System for streamlining the process of creating research findings

Abstract

This enables the provision of AI-powered systems, methods, and programs to streamline the process of handling clinical test requests and sharing pathology findings among multiple parties. [Solution] According to the present invention, by comparing the first report written by a medical professional with the second report written based on image data corresponding to the first report, it is possible to automatically verify whether the first report is correctly related to the requested examination and whether there are any typographical errors, omissions, or errors in the diagnostic content of the report, thereby providing a report writing efficiency system that streamlines the work related to reports written by medical professionals.

Description

Technical Field

[0001] The present invention relates to a finding work efficiency improvement system, a finding work efficiency improvement method, and a program.

Background Art

[0002] Clinical examination refers to evaluating the patient's condition by collecting specimens such as urine, blood, and tissue to evaluate the patient's health status, or directly examining the body functions through electroencephalogram, electrocardiogram, etc. Among them, pathological examination is an important examination used for definitive diagnosis. The pathological examination is carried out in the procedure that a doctor collects a specimen, a clinical laboratory technician or the like prepares a specimen, and a pathologist observes the specimen with a microscope and makes a diagnosis.

[0003] In addition, since only the collection of specimens in the process of pathological examination performed at an external examination facility such as a clinical examination center corresponds to a medical act, there are institutions in medical institutions that send the collected specimens to the clinical examination center and entrust clinical examinations after collecting the specimens.

[0004] Thus, in the process of clinical examination, exchanges of specimens, specimens, accompanying documents, etc. are carried out among multiple departments, facilities, and doctors. However, since clinical examination is carried out in a medical setting and causes a great impact by simply taking time and labor, it can be said that improving the efficiency of the working time of each person involved is important. Such procedures occur not only in clinical examinations but also in all scenes involving medical staff.

[0005] As a technique related to the efficient generation of findings in such clinical examinations and the like, Patent Document 1 discloses a method that makes it easy to grasp the output layout while editing a medical report or the like when creating it.

Prior Art Documents

Patent Documents

[0006]

Patent Document 1

[0007] Human errors are more likely to occur in work that spans departments and facilities, and discrepancies between text information submitted by doctors and diagnostic results, as well as spelling mistakes, typos, omissions, and diagnostic errors in findings, hinder efficiency. However, Patent Document 1 does not disclose how to exchange findings documents between multiple locations and multiple doctors.

[0008] Therefore, the present invention aims to provide a system, method, and program that utilizes artificial intelligence to improve the efficiency of processing findings in the workflow when sharing clinical test requests and pathological diagnosis findings among multiple parties. [Means for solving the problem]

[0009] The present invention relates to a system for improving the efficiency of observation work used between a first terminal and a second terminal, and is characterized by comprising: an acquisition unit that acquires a first observation statement and image data corresponding to the first observation statement from the first terminal; an observation generation unit that generates a second observation statement based on source data including at least the acquired image data; a similarity calculation unit that calculates the similarity between the first observation statement and the second observation statement; and an output unit that outputs the first observation statement as an output observation statement when the similarity between the first observation statement and the second observation statement calculated by the similarity calculation unit exceeds a predetermined threshold.

[0010] Furthermore, the present invention further comprises an editing unit that accepts editing of the first observation statement or the second observation statement when the similarity between the first observation statement and the second observation statement calculated by the similarity calculation unit does not exceed a predetermined threshold, and the output unit preferably outputs the observation statement edited by the editing unit as an output observation statement when the editing unit has functioned.

[0011] Furthermore, it is preferable that the present invention further comprises an integration unit that integrates and stores the output observation text output by the output unit and at least the image data.

[0012] Furthermore, it is preferable that the present invention further includes a reward determination unit that determines whether or not the output observation statement has been accepted by the user, and a library that stores the output observation statement and the image data stored in the integration unit when it is determined that the output observation statement has earned a reward.

[0013] Although this invention falls under the category of computer systems, it exhibits similar functions and effects in other categories such as electronic signature methods and programs, depending on the category. [Effects of the Invention]

[0014] According to the present invention, by calculating the similarity between a first report obtained by a medical professional and a second report generated by a report generation unit, it is possible to confirm whether the first report is correctly related to the requested examination, and to point out typographical errors, omissions, or errors in the diagnostic content of the report, thereby streamlining the work related to the report of medical professionals. [Brief explanation of the drawing]

[0015] [Figure 1] This is a schematic diagram of System 0 in the present invention. [Figure 2] This is an overall configuration diagram of System 0 of the first embodiment of the present invention. [Figure 3] This figure shows a flowchart of the process performed by System 0 of the first embodiment of the present invention. [Figure 4] This figure shows an example of an LLM4 model included in the findings generation module 21 of the present invention. [Figure 5] This is a schematic diagram illustrating the selection of LLM4, which is provided by the findings generation module 21 in the present invention. [Figure 6] This is an example of a second observation statement generated by the observation generation module 21 in the present invention. [Figure 7] It is a schematic diagram showing the comparison of text vectors performed in the similarity calculation module 22 in the present invention. [Figure 8] It is an overall configuration diagram of the system 0 according to the second embodiment in the present invention. [Figure 9] It is a diagram showing a flowchart of the process executed by the system 0 according to the second embodiment in the present invention.

Mode for Carrying Out the Invention

[0016] The present invention will be specifically described using preferred embodiments. However, the following embodiments are merely examples of the present invention, and the effects of the present invention are not limited to those described in the embodiments of the present invention.

[0017] First, based on FIG. 1, the outline of the finding work efficiency improvement system 0 in the present invention will be described. FIG. 1 is an outline diagram of the system 0 in the present invention. The system 0 is used when exchanging finding texts among a plurality of users, and is composed of a first terminal 1 used by a first user and a second terminal 2 used by a second user. In addition, the finding text in the present invention refers to a document showing an opinion in which medical professionals such as doctors have judged a disease or the state of the body from a diagnosis result, a medical report, etc. based on examination and test results. In addition, the first user and the second user are assumed to be medical professionals such as clinical laboratory technicians, pathologists, clinicians, and nurses belonging to an institution where pathological specimens are transmitted and received, such as a medical institution or a testing center. As a typical example, a clinician examining a patient is assumed to be the first user, and a pathologist performing a pathological diagnosis upon receiving a request from the clinician is assumed to be the second user. In addition, the system 0 may include other terminals and devices other than the first terminal 1 and the second terminal 2, and the number, type, and function thereof are not particularly limited and can be appropriately designed.

[0018] The first terminal 1 and the second terminal 2 may be implemented as, for example, mobile devices such as smartphones and tablet devices, or computers such as desktop PCs and laptop computers. Furthermore, the first terminal 1 and the second terminal 2 may be implemented as, for example, a single computer, or as multiple computers, such as in a cloud computing environment. In this specification, a cloud computer refers to a system that scalably utilizes any computer to perform a specific function, or a system that includes multiple functional modules to realize a certain system, possesses those functions, and includes a communication unit, such as a device that enables communication with other terminals and devices. In this case, system 0 will execute each of the processes described later using one or more combinations of the first terminal 1, the second terminal 2, and other included terminals and devices. Furthermore, the first terminal 1 and the second terminal 2 may be implemented using the same device.

[0019] Necessary data and information are transmitted and received between the first terminal 1 and the second terminal 2, or between the second terminal 2 and other terminals or devices, via information communication lines such as 4G / LTE and WiFi, or public networks (referred to as network 4 in the diagram). This communication may be wired or wireless.

[0020] Next, the system configuration of System 0 in the first embodiment, which is a preferred embodiment of the present invention, will be described based on Figure 2. Figure 2 is an overall configuration diagram of System 0 in the first embodiment of the present invention.

[0021] The first terminal 1 and the second terminal 2 are equipped with a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), RAM (Random Access Memory), ROM (Read Only Memory), etc., as control units. Furthermore, the first terminal 1 and the second terminal 2 are equipped with data storage as memory units, such as hard disks, semiconductor memory, recording media, and memory cards. The data may be stored in cloud services or databases. Furthermore, the first terminal 1 and the second terminal 2 are equipped with devices as communication units that enable communication with other terminals and equipment. The communication method may be wireless or wired. Furthermore, the first terminal 1 and the second terminal 2 shall be equipped with the necessary functions as input units to enable operation of the first terminal 1 and the second terminal 2. Examples include a liquid crystal display for touch panel functionality, a keyboard, a mouse, hardware buttons on the device, and a microphone for voice recognition. Furthermore, the first terminal 1 and the second terminal 2 shall be equipped with the necessary functions as output units to enable the operation of the first terminal 1 and the second terminal 2. Examples include display and audio output such as projection to an LCD display, PC display, or projector.

[0022] The second terminal 2, by having the control unit read a predetermined program, works in cooperation with the communication unit to realize the acquisition module 20, the findings generation module 21, the similarity calculation module 22, the output module 23, and the editing module 24. Furthermore, these modules may be implemented by terminals on the cloud, and if the second terminal 2 is composed of multiple computers, including a cloud computer, some modules may be implemented on the cloud computer. Furthermore, in the second terminal 2, it is not necessary to implement all modules at once; system 0 may be used with only specific modules implemented.

[0023] The overview of the processes performed by System 0 will be explained based on Figure 3. Figure 3 is a flowchart showing the processes performed by System 0 in the first embodiment of the present invention. As shown in Figure 3, the system 0 of the first embodiment consists of steps S1 to S9 and includes an acquisition module 20 (step S1) that acquires a first observation statement and image data corresponding to the first observation statement from a first terminal; an observation generation module 21 (steps S2 to S4) that generates a second observation statement based on source data that includes at least the acquired image data; a similarity calculation module 22 (steps S5 to S8) that calculates the similarity between the first observation statement and the second observation statement; an output module 23 (steps S9-1, S10) that outputs the first observation statement when the similarity between the first observation statement and the second observation statement calculated by the similarity calculation module 22 exceeds a predetermined threshold; and an editing module 24 (step S9-2) that accepts editing of the first observation statement or the second observation statement when the similarity between the first observation statement and the second observation statement calculated by the similarity calculation module 22 does not exceed a predetermined threshold.

[0024] In this specification, each process may be executed as a function of its own nature, or it may be executed through a predetermined application. Alternatively, it may be executed by loading a predetermined program that includes each process.

[0025] This section details the processes that System 0 will execute. The acquisition module 20 acquires the first observation statement and the source data 3 capable of generating the observation statement from the first terminal 1 (step S1). At this time, the communication method between the first terminal 1 and the second terminal 2 may be wired or wireless, and the acquisition of the first observation statement and the image data corresponding to the first observation statement may be done using external services or applications such as email or cloud drives. The initial findings document acquired by acquisition module 20 includes, for example, the patient's condition, symptoms, and possible disease name from whom the specimen was collected, and is prepared by the medical institution or clinician that requested the pathology examination. Furthermore, source data 3 includes at least image data corresponding to the first finding statement and is the data that will be used to generate the second finding statement in the subsequent flow. The image data corresponding to the first finding statement is taken for medical and research purposes and includes at least one of the following as the subject of the image: organ location, description of pathological characteristics, test results, or diagnostic results. The file format of the first observation statement and the source data 3 acquired by the acquisition module 20 is preferably some kind of image data, but the first observation statement may be text data, and the file format of the image data or text data is not specified.

[0026] The findings generation module 21 generates a second findings statement based on the source data 3 acquired by the acquisition module 20 (steps S2-S4). Note that this generation may be performed using only a portion of the source data 3. Specifically, the observation generation module 21 includes an LLM (Large Language Model) 4. The LLM included in the observation generation module 21 may be, for example, a GPT (Generative Pre-trained Transformer) or a BERT (Bidirectional Encoder Representations from Transformers), and the type of method is not restricted. Based on Figure 4, the configuration and functions of LLM4 will be explained. Figure 4 shows an example of the LLM4 model provided by the findings generation module 21. LLM4 takes words, sentences, or image data containing them as input data. First, the vector transformation layer converts the input data into operable vectors. The vector transformation layer also includes a process to transform the positional information of words within each input data into vectors. Next, attention layers 1 and 2 perform tasks such as determining which parts are important and normalizing these vector quantities. Then, the feedforward layer weights these calculation results. This process from attention layer 1 to the feedforward layer is repeated multiple times depending on the scale of the model. Finally, in the output layer, output data is generated by adjusting the calculation results using linear functions, softmax functions, etc., to make the calculation results usable as numerical values. In the present invention, LLM4 is pre-trained with sufficient image data, such as descriptions of organ locations, pathological characteristics, test results, and diagnostic results, as well as observation texts created based on that image data. By importing the source data 3 into the trained LLM4, a second observation text corresponding to the source data 3 is generated. LLM4 can be retrained at any time.

[0027] Furthermore, the findings generation module 21 may have multiple LLM4s. In this case, the second user using the second terminal 2 selects an LLM4 specialized for a particular case based on the information contained in the first findings statement and the source data 3 (step S2). Specifically, examples include selecting LLM4B, which belongs to the clinical test the second user is about to perform (let's say a blood test), from among LLM4A specialized for cancer tumors, LLM4B specialized for hematological diseases, LLM4C specialized for electroencephalogram abnormalities, etc. As shown in Figure 5, a Convolutional Neural Network (CNN) may also be used in the selection of LLM4 by the second user. In this case, by inputting the source data 3 into the CNN, features contained in source data 3 are extracted, and the LLM4 using the training data closest to the extracted features is selected. If different source data 3' is input, a different LLM4 than the one used when source data 3 was input is selected.

[0028] To summarize the processing of the findings generation module 21, first, the selection of LLM4 is received from the second user (step S2), the source data 3 is imported into the selected LLM4 (step S3), and a second findings statement is generated in accordance with the information described in the imported source data 3 (step S4). Figure 6 shows an example of a second finding statement generated by the processing of the finding generation module 21. In Figure 6(a), the green shaded area shows the source data 3 input to LLM4. By incorporating this data, a second finding statement like the one shown in Figure 6(b), "We see granulation tissue with glandular ductal epithelia and inflammatory cells; lamina propria infiltration; and inflammatory cells are prominently present, indicating active inflammation," is generated.

[0029] The similarity calculation module 22 calculates the similarity between the first observation statement obtained in the acquisition module 20 and the second observation statement generated in the observation generation module 21 (steps S5 to S8). An example of similarity calculation in the similarity calculation module 22 will be specifically explained with reference to Figure 7. Figure 7 is a schematic diagram showing the comparison of text vectors performed in the similarity calculation module 22. The similarity calculation module 22 extracts keywords related to positive and negative test results and pathology types from the first and second findings statements (step S5). Keywords extracted at this time include the type of histopathological type, positive / negative test results, and measures indicating the progression of the disease, such as the stage and category of Genomics England. A text vector is calculated for each extracted keyword, generating a text vector group 5A representing the feature quantities of the content described in the first observation statement and a text vector group 5B representing the content of the second observation statement (step S6). The calculated text vector groups may consist of a single text vector or may be represented in two or more dimensions. The distance between the region to which text vector group 5A belongs and the region to which text vector group 5B belongs is measured (Step S7). If this distance between regions exceeds a certain threshold, the first observation sentence and the second observation sentence are determined to be similar; if it is below the threshold, they are determined to be dissimilar (Step S8). Furthermore, the text vector group 5B may be compared not only with the text vector group 5A, but also with a text vector group similarly calculated from the training data used to train LLM4, which has been previously stored in the similarity calculation module 22.

[0030] Furthermore, in the findings generation module 21, it is also possible to intentionally create a second findings statement ' that differs in opinion from the second findings statement created by selecting a different LLM4 from the first LLM4 created, and to measure the distance between the text vector group 5B' and the text vector group 5A using a similar process, thereby confirming that the distance between the text vector group 5B and the text vector group 5A generated from the first second findings statement is shorter than the distance between the text vector group 5B' and the text vector group 5A.

[0031] The output module 23 outputs the first observation statement if the similarity calculation module 22 determines that the first observation statement and the second observation statement are similar (step S9-1). This output may be displayed on the screen of the second terminal 2, transmitted to another terminal by some means via the network, or printed through an external device connected to the second terminal 2. Furthermore, if the similarity calculation module 22 determines that the first and second observation statements are dissimilar, the output module 23 will operate after the editing module 24 has functioned.

[0032] The editing module 24 functions only when the similarity calculation module 22 determines that the first and second findings are dissimilar, and accepts edits to the first or second findings from the second user (step S9-2). This editing may be performed by someone other than the second user, and may be done using the input section provided on the second terminal 2, or using an external device connected to the second terminal 2. An example of edited content is to add "Inflammatory cells almost normal colonic mucosal epithelia." to the end of the sentence "We see granulation tissue with glandular ductal epithelia and inflammatory cells lamina propria infiltration inflammatory cells are prominently present active inflammation." In other words, possible edits include correcting typographical errors, changing the case, or modifying the category to which the case belongs. The first or second opinion document, once edited, will be designated as the third opinion document.

[0033] If the editing module 24 is functioning correctly, the output module 23 outputs the third-party observation statement after editing is complete (step S10). This output is performed in the same manner as in step S9-1 described above.

[0034] Through the above process, by comparing the first report written by a medical professional with the second report written based on the image data corresponding to the first report, it is possible to automatically verify whether the first report is correctly related to the requested examination and to check for typographical errors, omissions, or errors in the diagnosis within the report, thereby improving the efficiency of tasks related to medical professionals' reports. Furthermore, since errors can be directly edited, corrections to the written report can be made easily.

[0035] Hereafter, a second embodiment of System 0, which is another embodiment of the present invention, will be described. Note that the same functions and components as in the first embodiment are denoted by the same reference numerals, and their descriptions are omitted.

[0036] The system configuration of System 0 of the second embodiment, which is another embodiment of the present invention, will be described with reference to Figure 8. Figure 8 is an overall configuration diagram of System 0 of the second embodiment of the present invention. In the second embodiment, similar to the first embodiment, the second terminal 2, in cooperation with the communication unit, implements the acquisition module 20, the findings generation module 21, the similarity calculation module 22, the output module 23, and the editing module 24 by having the control unit read a predetermined program. Furthermore, unlike the first embodiment, the second terminal 2 implements an integration module 25, a determination module 26, and a library module 27. Furthermore, it is not necessary to implement all modules at once on the second terminal 2; system 0 can be used with only specific modules implemented.

[0037] An overview of the processes performed by System 0 of the second embodiment of the present invention will be described with reference to Figure 9. Figure 9 is a flowchart showing the processes performed by System 0 of the second embodiment of the present invention. As shown in Figure 9, the system 0 of the second embodiment consists of steps S1 to S13 and further comprises an integration module 25 (step S11) that integrates and stores the output observation text output by the output module and at least image data; a determination module 26 (step S12) that recognizes that the output observation text has been received by the user as a reward and determines whether or not the output observation text has received a reward; and a library module 27 (step S13) that stores the output observation text stored in the integration unit and the image data when it is determined that the output observation text has received a reward.

[0038] In the following explanation, it is assumed that, as a result of the processing up to step S10, the output module 23 has output either the first observation statement or the third observation statement that has been edited in the editing module 24.

[0039] The integration module 25 integrates the first or third findings statement (hereinafter referred to as the output findings statement) output by the output module 23 with the image data corresponding to the first findings statement and saves it in the temporary storage area (step S11). This integrated data may include data other than image data corresponding to the first findings statement.

[0040] The determination module 26 determines whether the output report has been received by a user, including a second user (step S12). One example of a situation in which an opinion is deemed to have been accepted is when compensation is exchanged between the organizations to which the first and second users belong, or the recipients of the opinion statement.

[0041] When the determination module 26 determines that the output report has earned a reward, the library module 27 copies or moves the integrated data stored in the primary storage area of ​​the integration unit and saves it (step S13). Furthermore, it is preferable to use the integrated data stored in the library module 27 for retraining the LLM4 provided in the findings generation module 21.

[0042] Through the above process, System 0 stores the outputted observation statements as correct data and uses them for learning, which can sometimes further optimize the LLM without having to prepare a large amount of additional training data.

[0043] The means and functions described above are realized by a computer (including a CPU, information processing unit, various terminals, etc.) incorporating and executing a predetermined program. This program may be provided, for example, via a network from the computer, or through a cloud service. Alternatively, this program may be provided in the form of a recording medium that can be read by a computer. In this case, the computer reads the program from this recording medium, transfers it to an internal or external recording device, records it, and executes it. Alternatively, the program may be pre-recorded on a recording device and provided to the computer from that recording device via a communication line.

[0044] Although embodiments of the present invention have been described above, the present invention is not limited to these embodiments. Furthermore, the effects described in the embodiments of the present invention are merely a list of the most preferred effects arising from the present invention, and the effects of the present invention are not limited to those described in the embodiments. [Explanation of symbols]

[0045] 0: System 1: First terminal 2: Second terminal 3: Source data 4:LLM 5: Text vector group 20: Acquisition Module 21: Findings Generation Module 22: Similarity Calculation Module 23: Output Module 24: Editing Module 25: Integration Module 26: Judgment Module 27: Library Modules

Claims

1. A system for improving the efficiency of findings processing used between the first terminal and the second terminal, An acquisition unit that acquires a first observation statement and image data corresponding to the first observation statement from the first terminal, A report generation unit generates a second report based on source data that includes at least the acquired image data, A similarity calculation unit that calculates the similarity between the first observation statement and the second observation statement, An output unit outputs the first observation statement as an output observation statement when the similarity between the first observation statement and the second observation statement calculated by the similarity calculation unit exceeds a predetermined threshold, A system for improving the efficiency of findings processing, characterized by comprising the following features.

2. An editing unit that accepts editing requests for the first or second observation statement if the similarity between the first and second observation statements calculated by the similarity calculation unit does not exceed a predetermined threshold, Furthermore, The report work efficiency system according to claim 1, characterized in that the output unit outputs the report edited by the editing unit as an output report when the editing unit is functioning.

3. An integration unit integrates and stores the output observation statement output by the output unit and at least the image data, The findings work efficiency system according to claim 1 or 2, further comprising:

4. A determination unit that determines whether the output report has been accepted by the user, When it is determined that the output report has received a reward, the library that stores the output report and image data stored in the integration unit, The findings work efficiency system according to claim 3, further comprising:

5. A system for improving the efficiency of findings processing used between the first terminal and the second terminal, The steps include obtaining a first report and image data corresponding to the first report from the first terminal, A step of generating a second observation statement based on source data that includes at least the acquired image data, A step of calculating the similarity between the first observation statement and the second observation statement, In the step of calculating the similarity, if the similarity between the first observation statement and the second observation statement calculated in the step of calculating the similarity exceeds a predetermined threshold, the first observation statement is output. A method for improving the efficiency of findings processing, characterized by comprising the following features.

6. On the computer, Steps include obtaining the first findings statement and image data corresponding to the first findings statement. A step of generating a second observation statement based on source data which includes at least the acquired image data, A step of calculating the similarity between the first observation statement and the second observation statement, In the step of calculating the similarity, if the similarity between the first observation statement and the second observation statement calculated in the step of calculating the similarity exceeds a predetermined threshold, the first observation statement is output. A computer-readable program for executing a command.

Images