Medical record summary report generation device using large language model, and medical record summary report generation method using same

Abstract

A medical record summary report generation device according to an embodiment of the present invention comprises: a summary format input unit for receiving a summary format of a medical record summary report; a medical record input unit for receiving various pieces of information about a medical record of a patient; a pre-processing unit connected to the medical record input unit; a large language model (LLM) unit connected to the summary format input unit; a control unit connected to the pre-processing unit and the LLM unit; and an output unit connected to the control unit.

Description

A device for writing a medical record summary report using a large-scale language model and a method for writing a medical record summary report using the same.

[0001] The present invention relates to a medical record summary report writing device using a large-scale language model and a medical record summary report writing method using the same, and more particularly, to a medical record summary report writing device that summarizes and reports various medical records of a patient using a large-scale language model of artificial intelligence and a medical record summary report writing method using the same.

[0002] In healthcare settings, various medical records exist, including a patient's past medical history, current condition, prescribed medications, and test results. Gathering all this information takes considerable time, and quickly gathering all the necessary information, especially during or before treatment, places a significant burden on medical staff.

[0003] As a solution to this, the need for a system that utilizes artificial intelligence to provide users with reports in the format they desire in advance, enabling medical staff to quickly assess a patient's condition is emerging.

[0004] The present invention has been devised to solve the above problems, and the purpose of the present invention is to provide a medical record summary report writing device that summarizes and reports various medical records of a patient using artificial intelligence LLM.

[0005] Another object of the present invention is to provide a method for preparing a medical record summary report by summarizing and reporting various medical records of a patient using a medical record summary report preparation device.

[0006] A medical record summary report writing device according to one embodiment of the present invention includes a summary format input unit for receiving a summary format of a medical record summary report; a medical record input unit for receiving various information regarding a patient's medical records; a preprocessing unit connected to the medical record input unit; an LLM (large language model) unit connected to the summary format input unit; a control unit connected to the preprocessing unit and the LLM unit; and an output unit connected to the control unit.

[0007] In a medical record summary report writing device according to one embodiment of the present invention, the preprocessing unit is characterized in that it refines the medical record information of the patient received using a template and extracts key information.

[0008] In a medical record summary report writing device according to one embodiment of the present invention, the LLM unit is characterized in that it summarizes the main information according to the summary format of the medical record summary report.

[0009] In addition, a method for creating a medical record summary report according to another embodiment of the present invention includes: (A) a step in which a control unit receives a summary format of a medical record summary report desired by a user through a summary format input unit, and receives information about a patient's medical record through a medical record input unit; (B) a step in which a preprocessing unit performs a preprocessing of a purification process to extract key information from the medical record information received from the medical record input unit; (C) a step in which the control unit transmits the purified medical record information to an LLM unit, and the LLM unit summarizes the transmitted purified medical record information according to the summary format of the medical record summary report; and (D) a step in which the control unit displays the received medical record summary report through an output unit.

[0010] In a method for preparing a medical record summary report according to another embodiment of the present invention, step (B) is characterized in that the preprocessing unit refines the patient's medical record information received using a template.

[0011] In a method for preparing a medical record summary report according to another embodiment of the present invention, step (C) is characterized in that the patient's main information is extracted from the refined medical record information using a natural language processing (NLP) technique for processing LLM additional natural language data and a large-scale language model.

[0012]

[0013] The features and advantages of the present invention will become more apparent from the following detailed description based on the attached drawings.

[0014] Prior to this, the terms and words used in this specification and claims should not be interpreted in their usual or dictionary meanings, but should be interpreted in the sense and concept that is consistent with the technical idea of ​​the present invention, based on the principle that the inventor can appropriately define the concept of the term to explain his or her own invention in the best way.

[0015] The medical record summary report writing device according to an embodiment of the present invention uses LLM to refine and process information on various medical records of a patient, summarize it, and output it as a medical record summary report, thereby enabling medical staff to quickly understand the patient's condition.

[0016] Figure 1 is a configuration diagram of a medical record summary report writing device according to one embodiment of the present invention.

[0017] FIG. 2 is an exemplary diagram showing the structure of a transformer constituting a medical record summary report writing device according to one embodiment of the present invention.

[0018] Figure 3 is a flowchart illustrating a method for creating a medical record summary report according to another embodiment of the present invention.

[0019]

[0020] The purpose, specific advantages, and novel features of the present invention will become more apparent from the following detailed description and preferred embodiments in conjunction with the accompanying drawings. In this specification, when assigning reference numbers to components in each drawing, it should be noted that, as much as possible, identical components are given the same numbers even if they are shown in different drawings. In addition, while terms such as first, second, etc. may be used to describe various components, the components should not be limited by the terms. The terms are used only for the purpose of distinguishing one component from another. In addition, in describing the present invention, if a detailed description of a related known technology is judged to unnecessarily obscure the gist of the present invention, the detailed description thereof will be omitted.

[0021] Hereinafter, a preferred embodiment of the present invention will be described in detail with reference to the attached drawings. Fig. 1 is a configuration diagram of a medical record summary report creation device according to an embodiment of the present invention, and Fig. 2 is an exemplary diagram showing the structure of a transformer constituting a medical record summary report creation device according to an embodiment of the present invention.

[0022] A medical record summary report writing device according to one embodiment of the present invention includes, as illustrated in FIG. 1, a summary format input unit (110), a medical record input unit (120), a preprocessing unit (130) connected to the medical record input unit (120), an LLM unit (150) connected to the summary format input unit (110), a control unit (140) connected to the preprocessing unit (130) and the LLM unit (150), and an output unit (160) connected to the control unit (140).

[0023] The summary format input unit (110) is a module that receives the summary format of the medical record summary report desired by the user through the user interface (UI), and processes the input of the document format of the medical record summary report desired by the user in Hangul, MS Word, Excel, etc. The user can write and input the desired medical record summary report format or upload an existing medical record summary report format. The medical record summary report format can be selected from a summary of key information of a specific patient, a summary of the entire medical record, etc., and the medical record summary report format entered in this way is processed by the summary format input unit (110) and transmitted to the LLM unit (150).

[0024] The medical record input unit (120) is a module that receives various information regarding a patient's medical records by being connected to an external hospital information system, medical database, document management system, etc., either wired or wirelessly. This medical record input unit (120) is connected to an external hospital information system, medical database, document management system, etc., by being connected wired or wirelessly, and receives and processes the patient's personal information, disease name, medical record, examination information, prescription drug information, etc., and transmits them to the preprocessing unit (130).

[0025] The preprocessing unit (130) is a module that extracts key information from the medical record information received from the medical record input unit (120) and checks for missing information. It refines the received medical record information using a template, extracts key information including the patient's personal information, disease name, medical record, examination information, prescription drug information, etc., and determines whether there is missing medical record information and transmits the determined information to the control unit (140). At this time, the preprocessing unit (130) may be optionally connected to the LLM unit (150) to analyze text from the medical record information using a natural language processing (NLP) technique or a large-scale language model and structure necessary information.

[0026] The control unit (140) is a module that controls the overall medical record summary report writing device, and one side is connected to the preprocessing unit (130) and the other side is linked to the LLM (large language model) unit (150) via an API (application programming interface). This control unit (140) converts preprocessed data transmitted from the preprocessing unit (130) into refined data and transmits it to the LLM unit (150) for processing.

[0027] The LLM (large language model) unit (150) is a module that includes an artificial neural network with numerous parameters (usually several billion weights or more) and a large-scale language model applied thereto, and can learn and train a significant amount of unlabeled text using self-supervised learning or semi-self-supervised learning. This LLM unit (150) processes a massive amount of natural language data, processes and summarizes medical record information transmitted from the control unit (140), and creates and generates a summary report according to the summary format transmitted from the summary format input unit (110). At this time, the LLM unit (150) can learn about a massive amount of text data of medical record information transmitted from the control unit (140) using deep learning technology and derive a medical record summary report according to a user's request. Examples of such LLMs include OpenAI's GPT (Generative Pre-trained Transformer) series and Google's BERT (Bidirectional Encoder Representations from Transformers) models. These models are used in various applications including language translation, content generation, and chatbots.

[0028] In particular, the LLM unit (150) may be configured to be linked to an external large-scale language model, or may be configured to include a transformer that generates a large-scale language model as illustrated in FIG. 2. At this time, the transformer is configured with an encoder (151) and a decoder (152), and inputs each refined medical record sentence into the input embedding layer of the encoder (151), and performs subsequent extraction of key information by assigning vector weights to each word in the attention layer, and summarizes the key information through the decoder (152), and creates and generates a summary report according to the summary format transmitted from the summary format input unit (110).

[0029] The output unit (160) includes a display device connected to the control unit (140) and can output a medical record summary report transmitted from the LLM unit (150) under the control of the control unit (140).

[0030] The medical record summary report writing device according to one embodiment of the present invention configured as described above uses LLM to refine and process information on various medical records of a patient, summarize it, and output it as a medical record summary report, so that medical staff can quickly understand the patient's condition.

[0031]

[0032] Hereinafter, a method for creating a medical record summary report according to another embodiment of the present invention will be described with reference to FIG. 3. FIG. 3 is a flowchart for explaining a method for creating a medical record summary report according to another embodiment of the present invention.

[0033] In a method for creating a medical record summary report according to another embodiment of the present invention, the control unit (140) first receives a summary format of a medical record summary report desired by the user through the summary format input unit (110), and receives various information (10) regarding the patient's medical records through the medical record input unit (120) (S310).

[0034] Specifically, the control unit (140) receives the summary format of the medical record summary report desired by the user through the summary format input unit (110), and at this time, receives and processes the summary format of the medical record summary report desired by the user in a document format such as Hangul, MS Word, Excel, etc. The user can write and input the desired medical record summary report format or upload an existing medical record summary report format. The format of the medical record summary report can be selected from a summary of key information of a specific patient, a summary of the entire medical record, etc., and the format of the medical record summary report input in this way is processed by the summary format input unit (110) and transmitted to the LLM unit (150).

[0035] Meanwhile, the control unit (140) receives various medical record information (10) of the patient from an external hospital information system, medical database, document management system, etc. connected wired or wirelessly through the medical record input unit (120). Here, the medical record information (10) may include the patient's personal information, disease name, medical record, CT, MRI examination information, prescription drug information, etc.

[0036] Thereafter, under the control of the control unit (140), the preprocessing unit (130) extracts key information from the medical record information (10) received from the medical record input unit (120) and performs preprocessing of a purification process to check for missing information (S320).

[0037] At this time, in order to perform preprocessing of the purification process, the preprocessing unit (130) refines the received medical record information using a template to extract key information including the patient's personal information, disease name, medical record, examination information, prescription information, etc., and determines whether there is any missing medical record information and transmits the determined information to the control unit (140). Here, the template is information in various example formats corresponding to each of the received medical record information, and the preprocessing unit (130) can compare the received medical record information corresponding to each of these example formats, extract key information, determine whether there is any missing medical record information, and transmit the determined information to the control unit (140).

[0038] At this time, the preprocessing unit (130) may be optionally connected to the LLM unit (150) to analyze text from medical record information and structure necessary information using natural language processing (NLP) techniques or large-scale language models.

[0039] After performing the preprocessing of the purification process, the control unit (140) transmits the purified medical record information (20) to the LLM unit (150), and the LLM unit (150) summarizes the transmitted purified medical record information (20) (S330).

[0040] That is, the LLM unit (150) can extract and summarize key information, including the patient's personal information, disease name, medical record, examination information, and prescription drug information, from the refined medical record information (20) transmitted from the control unit (140) using a natural language processing (NLP) technique and a large-scale language model that process a large amount of natural language data.

[0041] At this time, the information summarized by the LLM department (150) from the refined medical record information (20) is in the form of key information including the patient's personal information, disease name, medical record, examination information, prescription drug information, etc., as shown below.

[0042]

[0043] After summarizing the refined medical record information (20), the LLM unit (150) creates a summary report according to the summary format transmitted from the summary format input unit (110). At this time, the LLM unit (150) uses deep learning technology to learn about the massive amount of text data of the medical record information transmitted from the control unit (140), and can derive a medical record summary report according to a user's request.

[0044] The control unit (140) that receives the medical record summary report generated in this way displays the medical record summary report (30) received through the output unit (160) (S340).

[0045] Of course, the control unit (140) can display the medical record summary report (30) received through the output unit (160) and simultaneously output the medical record summary report (30) to a printer.

[0046] Accordingly, a method for creating a medical record summary report according to another embodiment of the present invention uses LLM to refine and summarize information on various medical records of a patient and output it as a medical record summary report, so that medical staff can quickly understand the patient's condition.

[0047]

[0048] Although the technical idea of ​​the present invention has been specifically described in accordance with the above preferred embodiments, it should be noted that the above-described embodiments are for illustrative purposes only and are not intended to limit the invention.

[0049] In addition, a person skilled in the art will understand that various implementations are possible within the scope of the technical idea of ​​the present invention.

Claims

1. Summary format input section for entering the summary format of the medical record summary report; A medical record entry unit that receives various information regarding a patient's medical records; A preprocessing unit connected to the above medical record input unit; LLM (large language model) part connected to the above summary format input part; a control unit connected to the above preprocessing unit and the LLM unit; and An output unit connected to the above control unit; A device for creating a summary report of medical records including:

2. In paragraph 1, A medical record summary report writing device characterized in that the above preprocessing unit extracts key information by refining the medical record information of the patient received using a template.

3. In paragraph 2, A medical record summary report writing device characterized in that the above LLM department summarizes the above key information in the summary format of the medical record summary report. 4.(A) A step in which the control unit receives a summary format of a medical record summary report desired by the user through a summary format input unit, and receives information about the patient's medical record through a medical record input unit; (B) A step in which the preprocessing unit performs a preprocessing process for refining the medical record information received from the medical record input unit to extract key information; (C) The control unit transmits the refined medical record information to the LLM unit, and the LLM unit summarizes the transmitted refined medical record information according to the summary format of the medical record summary report; and (D) a step of the control unit displaying the summary report of the medical records received through the output unit; How to prepare a medical records summary report including:

5. In paragraph 4, A method for creating a medical record summary report, characterized in that the above step (B) refines the medical record information of the patient received by the preprocessing unit using a template.

6. In paragraph 4, The above step (C) is a method for creating a medical record summary report, characterized in that it extracts key information of a patient from the refined medical record information using a natural language processing (NLP) technique and a large-scale language model that processes LLM additional natural language data.

Images