Medical institution analysis system, management device, medical institution analysis method, and medical institution analysis program

Abstract

The objective is to efficiently conduct qualitative evaluations that delve into the diagnostic content of medical doctors and other medical institutions. [Solution] A local server 20 receives organizational information of a medical institution and medical information, such as a diagnostic report and / or test images, anonymizes the medical information, and transmits the anonymized data to a cloud management server 10. The cloud management server 10 extracts multiple pieces of label information from text included in the received anonymized data, vectorizes each of the multiple pieces of label information, and generates vector information. The cloud management server 10 then generates a hierarchical knowledge graph based on the multiple pieces of vector information. When the cloud management server 10 receives a search request from a doctor terminal 30 via the local server 20, it searches hierarchical knowledge graph data based on the search request and notifies the doctor terminal 30 of the results of a difference analysis of the hierarchical knowledge graph corresponding to the search request.

Description

[Technical Field]

[0001] The present invention relates to a medical institution analysis system, a management device, a medical institution analysis method, and a medical institution analysis program that can efficiently perform in-depth qualitative evaluation of diagnostic contents by doctors or the like at medical institutions. [Background technology]

[0002] Conventionally, medical institutions such as hospitals often have different characteristics from other medical institutions. For this reason, techniques for evaluating medical institutions by comparing multiple medical institutions are known. For example, Patent Literature 1 discloses a technique for understanding the financial status and service quality of a specific medical institution by comparing indicators such as the number of inpatients, length of stay, and illnesses of inpatients at a specific medical institution with indicators of other medical institutions. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2010-170430 Summary of the Invention [Problem to be solved by the invention]

[0004] However, while the above-mentioned Patent Document 1 compares medical institutions based on quantitative indicators, it is important to evaluate them using qualitative and quantitative indicators that delve into the diagnostic content of the medical institution's doctors. For example, doctors and radiologists at medical institutions interpret test images captured by modalities such as CT (Computed Tomography) and MRI (Magnetic Resonance Imaging), and compile the interpretation results into a diagnostic report. It is desirable to identify the strengths and weaknesses of specific medical institutions based on these medical examination results.

[0005] The present invention has been made to solve the problems (issues) of the prior art, and provides a qualitative analysis of the diagnosis content by doctors and others at medical institutions. Evaluate using metrics and quantitative indicators to identify the strengths and weaknesses of specific medical institutions The present invention aims to provide a medical institution analysis system, a management device, a medical institution analysis method, and a medical institution analysis program that can perform the above-mentioned analysis. [Means for solving the problem]

[0006] In order to solve the above-mentioned problems and achieve the object, the present invention provides a medical institution analysis system having a management device that manages medical institution information including medical images captured by modality equipment installed in a medical institution and diagnostic reports of the medical images, the management device comprising: The aforementioned From the text included in the medical institution information, for each input item of medical department, test request, findings, and diagnosis The information consists of qualitative information such as the location and disease name, and quantitative information such as the size of the object to be detected. The system is characterized by comprising a label information extraction means for extracting each piece of label information, a vector information generation means for vectorizing each piece of label information to generate a plurality of pieces of vector information, a graph network generation means for generating a graph network of medical institutions based on the plurality of pieces of vector information, a management means for managing the graph networks of the plurality of medical institutions, and a difference analysis means for performing a difference analysis between the graph network of a specified medical institution included in the plurality of medical institutions and the graph networks of other medical institutions.

[0007] In the above invention, the graph network is a graph network in which sub-graph networks for each medical department forming the medical institution are associated with the medical institution.

[0008] Further, in the present invention, in the above invention, the label information extraction means extracts at least a site, a disease name, a degree of certainty of the disease name, and a size of the object to be detected from text included in the medical institution information. Z The label information is extracted from the medical institution information.

[0009] In the present invention, the label information extraction means extracts the label information for each input item of test request, findings, and diagnosis.

[0010] Further, in the present invention, in the above invention, the label information extraction means extracts at least one or more of the type of examination equipment, the presence or absence of a contrast agent, and the imaging conditions from text accompanying the medical image included in the plurality of pieces of medical institution information as the label information for the medical institution information. moreover The present invention is characterized by extracting

[0011] Further, in the present invention, in the above invention, the label information extraction means extracts at least one or more of patient attributes, clinical background, and progress records from text included in the plurality of pieces of medical institution information as the label information for the medical institution information. moreover The present invention is characterized by extracting

[0013] In the present invention, the vector information generating means generates a plurality of pieces of vector information by vectorizing each of the plurality of pieces of label information using a natural language processing technique.

[0014] In the above invention, the natural language processing technology is a predetermined large-scale language model.

[0015] In addition, in the present invention, the graph network generation means generates a graph network in which one or more of the parts, disease names, test types, and imaging conditions contained in the plurality of vector information or the text corresponding to the vector information are used as nodes, and the relationships between the nodes are expressed as edges.

[0016] Furthermore, in the above invention, the differential analysis means, upon receiving a request for differential analysis of the graph networks of a first medical institution and a second medical institution, generates a graph network of the first medical institution that makes it possible to distinguish the difference between the first graph network corresponding to the first medical institution and the second graph network corresponding to the second medical institution.

[0017] The present invention also provides a management device for managing medical institution information including medical images captured by modality equipment installed in a medical institution and diagnostic reports of the medical images, the management device comprising: The aforementioned From the text included in the medical institution information, for each input item of medical department, test request, findings, and diagnosis The information consists of qualitative information such as the location and disease name, and quantitative information such as the size of the object to be detected. The system is characterized by comprising a label information extraction means for extracting each of a plurality of pieces of label information, a vector information generation means for vectorizing each of the plurality of pieces of label information to generate a plurality of pieces of vector information, a graph network generation means for generating a graph network of medical institutions based on the plurality of pieces of vector information, a management means for managing the graph networks of the plurality of medical institutions, and a difference analysis means for performing a difference analysis between the graph network of a specified medical institution included in the plurality of medical institutions and the graph networks of other medical institutions.

[0018] The present invention also provides a medical institution analysis method in a medical institution analysis system having a management device that manages medical institution information including medical images captured by modality equipment installed in a medical institution and diagnostic reports of the medical images, the management device comprising: The aforementioned From the text included in the medical institution information, for each input item of medical department, test request, findings, and diagnosis The information consists of qualitative information such as the location and disease name, and quantitative information such as the size of the object to be detected.The method includes a label information extraction step of extracting each of a plurality of pieces of label information, a vector information generation step in which the management device vectorizes each of the plurality of pieces of label information to generate a plurality of pieces of vector information, a graph network generation step in which the management device generates a graph network of medical institutions based on the plurality of pieces of vector information, a management step in which the management device manages the graph networks of the plurality of medical institutions, and a difference analysis step in which the management device performs a difference analysis between the graph network of a specified medical institution included in the plurality of medical institutions and the graph networks of other medical institutions.

[0019] The present invention also provides a medical institution analysis program executed in a management device that manages medical institution information including medical images captured by modality equipment installed in a medical institution and diagnostic reports of the medical images, the program comprising: The aforementioned From the text included in the medical institution information, for each input item of medical department, test request, findings, and diagnosis The information consists of qualitative information such as the location and disease name, and quantitative information such as the size of the object to be detected. The method is characterized in that the computer executes a label information extraction procedure for extracting each piece of label information, a vector information generation procedure for vectorizing each of the plurality of label information to generate a plurality of pieces of vector information, a graph network generation procedure for generating a graph network of medical institutions based on the plurality of pieces of vector information, a management procedure for managing the graph networks of the plurality of medical institutions, and a differential analysis procedure for performing a differential analysis of the graph network of a specified medical institution included in the plurality of medical institutions with the graph networks of other medical institutions. [Effects of the Invention]

[0020] According to the present invention, a qualitative evaluation that goes into the details of a diagnosis by a doctor or the like at a medical institution can be efficiently performed. [Brief explanation of the drawings]

[0021] [Figure 1] FIG. 1 is a diagram illustrating an overview of a medical institution analysis system according to an embodiment. [Figure 2]FIG. 2 is a functional block diagram illustrating the configuration of the cloud management server illustrated in FIG. [Figure 3] FIG. 3 is a diagram showing an example of a hierarchical knowledge graph. [Figure 4] FIG. 4 is a functional block diagram showing the configuration of the local server shown in FIG. [Figure 5] FIG. 5 is a diagram showing an example of a menu screen of the medical institution analysis system. [Figure 6] FIG. 6 is a diagram (part 1) showing an example of a display screen for comparing medical institutions. [Figure 7] FIG. 7 is a diagram (part 2) showing an example of the display screen for comparing medical institutions. [Figure 8] FIG. 8 is a diagram (part 1) showing an example of a display screen for comparing disease cases. [Figure 9] FIG. 9 is a diagram (part 2) showing an example of the display screen for comparing disease cases. [Figure 10] FIG. 10 is a diagram (part 1) showing an example of a display screen for searching specialized medical institutions. [Figure 11] FIG. 11 is a diagram (part 2) showing an example of a display screen for searching specialized medical institutions. [Figure 12] FIG. 12 is a sequence chart showing the processing procedure for updating hierarchical knowledge graph data. [Figure 13] FIG. 13 is a sequence chart showing the procedure for comparing medical institutions. [Figure 14] FIG. 14 is a sequence chart showing the procedure for comparing disease cases. [Figure 15] FIG. 15 is a sequence chart showing the processing procedure for searching for specialized medical institutions. [Figure 16] FIG. 16 is a diagram illustrating an example of a hardware configuration. DETAILED DESCRIPTION OF THE INVENTION

[0022] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of a medical institution analysis system, a management device, a medical institution analysis method, and a medical institution analysis program according to the present invention will be described in detail with reference to the accompanying drawings.

[0023] <Overview of Medical Institution Analysis System> An overview of the medical institution analysis system according to this embodiment will be described. This medical information analysis system receives organizational information of multiple medical institutions and anonymized information of multiple medical institutions (hereinafter referred to as "medical information"). The medical information management system then extracts multiple pieces of label information from text included in the organizational information of the medical institutions and the multiple pieces of medical information, vectorizes each piece of label information to generate vector information, and generates a hierarchical graph network (hereinafter referred to as "hierarchical knowledge graph") for each medical institution based on the vectorized information.

[0024] Based on the hierarchical knowledge graph, the medical information analysis system can display a hierarchical knowledge graph comparing medical institutions, display comparative data on the number of cases, and display the results of narrowing down specialized hospitals based on descriptions of symptoms, etc.

[0025] Fig. 1 is a diagram showing an overview of a medical institution analysis system according to an embodiment. As shown in Fig. 1, this medical information analysis system includes a cloud management server 10, which is a virtual management server that provides cloud computing services for storing, managing, and analyzing medical information on the cloud, and local servers 20 in medical institutions 1 and 2, and the cloud management server 10 and the local servers 20 in medical institutions 1 and 2 are connected via a network N. Note that, for simplicity of explanation, a case in which there are two medical institutions will be described here, but the number of medical institutions may be three or more.

[0026] Furthermore, medical institution 1 and medical institution 2 further have a doctor terminal 30 and modality equipment (not shown), which are communicably connected to each other via LANs within medical institution 1 and medical institution 2. The local server 20 of the medical information analysis system receives diagnostic reports from the doctor terminal 30, etc. The local server 20 within medical institution 1 receives examination image data from the modality equipment (not shown). The local server 20 within medical institution 1 then performs anonymization processing on medical information data (diagnostic reports, examination image data, etc.) within medical institution 1, and transmits the anonymized data together with organizational information within medical institution 1 to the cloud management server 10 (S1). The local server 20 within medical institution 2 also transmits anonymized data to the cloud management server 10 in a similar manner.

[0027] The cloud management server 10 extracts multiple pieces of label information from the received organizational information of the medical institution and the text included in the medical information of the anonymized data, and vectorizes the multiple pieces of label information and the multiple sentences included in the medical information to generate multiple pieces of vector information (S2). The cloud management server 10 then updates (generates) the hierarchical knowledge graph data of each medical institution based on the multiple pieces of vector information (S3). Then, for example, when a search request for a comparison of medical institutions is sent from the doctor terminal 30 via the local server 20 (S4), the cloud management server 10 searches the hierarchical knowledge graph data and performs a difference analysis of the hierarchical knowledge graph corresponding to the search request (S5). The cloud management server 10 then sends the search result data to the local server 20 (S6), and the local server stores the search result data in a memory unit and displays it on the doctor terminal 30 (S7).

[0028] In addition to the medical image, the medical information generated by the modality equipment (not shown) includes "personal information of the recipient corresponding to the medical image (name, date of birth, age, etc.)" and "information related to image generation (examination information such as the date and time of image generation, name and number of the modality equipment that generated the image, type of examination, and examination area)" as appropriate in accordance with the DICOM (Digital Imaging and Communications in Medicine) standard.

[0029] As described above, in this embodiment, the local servers 20 in the medical institutions 1 and 2 receive organizational information of the medical institutions and medical information such as diagnostic reports and / or test images, anonymize the medical information, and transmit the anonymized data to the cloud management server 10. The cloud management server 10 extracts multiple pieces of label information from text included in the received anonymized data, vectorizes each of the multiple pieces of label information, and generates vector information. The cloud management server 10 then generates a hierarchical knowledge graph based on the multiple pieces of vector information. When the cloud management server 10 receives a search request from the doctor terminal 30 of the medical institution 1 via the local server 20, it searches the hierarchical knowledge graph data based on the search request and notifies the doctor terminal 30 of the medical institution 1 of the results of a difference analysis of the hierarchical knowledge graph corresponding to the search request.

[0030] <Configuration of Cloud Management Server 10> Next, the configuration of the cloud management server 10 of the medical information analysis system according to this embodiment will be described. Fig. 2 is a functional block diagram showing the configuration of the cloud management server 10 shown in Fig. 1. As shown in Fig. 2, the cloud management server 10 has a communication I / F unit 13, a storage unit 15, and a control unit 16, and is connected to an input unit 11 and a display unit 12.

[0031] The input unit 11 is an input device such as a keyboard or a mouse, the display unit 12 is a display device such as a liquid crystal panel, and the communication I / F unit 13 is a communication interface unit for connecting to the local server 20 via the network N.

[0032] The storage unit 15 is a storage device such as a hard disk drive or nonvolatile memory, and stores anonymized data 15a, label data 15b, Hospital A hierarchical knowledge graph data 15c, Hospital B hierarchical knowledge graph data 15d, and XX Association XX Hospital hierarchical knowledge graph data 15e. The anonymized data 15a is data of organizational information of medical institutions and anonymized medical information received from each local server 20 of medical institution 1 and medical institution 2. The anonymized data 15a includes diagnostic reports, radiology reports, and examination images.

[0033] The label data 15b is data of label information extracted from text included in the anonymized data 15a. The label data 15b stores data of labels extracted for each anonymized data received from each medical institution. The label information is extracted, for example, for each input item of the medical department or division of the medical institution 1, such as a test request, findings, or diagnosis. The label information is information obtained by extracting labels corresponding to the body part, disease name, certainty of the disease name, size of the detected object, patient attributes, clinical background, and progress record from text included in multiple pieces of medical information, and information obtained by extracting labels corresponding to the model of the examination machine, the presence or absence of a contrast agent, and imaging conditions from text accompanying the medical images included in the medical information.

[0034] The Hospital A hierarchical knowledge graph data 15c is hierarchical knowledge graph data of Hospital A, which represents multiple labels extracted based on data including organizational information of Hospital A received from Hospital A in the label data 15b and the relationships between the labels using nodes and edges. For example, as shown in FIG. 3, the hierarchical knowledge graph data of Hospital A is composed of three layers: a first layer L1, a second layer L2, and a third layer L3. The first layer L1 is composed of a node A1 of Hospital A, and the second layer L2 is composed of nodes of departments of Hospital A (cardiology B1, cardiovascular surgery B2, internal medicine B3, surgery B4, neurosurgery B5). The third layer L3 is composed of nodes (C1, C2, ... C11) representing body parts, disease names, symptoms, etc. The number of nodes in the second layer L2 and the third layer L3 varies depending on the number of departments and the number of body parts, disease states, symptoms, etc. of the medical institution.

[0035] Similar to the Hospital A hierarchical knowledge graph data 15c, the Hospital B hierarchical knowledge graph data 15d is hierarchical knowledge graph data of Hospital B, which represents, with nodes and edges, multiple labels and the relationships between the labels extracted based on data including organizational information of Hospital B received from Hospital B in the label data 15b. Similar to the Hospital A hierarchical knowledge graph data 15c, the XX Association XX Hospital hierarchical knowledge graph data 15e is hierarchical knowledge graph data of XX Association XX Hospital, which represents, with nodes and edges, multiple labels and the relationships between the labels extracted based on data including organizational information of XX Association XX Hospital received from Hospital A in the label data 15b.

[0036] The control unit 16 is a control unit that performs overall control of the cloud management server 10, and includes an anonymized data receiving unit 16a, a label generating unit 16b, a vectorization processing unit 16c, a hierarchical knowledge graph data updating unit 16d, a hierarchical knowledge graph data managing unit 16e, a search request receiving processing unit 16f, a hierarchical knowledge graph data search processing unit 16g, a vector search processing unit 16h, a difference analysis processing unit 16i, and a search result data sending unit 16j. In practice, by loading these programs into the CPU and executing them, the anonymized data receiving unit 16a, the label generating unit 16b, the vectorization processing unit 16c, the hierarchical knowledge graph data updating unit 16d, the hierarchical knowledge graph data managing unit 16e, the search request receiving processing unit 16f, the hierarchical knowledge graph data search processing unit 16g, the vector search processing unit 16h, the difference analysis processing unit 16i, and the search result data sending unit 16j are caused to execute processes corresponding to the respective programs.

[0037] The anonymized data receiving unit 16a is a processing unit that receives anonymized data obtained by anonymizing medical information transmitted from the local server 20. The anonymized data receiving unit 16a stores the received anonymized data in the storage unit 15.

[0038] The label generation unit 16b is a processing unit that reads the anonymized data 15a from the storage unit 15 and generates label information from the text of the tissue information and medical information included in the anonymized data. Examples of the label information include the medical department included in the tissue information, the site included in the image diagnosis report, the disease name, the certainty of the disease name (positive, negative, suspected), the size of the detected object (tumor, etc.), the model of the examination machine (modality) included in the information accompanying the examination image, the presence or absence of a contrast agent, the imaging conditions, patient attributes included in the information in the clinician's diagnostic report (gender, age, family history, medical history, treatment history, medication information, allergy information), clinical history, and progress records. Note that the label generation unit 16b preferably automatically assigns labels using a machine learning model that has been trained in advance using an existing labeled dataset.

[0039] The vectorization processing unit 16c is a processing unit that converts the plurality of pieces of label information generated by the label generating unit 16b, the medical information text included in the anonymized data 15a, and the like into corresponding vector information.

[0040] The hierarchical knowledge graph data update unit 16d is a processing unit that extracts entities (such as medical departments, body parts, symptoms, and disease names) from the vectorized data based on the organizational information and medical information vectorized by the vectorization processing unit 16c, identifies relationships between the extracted entities, and generates a hierarchical knowledge graph with the entities as nodes and the relationships as edges. The hierarchical knowledge graph data generated by the hierarchical knowledge graph data update unit 16d is composed of, for example, three layers: a first layer L1, a second layer L2, and a third layer L3. The first layer L1 is composed of nodes for medical institutions, and the second layer L2 is composed of nodes for medical departments at the medical institutions. The third layer L3 is composed of nodes for body parts, disease names, symptoms, etc.

[0041] The hierarchical knowledge graph data management unit 16e is a processing unit that manages the generated hierarchical knowledge graph data. Since new medical information is generated daily at each medical institution, the hierarchical knowledge graph data also needs to be updated according to the situation. For example, the hierarchical knowledge graph data management unit 16e monitors updates to the anonymized data 15a, and when the anonymized data 15a is updated, performs label generation processing and vectorization processing on the newly updated differential data, thereby updating the hierarchical knowledge graph data. Note that the hierarchical knowledge graph data may be updated all at once by performing batch processing overnight, or may be updated whenever the anonymized data 15a is updated.

[0042] The search request reception processing unit 16f is a processing unit that receives search request data transmitted from the doctor terminal 30 via the local server 20, and extracts labels and vectorizes them based on the search request data.

[0043] The hierarchical knowledge graph data search processing unit 16g is a processing unit that references the hierarchical knowledge graph data and generates a hierarchical sub-knowledge graph corresponding to the label. The vector search processing unit 16h is a processing unit that searches the hierarchical knowledge graph data based on vector values ​​that are vectorized based on the text included in the search request data generated by the search request reception processing unit 16f. The vector search uses the similarity between vectors based on the vector values ​​using algorithms such as cosine similarity and Euclidean distance.

[0044] The difference analysis processing unit 16i is a processing unit that performs a difference analysis between predetermined hierarchical knowledge graph data contained in a plurality of medical institutions and hierarchical knowledge graph data of other medical institutions. For example, when comparing Hospital A and Hospital B, the hierarchical knowledge graph data of Hospital A is compared with the hierarchical knowledge graph data of Hospital B, and hierarchical knowledge graph data that can distinguish the difference between Hospital A and Hospital B is generated.

[0045] The search result data transmission unit 16j is a processing unit that transmits to the local server 20 the search result data searched by the vector search processing unit 16h or the difference analysis processing unit 16i.

[0046] <Configuration of local server 20> Next, the configuration of the local server 20 of the medical information analysis system will be described. Fig. 4 is a functional block diagram showing the configuration of the local server 20 shown in Fig. 1. As shown in Fig. 4, the local server 20 has a communication I / F unit 23, a storage unit 25, and a control unit 26, and is connected to an input unit 21 and a display unit 22.

[0047] The input unit 21 is an input device such as a keyboard or a mouse, the display unit 22 is a display device such as an LCD panel, and the communication I / F unit 23 is a communication interface unit for connecting to the cloud management server 10 and the doctor terminal 30 via the network N.

[0048] The storage unit 25 is a storage device such as a hard disk drive or nonvolatile memory, and stores medical information data 25a, anonymized data 25b, and search result data 25c. The medical information data 25a is medical information data such as diagnostic reports, radiology reports, and examination images received from the doctor terminal 30 and modality devices in the medical institution 1. The anonymized data 25b is data obtained by deleting personal information such as name, address, and telephone number from the organizational information of the medical institution 1 and the medical information data 25a so that the personal information cannot be identified.

[0049] The control unit 26 is a control unit that performs overall control of the local server 20, and includes a medical information data receiving unit 26a, an anonymization processing unit 26b, an anonymized data transmitting unit 26c, a search request data transmitting unit 26d, a search result data receiving unit 26e, and a display control data transmitting unit 26f. In practice, by loading and executing these programs into the CPU, the medical information data receiving unit 26a, the anonymization processing unit 26b, the anonymized data transmitting unit 26c, the search request data transmitting unit 26d, the search result data receiving unit 26e, and the display control data transmitting unit 26f execute their respective corresponding processes.

[0050] The medical information data receiving unit 26a is a processing unit that receives medical information data such as diagnostic reports, image interpretation reports, and examination images transmitted from the doctor terminal 30 and modality devices in the medical institution.

[0051] The anonymized data transmission unit 26c performs a process of transmitting the anonymized data 25b to the cloud management server 10. The anonymized data transmission unit 26c preferably transmits the anonymized data 25b at night, for example, when outpatient consultations at medical institutions have been completed and there is less work to create diagnostic reports and take examination images. However, the local server 20 may receive medical information data, perform anonymization processing sequentially, and transmit the anonymized medical information data to the cloud management server 10 once the anonymization processing is complete.

[0052] The search request data sending unit 26d is a processing unit that, upon receiving a search request from the doctor terminal 30, sends the data of the search request to the cloud management server 10. The search result data receiving unit 26e is a processing unit that receives search result data searched by the cloud management server 10 based on the search request data. The search result data receiving unit 26e stores the search result data received from the cloud management server 10 as search result data 25c in the memory unit 25. The doctor terminal 30 can check past search results by referring to the search result data 25c.

[0053] The display control data transmission unit 26f is a processing unit that transmits display control data so that the operation screen of the medical institution analysis system transmitted by the cloud management server 10 can be displayed on the doctor terminal 30. Communication between the doctor terminal 30 and the cloud management server 10 is performed via the local server 20.

[0054] <Menu screen of medical institution analysis system> Next, an example of a menu screen of the medical institution analysis system will be described. Figure 5 is a diagram showing an example of the menu screen of the medical institution analysis system. As shown in Figure 5, when the doctor terminal 30 accesses the cloud management server 10 via the local server 20, the medical institution analysis system displays a search menu for comparing medical institutions, comparing disease cases, and searching specialized medical institutions on the display screen D of the doctor terminal 30.

[0055] Medical institution comparison displays organizational differences between medical institutions in a hierarchical knowledge graph. Disease case comparison searches for and displays differences in the number of cases, facilities, etc. between medical institutions for a specified disease case. Specialized medical institution search allows you to input the disease condition you want to search for, and searches and displays the disease name that can be inferred from the condition and hospitals that can treat that disease.

[0056] <Example of medical institution comparison> Next, a display example relating to medical institution comparison in the medical institution analysis system will be described. Figures 6 and 7 are diagrams showing an example of the display screen for medical institution comparison. As shown in Figure 6, when medical institution comparison is selected on the menu screen of the doctor terminal 30, the medical institution analysis system displays a medical institution selection area D1, a target medical institution selection area D2, and a comparison execution button D3 for executing the comparison on the display screen D of the doctor terminal 30.

[0057] In the medical institution selection area D1, Hospital A, Hospital B, XX Association XX Hospital, etc. are displayed as selectable options. In addition, in the target medical institution selection area D2, all of Hospital A, Hospital B, XX Association XX Hospital, etc. are displayed as selectable options. Here, if a doctor selects Hospital A in the medical institution selection area D1, selects Hospital B in the target medical institution selection area D2, and presses the comparison execution button D3, a search request to "compare Hospital A and Hospital B" is sent to the cloud management server 10.

[0058] When the cloud management server 10 receives a search request to “compare Hospital A and Hospital B,” it compares the Hospital A hierarchical knowledge graph data 15c with the Hospital B hierarchical knowledge graph data 15d. Then, the cloud management server 10, in the difference analysis processing unit 16i, analyzes the differences in the nodes and edges between the Hospital A hierarchical knowledge graph data 15c and the Hospital B hierarchical knowledge graph data 15d, generates hierarchical knowledge graph data in which the differences can be distinguished, and displays it on the doctor terminal 30.

[0059] For example, as shown in Figure 7, nodes and edges that are the same in the hierarchical knowledge graph data of Hospital A and the hierarchical knowledge graph data of Hospital B are displayed with solid lines, nodes and edges that do not exist in the hierarchical knowledge graph data of Hospital A but exist in the hierarchical knowledge graph data of Hospital B are displayed with dotted lines, and if the difference in the number of disease cases, for example, linked to a node exceeds a predetermined ratio, the node is displayed with a thick solid line. Note that by selecting a node displayed with a thick solid line, data such as the number of disease cases linked to the node can be displayed. This makes it possible to evaluate the strengths and weaknesses of one's own facility.

[0060] <Display of case comparison> Next, a display example for case comparison in the medical institution analysis system will be described. Note that the same components as those in the display example for medical institution comparison in Fig. 6 are given the same reference numerals, and detailed description thereof will be omitted. Figs. 8 and 9 are diagrams showing an example of a display screen for case comparison. As shown in Fig. 8, when case comparison is selected on the menu screen of the doctor terminal 30, the medical institution analysis system displays, on the display screen D of the doctor terminal 30, a case input area D4, a medical institution selection area D1, a target medical institution selection area D2, and a comparison execution button D3 for executing the comparison.

[0061] When the doctor inputs "malignant neoplasm of the pancreatic body" into the symptom input field D4 and confirms it, search request data for "comparison of cases of malignant neoplasm of the pancreatic body" is sent to the cloud management server 10. The cloud management server 10 performs label generation and vectorization processing based on the search request data, and searches for hierarchical knowledge graph data based on the vector values ​​generated by vectorization. The cloud management server 10 then displays the searched medical information on the doctor terminal 30.

[0062] Here, as shown in Figure 9, the comparison result area D5 displays "2021," "2022," "2023," and "2024" corresponding to the number of cases, "Hospital A," "45," "37," "42," and "41," and "Hospital B," "114," "132," "129," and "135," and displays "Hospital A," "CT, MRI, CR, XA, US," and "Hospital B," "PET-CT, CT, MRI, CR, XA, US" corresponding to "Equipment."

[0063] <Display of specialized medical institution search> Next, a display example for specialized medical institution search of the medical institution analysis system will be described. Figures 10 and 11 are diagrams showing an example of the display screen for specialized medical institution search. As shown in Figure 10, when specialized medical institution search is selected on the menu screen of the doctor terminal 30, the medical institution analysis system displays a description area D6 for the condition etc. to be searched for and a search execution button D7 for executing the search on the display screen D of the doctor terminal 30.

[0064] If a doctor enters the medical condition he or she wants to search for in the description area D6, such as "Hypoxemia at rest and on exertion is pronounced. Also, shortness of breath on exertion, chronic cough, and edema of the lower limbs are observed," and presses the search execution button D7, the text entered in the description area D6 regarding the medical condition he or she wants to search for is sent to the cloud management server 10 as search request data.

[0065] The cloud management server 10 performs label generation and vectorization processing based on the text included in the search request data, searches the hierarchical knowledge graph data, and extracts the predicted disease name, the names of specialized medical institutions that treat the disease name, the number of cases, etc. After that, the cloud management server 10 displays the extracted search result data on the doctor terminal 30.

[0066] Here, as shown in Figure 11, the predicted disease name area D8 displays "Suspected pulmonary venous obstruction or pulmonary capillary hemangiomatosis," and the specialized hospital institution area D9 displays "2021," "2022," "2023," and "2024," as well as "Number of cases," "12," "16," "18," and "20," corresponding to the searched hospital name "XX Association XX Hospital," and "CT, PETCT, MRI, DR, CR, XA, US" corresponding to "Equipment."

[0067] <Hierarchical knowledge graph data update procedure> Next, the processing procedure for updating hierarchical knowledge graph data in the medical information analysis system will be described. FIG. 12 is a sequence chart showing the processing procedure for updating hierarchical knowledge graph data. As shown in FIG. 12, the local server 20 receives medical information data from the doctor terminal 30 and modality devices in the medical institution 1 (step S101). Then, the local server 20 anonymizes the medical information data (step S102). Thereafter, the local server 20 transmits the anonymized data to the cloud management server 10 (step S103). The cloud management server 10 receives the anonymized data transmitted from the local server 20 (step S104).

[0068] Then, the cloud management server 10 generates label data from the anonymized data (step S105). After that, the cloud management server 10 vectorizes the anonymized data and the label data (step S106). Then, the cloud management server 10 updates the hierarchical knowledge graph data (step S107). If hierarchical knowledge graph data already constructed exists, the cloud management server 10 updates the knowledge graph data using new data, and if no knowledge graph data already constructed exists, the cloud management server 10 generates new knowledge graph data. Furthermore, the cloud management server 10 performs the update (generation) process of the hierarchical knowledge graph data for each medical institution.

[0069] <Processing procedure for comparing medical institutions> Next, the processing procedure for comparing medical institutions in the medical information analysis system will be described. Fig. 13 is a sequence chart showing the processing procedure for comparing medical institutions. As shown in Fig. 13, the doctor terminal 30 transmits search request data, which is the search conditions entered by the doctor to the local server 20 (step S201). The local server 20 transmits the search request data received from the doctor terminal 30 to the cloud management server 10 (step S202).

[0070] The cloud management server 10 receives the search request data sent from the local server 20 (step S203). The cloud management server 10 selects the hierarchical knowledge graph data of the medical institution from the information of the medical institution included in the search request data (step S204). Then, the cloud management server 10 searches the hierarchical knowledge graph data (step S205). Here, searching means generating sub-hierarchical knowledge graph data for each hierarchical knowledge graph data of each medical institution based on the search request data.

[0071] Thereafter, the cloud management server 10 analyzes the differences between the plurality of sub-hierarchical knowledge graph data (step S206). Then, the cloud management server 10 transmits the analysis results of the differences as search result data to the local server 20 (step S207). Upon receiving the search result data, the local server 20 stores the search result data in the storage unit 25 (step S208).

[0072] Then, the local server 20 transmits display data of the search result data to the doctor terminal 30 (step S209). Upon receiving the display data, the doctor terminal 30 controls the display of the search result data (step S210).

[0073] <Case comparison procedure> Next, the processing procedure for comparing disease cases in the medical information analysis system will be described. Fig. 14 is a sequence chart showing the processing procedure for comparing disease cases. As shown in Fig. 14, the doctor terminal 30 transmits search request data, which is the search conditions entered by the doctor to the local server 20 (step S301). The local server 20 transmits the search request data received from the doctor terminal 30 to the cloud management server 10 (step S302).

[0074] The cloud management server 10 receives the search request data sent from the local server 20 (step S303). The cloud management server 10 selects the hierarchical knowledge graph data of the medical institution from the information of the medical institution included in the search request data (step S304). Then, the cloud management server 10 searches the hierarchical knowledge graph data (step S305). After that, the cloud management server 10 sends the search result data to the local server 20 (step S306).

[0075] When the local server 20 receives the search result data, it stores the search result data in the storage unit 25 (step S307). Then, the local server 20 transmits display data of the search result data to the doctor terminal 30 (step S308). When the doctor terminal 30 receives the display data, it controls the display of the search result data (step S309).

[0076] <Procedure for searching specialized medical institutions> Next, the processing procedure for searching specialized medical institutions in the medical information analysis system will be described. Fig. 15 is a sequence chart showing the processing procedure for searching specialized medical institutions. As shown in Fig. 15, the doctor terminal 30 transmits search request data, which is the search conditions entered by the doctor to the local server 20, as search request data (step S401). The local server 20 transmits the search request data received from the doctor terminal 30 to the cloud management server 10 (step S402).

[0077] The cloud management server 10 receives the search request data sent from the local server 20 (step S403). Then, the cloud management server 10 searches for the hierarchical knowledge graph data (step S404). After that, the cloud management server 10 sends the search result data to the local server 20 (step S405).

[0078] When the local server 20 receives the search result data, it stores the search result data in the storage unit 25 (step S406). Then, the local server 20 transmits display data of the search result data to the doctor terminal 30 (step S407). When the doctor terminal 30 receives the display data, it controls the display of the search result data (step S408).

[0079] As described above, in this embodiment, the cloud management server 10 receives anonymized data of organizational information and medical information of a medical institution from the local server 20, extracts label information and vector information based on the anonymized data, and generates hierarchical knowledge graph data. In response to a search request from the doctor terminal 30, the cloud management server 10 extracts a search instruction from the search request data and searches multiple hierarchical knowledge graph data to generate multiple sub-hierarchical knowledge graph data. The cloud management server 10 is configured to analyze the differences between the multiple sub-knowledge graph data, generate hierarchical knowledge graph data in which the differences can be distinguished, and display the hierarchical knowledge graph data on the doctor terminal 30, etc. via the local server 20.

[0080] <Relationship with hardware> Next, the correspondence between the cloud management server 10 of the medical information analysis system according to this embodiment and the main hardware configuration of the computer will be described. Fig. 16 is a diagram showing an example of the hardware configuration.

[0081] Generally, a computer is configured such that a CPU 81, a ROM 82, a RAM 83, and a non-volatile memory 84 are connected via a bus 85. A hard disk drive may be provided instead of the non-volatile memory 84. For the sake of convenience of explanation, only the basic hardware configuration is shown.

[0082] Here, the ROM 82 or non-volatile memory 84 stores programs required to start the operating system (hereinafter simply referred to as "OS"), and the CPU 81 reads and executes the OS program from the ROM 82 or non-volatile memory 84 when the power is turned on.

[0083] On the other hand, various application programs executed on the OS are stored in non-volatile memory 84, and the CPU 81 executes the application programs while using RAM 83 as the main memory, thereby executing processes corresponding to the applications.

[0084] The medical information management program of the cloud management server 10 of the medical information analysis system according to this embodiment is also stored in the nonvolatile memory 84 or the like, like other application programs, and the CPU 81 loads and executes the analysis program. In the case of the cloud management server 10 of the medical information management system according to this embodiment, a medical information analysis program including routines corresponding to the anonymized data receiving unit 16a, the label generating unit 16b, the vectorization processing unit 16c, the hierarchical knowledge graph data updating unit 16d, the hierarchical knowledge graph data managing unit 16e, the search request receiving processing unit 16f, the hierarchical knowledge graph data search processing unit 16g, the vector search processing unit 16h, the difference analysis processing unit 16i, and the search result data sending unit 16j shown in FIG. 2 is stored in the nonvolatile memory 84 or the like. When the medical information analysis program is loaded and executed by the CPU 81, a medical information analysis process corresponding to an anonymized data receiving unit 16a, a label generating unit 16b, a vectorization processing unit 16c, a hierarchical knowledge graph data updating unit 16d, a hierarchical knowledge graph data managing unit 16e, a search request receiving processing unit 16f, a hierarchical knowledge graph data search processing unit 16g, a vector search processing unit 16h, a difference analysis processing unit 16i, and a search result data sending unit 16j is generated.

[0085] In the above embodiment, the doctor terminal 30 is described as being connected to the cloud management server 10 via the local server 20, but the doctor terminal 30 may be directly connected to the cloud management server 10 via the network N, and once the doctor terminal 30 receives search result data from the cloud management server 10, the doctor terminal 30 may transmit the search result data to the local server 20.

[0086] The components illustrated in the above embodiments are merely functional schematics and are not necessarily physically configured as shown. In other words, the distribution and integration of each device is not limited to the illustrated configuration, and all or part of the devices can be functionally or physically distributed and integrated in any unit depending on various loads, usage conditions, etc. [Industrial Applicability]

[0087] The medical institution analysis system, management device, medical institution analysis method, and medical institution analysis program according to the present invention are suitable for efficiently conducting in-depth qualitative evaluation of the diagnosis content by doctors and others at medical institutions. [Explanation of symbols]

[0088] 1, 2 Medical institutions 10 Cloud Management Server 11 Input section 12 Display section 13 Communication I / F section 15 Storage section 15a Anonymized Data 15b Label Data 15c A Hospital Hierarchical Knowledge Graph Data 15d B Hospital Hierarchical Knowledge Graph Data 15e XX Association ○○ Hospital Hierarchical Knowledge Graph Data 16 Control Unit 16a Anonymized data receiving unit 16b Level Generation Section 16c Vectorization processing section 16d Hierarchical knowledge graph data update unit 16e Hierarchical Knowledge Graph Data Management Department 16f Search request reception processing unit 16g Hierarchical knowledge graph data search processing unit 16h Vector search processing section 16i Differential analysis processing section 16j Search result data transmission section 20 Local Server 21 Input section 22 Display section 23 Communication I / F section 25 Memory section 25a Medical Information Data 25b Anonymized Data 25c Search result data 26 Control Unit 26a Medical information data receiving unit 26b Anonymization processing unit 26c Anonymized data transmission unit 26d Search request data transmission unit 26e Search result data receiving unit 26f Display control data transmission unit 30 Doctor's terminal 81 CPU 82 ROM 83 RAM 84 Non-volatile memory 85 Bus

Claims

1. A medical institution analysis system having a management device that manages medical institution information including medical images captured by modality devices installed in a medical institution and diagnostic reports of the medical images, The management device a label information extraction means for extracting label information consisting of qualitative information such as a location and a disease name and quantitative information such as a size of a detection target object for each input item of a medical institution's medical department, a test request, a finding, and a diagnosis from text included in the plurality of medical institution information; vector information generating means for vectorizing each of the plurality of pieces of label information to generate a plurality of pieces of vector information; a graph network generation means for generating a graph network of medical institutions based on the plurality of vector information; a management means for managing a graph network of multiple medical institutions; a differential analysis means for performing a differential analysis of the graph network of a predetermined medical institution included in the plurality of medical institutions with the graph networks of other medical institutions; A medical institution analysis system comprising:

2. The graph network is The medical institution analysis system according to claim 1 , wherein the medical institution is a graph network in which a subgraph network for each medical department that constitutes the medical institution is associated with the medical institution.

3. The label information extraction means The medical institution analysis system according to claim 1, characterized in that at least the location, disease name, certainty of the disease name, and size of the detected object are extracted as the label information for the medical institution information from the text contained in the medical institution information.

4. The label information extraction means 4. The medical institution analysis system according to claim 3, wherein the label information is extracted for each input item of test request, findings, and diagnosis.

5. The label information extraction means The medical institution analysis system according to claim 1, further extracting at least one or more of the type of examination equipment, the presence or absence of contrast agent, and the imaging conditions from text accompanying medical images contained in a plurality of medical institution information as the label information for the medical institution information.

6. The label information extraction means The medical institution analysis system according to claim 1, further extracting at least one or more of patient attributes, clinical background, and progress records from text contained in a plurality of medical institution information as the label information for the medical institution information.

7. The vector information generating means The medical institution analysis system according to claim 1 , wherein the plurality of pieces of label information are vectorized using natural language processing technology to generate a plurality of pieces of vector information.

8. The natural language processing technology is 8. The medical institution analysis system according to claim 7, wherein the predetermined large-scale language model is used.

9. The graph network generation means The medical institution analysis system of claim 1, wherein a graph network is generated in which one or more of the plurality of vector information or the body part, disease name, test type, and imaging condition contained in the text corresponding to the vector information are used as nodes, and the relationships between the nodes are expressed as edges.

10. The differential analysis means The medical institution analysis system described in claim 1, characterized in that when a request for differential analysis of graph networks of a first medical institution and a second medical institution is received, a graph network of the first medical institution is generated that makes it possible to distinguish the difference between a first graph network corresponding to the first medical institution and a second graph network corresponding to the second medical institution.

11. A management device that manages medical institution information including medical images captured by modality devices installed in a medical institution and diagnostic reports of the medical images, a label information extraction means for extracting label information consisting of qualitative information such as a location and a disease name and quantitative information such as a size of a detection target object for each input item of a medical institution's medical department, a test request, a finding, and a diagnosis from text included in the plurality of medical institution information; vector information generating means for vectorizing each of the plurality of pieces of label information to generate a plurality of pieces of vector information; a graph network generation means for generating a graph network of medical institutions based on the plurality of vector information; a management means for managing a graph network of multiple medical institutions; a differential analysis means for performing a differential analysis of the graph network of a predetermined medical institution included in the plurality of medical institutions with the graph networks of other medical institutions; A management device comprising:

12. A medical institution analysis method in a medical institution analysis system having a management device that manages medical institution information including medical images captured by modality equipment installed in a medical institution and diagnostic reports of the medical images, comprising: a label information extraction step in which the management device extracts label information consisting of qualitative information such as a location and a disease name and quantitative information such as a size of a detection target object for each input item of a medical institution's medical department, a test request, a finding, and a diagnosis from text included in the plurality of pieces of medical institution information; a vector information generating step in which the management device vectorizes each of the plurality of pieces of label information to generate a plurality of pieces of vector information; a graph network generation step in which the management device generates a graph network of medical institutions based on the plurality of vector information; a management step in which the management device manages graph networks of multiple medical institutions; a difference analysis step in which the management device performs a difference analysis between the graph network of a predetermined medical institution included in the plurality of medical institutions and the graph networks of other medical institutions; A medical institution analysis method comprising:

13. A medical institution analysis program executed in a management device that manages medical institution information including medical images captured by modality devices installed in a medical institution and diagnostic reports of the medical images, a label information extraction step of extracting label information consisting of qualitative information such as a location and a disease name and quantitative information such as a size of a detection target object for each input item of a medical institution's medical department, a test request, a finding, and a diagnosis from text included in the plurality of pieces of medical institution information; a vector information generating step of generating a plurality of pieces of vector information by vectorizing each of the plurality of pieces of label information; a graph network generation step of generating a graph network of medical institutions based on the plurality of vector information; Management procedures for managing a graph network of multiple medical institutions; a differential analysis step of performing a differential analysis of the graph network of a predetermined medical institution included in the plurality of medical institutions with the graph networks of other medical institutions; A medical institution analysis program characterized by causing a computer to execute the above.

Images