example, an MRI device that outputs a DICOM file containing multiple medical images representing axial cross sections of a patient's body. A technician operates the modality 20 based on requests from the attending physician. Images captured at desired positions are then created as medical images, and the created medical images are stored in a DICOM file. The DICOM file created in this way has the data structure shown in Figure 3 and is stored in the PACS 30 via the hospital LAN 50.

[0018] 3, a DICOM file contains multiple medical images as well as additional information, which consists of multiple items that each indicate an examination attribute (examination date, patient name, patient age at the time of examination, accession number, etc.).

[0019] The CPU 12pr provided in the medical support device 10 executes the support processing shown in Figures 7 and 8 to support the creation of an explanatory document for a patient based on a report prepared by a radiologist by referring to the report database 40. The program corresponding to this flowchart is stored in the HDD 12hd as a medical support program that supports document creation.

[0020] Referring to Fig. 7, in step S1, the top screen is loaded into the DRAM 12m. The loaded top screen is read out by the main monitor 12m1 and displayed as shown in Fig. 4. According to Fig. 4, the top screen has a header field IDX in which a plurality of headers representing a plurality of examinations are listed. Also, in the top screen shown in Fig. 4, the radiology reports corresponding to all of the headers are also set to "pending."

[0021] In step S3, it is determined whether a heading selection operation (= an operation to select one of the multiple headings listed in the heading column IDX) has been performed using the keyboard / mouse 12km. In addition, in step S5, it is determined whether another operation has been performed using the keyboard / mouse 12km. If both the determination results of step S3 and step S5 are NO, the process returns to step S3; if the determination result of step S3 is YES, the process proceeds to step S7; if the determination result of step S5 is YES, the process proceeds to other processing.

[0022] In step S7, the image interpretation report screen corresponding to the heading selected by the heading selection operation is written to the DRAM 12 mm. The top screen on the DRAM 12 mm is updated with the image interpretation report screen.

[0023] The updated image interpretation report screen is read out by the main monitor 12m1 and displayed as shown in Fig. 5. According to Fig. 5, the image interpretation report screen has a findings column 501 and a patient summary column 502.

[0024] In step S9, the PACS 30 is accessed via the communication I / F 12cm, and a DICOM file corresponding to the accession number assigned to the selected heading is obtained from the PACS 30. In step S9, one of the single or multiple medical images stored in the obtained DICOM file is loaded into the DRAM 12mm. The loaded medical image is read out and displayed on the sub-monitor 12m2.

[0025] In step S11, it is determined whether or not a finding input operation (an operation of specifying the finding field 501 and inputting a desired character string) has been performed using the keyboard / mouse 12km. If the determination result is NO, the process proceeds directly to step S15.

[0026] On the other hand, if the determination result is YES, the process proceeds to step S13, where the character string (text) acquired by the finding input operation is displayed in the finding field 501 as shown in Fig. 5. According to Fig. 5, the finding field 501 displays the following character string: "Bilateral pleural effusion (right < left) has appeared. As a result, the volume of the left lower lobe has decreased, resulting in subpleural lung collapse. No clear infiltrates have been noted in the lung fields where air is retained. Small intestinal dilation and intramural gas are observed. Increased mesenteric fat concentration is also observed nearby. In addition, gas images are observed in the mesenteric vein and intrahepatic portal vein. Furthermore, compared with the previous CT scan, both the superior mesenteric artery and superior mesenteric vein have narrowed. The abnormal findings strongly suggest non-occlusive intestinal ischemia (NOMI). The inferior vena cava is generally flattened, suggesting dehydration. Ascites has appeared. Gas images have appeared subcutaneously in the abdominal wall. A relationship with the above pathology is suggested, but the details are unknown." When the display of the character string is completed, the process proceeds to step S15.

[0027] In step S15 (FIG. 8), it is determined whether the "Create Patient Summary" button (shown by a bold circle in FIG. 5) displayed in the upper right corner of the findings field 501 has been clicked with the keyboard / mouse 12km. If the determination result is NO, the process proceeds directly to step S27.

[0028] If the determination result in step S15 is YES, the process proceeds to a step of creating a patient information document. The patient information document is created by referring to the report database 40. In step S17, the report database 40 is referenced to extract character strings (input technical terms) stored (registered) in the table of FIG. 2 from the findings (text data) entered in the findings column 501.

[0029] In step S19, the character string extracted in step S17 is changed to the output character string in the table of FIG. 2, and the process proceeds to step S21 (terminology rewriting preprocessing).

[0030] In step S21, the character string (sentence data) converted into a simplified representation in step S19 is input into a large-scale language model (LLM), an appropriate prompt is created, and the sentence is converted for the patient, and the process proceeds to step S23. Various generative AIs, such as ChatGPT (registered trademark) and rinna (registered trademark), can be used as the large-scale language model. It is also preferable to fine-tune the LLM for each clinical department. In the present invention, sentences may be converted using a publicly available trained LLM, or optimization may be achieved by training the LLM.

[0031] In step S23, the sentence converted in step S21 is converted to polite language, and the process proceeds to step S25. In step S25, the data of the sentence converted in step S23 is loaded into the DRAM 12m1. The loaded data is read by the main monitor 12m1 and is added to and displayed in the patient summary text field 502 on the radiology report screen as shown in FIG.

[0032] After the display process is complete, the medical professional can confirm and correct the final text. Specifically, the medical professional clicks the "Edit Text" button (shown by a bold circle in FIG. 5) displayed in the upper right corner of the patient summary text field 502 using the keyboard / mouse 12k to perform the patient summary text editing operation (specifying the patient summary text field 502 and inputting / deleting the desired character string). The edited text may be fed back to the large-scale language model and report database 40.

[0033] In step S27, it is determined whether or not the "Confirm" button (shown by the thick circle in the upper right corner in FIG. 5) displayed in the upper right corner of the image interpretation report screen has been clicked by the keyboard / mouse 12km. In addition, in step S29, it is determined whether or not another operation has been performed by the keyboard / mouse 12km. If both the determination results of step S27 and step S29 are NO, the process returns to step S27; if the determination result of step S27 is YES, the process proceeds to step S31; if the determination result of step S29 is YES, the process proceeds to other processing.

[0034] In step S31, a confirmation process is performed to confirm the input contents. Once the confirmation process is performed, the process returns to step S1. As a result, the display on the main monitor 12m1 returns from the radiology report screen to the top screen (see FIG. 6). However, as shown by the bold circle in FIG. 6, the heading corresponding to the radiology report that has been confirmed has the word "confirmed" added instead of "pending."

[0035] In the example shown in this embodiment, the following string is created and displayed in the patient summary document field 502: "There is a lot of fluid around the left lung, preventing it from expanding normally. However, there are no signs of inflammation or infection in the lung itself. The small intestine is abnormally swollen, which may be causing poor blood flow. This suggests that there is insufficient blood flow to the small intestine. The body is dehydrated, and excess fluid is accumulating in the abdomen. There is gas under the abdominal wall, but it is unclear how this is related to other medical conditions. There is abnormal fat accumulation in the liver, but there are no major problems with the other internal organs. The kidneys are slightly shrunk, and there are small fluid sacs (cysts) in the kidneys, but this is a pre-existing condition. In short, there are multiple health problems, including fluid accumulation in the lungs, insufficient blood flow to the small intestine, dehydration, ascites, gas under the abdominal wall, fat accumulation in the liver, and kidney problems."

[0036] As can be seen from the above description, the medical support device 10 is a device that supports the creation of patient information documents, and the following processing is executed by the CPU 12pr that constitutes the device.

[0037] First, at least one medical image is displayed on the sub-monitor 12m2 facing the radiologist (S9). The radiologist's opinion (comments on findings) is acquired via the keyboard / mouse 12km and displayed on the radiology report screen on the main monitor 12m1 (S11 to S13).

[0038] Then, an explanatory document for the patient corresponding to the opinion (finding comment) is automatically generated, and the generated (converted) text data is expanded into DRAM 12mm and added to the patient summary document field 502 on the radiology report screen of the main monitor 12m1 (S15 to S25).

[0039] According to the present invention, it is possible to automatically convert text data of medical-related documents such as radiology reports, which often contain a large amount of technical terminology, into expressions that are easy for patients to understand. By easily creating explanatory documents that are rewritten from medical documents that are difficult for patients to understand in their original state, it is possible to reduce the burden on medical professionals.

[0040] Although the above description has been given using a radiology report as an example of a medical-related document, the present invention is not limited to this and can be applied to various other documents. For example, the present invention can also be applied to medication instructions, treatment plan explanations, etc. In these cases, medication-related character strings such as the name of the drug and the time of administration, as well as the name of the treatment, can be stored as technical terms. Furthermore, in these instructions, a database can be constructed so that, for example, the drug name is displayed in a different color when output. [Explanation of symbols]

[0041] 10...Medical support equipment 12pr...CPU 12km...Keyboard / Mouse 12hd...HDD 12m1 ...Main monitor 12m2 ... Sub-monitor 40...Reporting database

Claims

1. a storage means for storing a plurality of dictionaries including a language model for converting technical terms into simple expressions, and correspondences between pre-registered words and the dictionaries; a data acquisition means for acquiring text data of the medical-related documents created; an extraction means for extracting technical terms contained in the text data by referring to the storage means; a term conversion means for converting the extracted technical terms into simple expressions; and A medical support device characterized by comprising a document creation means for inputting the sentence data converted into simplified expressions into a large-scale language model and generating and outputting an explanatory document for a patient.

2. a storage step of storing a plurality of dictionaries including a language model for converting technical terms into simple expressions and correspondences between pre-registered words and the dictionaries; a data acquisition step of acquiring text data of the created medical-related documents; an extraction step of extracting technical terms contained in the text data by referring to the information stored in the storage step; a term conversion step of converting the extracted technical terms into simple expressions; and A medical support method characterized by comprising a document creation step of inputting the sentence data converted into simplified expressions into a large-scale language model and generating and outputting an explanatory document for a patient.

3. a storage step of storing a plurality of dictionaries including a language model for converting technical terms into simple expressions and correspondences between pre-registered words and the dictionaries; a data acquisition step of acquiring text data of the created medical-related documents; an extraction step of extracting technical terms contained in the text data by referring to the information stored in the storage step; a term conversion step of converting the extracted technical terms into simple expressions; and A medical support program for executing a document creation step of inputting the sentence data converted into simplified expressions into a large-scale language model and generating and outputting an explanatory document for a patient.

Images