Data processing device, data processing method, and data processing system

The data processing device and system address the limitations of existing technologies by setting evaluation groups and evaluating results for accuracy, facilitating precise patient identification through comprehensive search conditions.

WO2026239755A1PCT designated stage Publication Date: 2026-11-19HITACHI HIGH TECH CORP
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
PCT/JP2026/011899
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2025-05-15
Filing Date
2026-03-25
Publication Date
2026-11-19

AI Technical Summary

Technical Problem

Existing technologies fail to perform specific evaluation group setting, search for an optimal evaluation value calculation formula, or process to extract valid data from search results, limiting the accuracy of identifying patients likely to have a specific disease.

Method used

A data processing device and system that sets up multiple evaluation groups, applies search conditions based on medical data, and evaluates results for accuracy, enabling comprehensive patient extraction with desired accuracy.

Benefits of technology

Enables automatic and accurate identification of patients likely to have a specific disease by creating search conditions using multiple test and treatment-related information, improving the precision of patient extraction.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure JP2026011899_19112026_PF_FP_ABST
    Figure JP2026011899_19112026_PF_FP_ABST
Patent Text Reader

Abstract

The present invention enables search condition creation and patient extraction that achieve desired extraction accuracy. A calculation unit sets, from among a first evaluation group, a second evaluation group, and a healthy group, two or more evaluation groups including at least the first evaluation group. The calculation unit sets, with the two or more evaluation groups being regarded as objects to be searched, a plurality of search conditions in which at least two or more pieces of data among examination result data, data obtained in medical care, and data relating to attributes of a patient are combined, and performs a search of the two or more evaluation groups serving as objects to be searched. The calculation unit identifies data for evaluation from search result data and a process in a designated medical cycle, evaluates the accuracy of a search result, and selectably presents an optimal search condition from among a plurality of search conditions on the basis of the result of the accuracy evaluation.
Need to check novelty before this filing date? Find Prior Art

Description

Data processing device, data processing method, and data processing system

[0001] The present disclosure relates to technologies such as a data processing device, a data processing method, and a data processing system.

[0002] To provide high-quality medical care, various efforts are required, including fostering excellent medical personnel, developing medical devices that realize highly accurate diagnosis, establishing regional medical care systems, and developing guidelines and implementing training for safety management related to medical care. In addition, the development and dissemination of business process reform solution services, Personal Health Records (PHR), and clinical decision support software utilizing Information and Communication Technology (ICT) and artificial intelligence technology are being promoted to further improve the quality and stability of medical care.

[0003] In recent years, further development of medical care through the utilization of information communication technology and artificial intelligence technology is expected. One of the application fields with high affinity for these technologies is clinical testing. Clinical testing is a type of test mainly used for disease diagnosis, treatment, and medical checkups.

[0004] Against this background, various technologies have been developed. For example, a known technology stores a data catalog that stores patient medical data related to cancer and information related to urinary metabolites, searches for a combination of at least one of test conditions related to medical information and urinary metabolite information based on information input via an input device, and outputs an optimal combination of test conditions related to medical information based on a predetermined evaluation value (refer to Patent Document 1 below).

[0005] Japanese Unexamined Patent Application Publication No. 2021-128115

[0006] The technology of Patent Document 1 described above does not perform specific evaluation group setting, search for an optimal evaluation value calculation formula, or processing to extract valid data from search results.

[0007] The purpose of this disclosure is to provide a technology that enables the creation of search conditions that achieve a desired extraction accuracy and patient extraction by automatically and comprehensively searching for patients who are highly likely to have a specific disease, using search conditions based on information related to multiple tests and treatments, information related to patient attributes, and the resulting values.

[0008] A typical embodiment of this disclosure has the configuration shown below. The data processing device of one embodiment includes a processor that executes a program and a storage device that stores the program, and analyzes medical data for each examination item. The processor performs an evaluation group setting process to set up at least two evaluation groups, including at least the first evaluation group, from a first evaluation group consisting of a plurality of patients suffering from or exhibiting a first disease, a second evaluation group consisting of a plurality of patients suffering from or exhibiting a second disease, and a third evaluation group consisting of a plurality of patients suffering from or exhibiting the same disease as the first or second evaluation group, and at least two data from a plurality of examination result data, data obtained in medical treatment, and data relating to the attributes of the patients, using the two or more evaluation groups set up by the evaluation group setting process as search targets. The system performs the following steps: a condition setting process for setting multiple search conditions; a search execution process for applying the multiple search conditions set by the condition setting process to each of the two or more evaluation groups that are the target of the search; a test result evaluation preparation process for identifying evaluation data from the search result data extracted as a result of the search process and the process in a specified medical cycle; and a search result analysis process for performing an accuracy evaluation of the search results on the evaluation data identified by the test result evaluation preparation process, and presenting the optimal search condition from among the multiple search conditions based on the results of the accuracy evaluation.

[0009] One embodiment of the data processing device includes a processor that executes a program and a storage device that stores the program, and analyzes medical data for each examination item. The data processing method of the data processing device includes the steps of the processor setting two or more evaluation groups, including at least the first evaluation group, from a first evaluation group consisting of a plurality of patients suffering from or exhibiting a first disease, a second evaluation group consisting of a plurality of patients suffering from or exhibiting a second disease, and a third evaluation group consisting of a plurality of patients suffering from or exhibiting the same disease as the first or second evaluation group, and the processor searching for at least two of the set two or more evaluation groups, including a plurality of examination result data, data obtained in medical treatment, and data relating to the attributes of the patients. The process includes: setting multiple search conditions by combining the above data; the processor applying the set multiple search conditions to each of the two or more evaluation groups to be searched; the processor identifying evaluation data from the search result data extracted as a result of the search process and the processes in a specified medical cycle; and the processor performing an accuracy evaluation of the search results on the identified evaluation data and, based on the results of the accuracy evaluation, presenting the optimal search condition from among the multiple search conditions in a selectable manner.

[0010] A data processing system according to one embodiment includes a data processing device that analyzes medical data for each test item, and a medical system that is connected to medical devices including an analyzer and a diagnostic device for analyzing specimens, and stores medical data acquired by the medical devices. The processor includes an evaluation group setting process that sets up at least two evaluation groups, including at least the first evaluation group, from a first evaluation group consisting of multiple patients suffering from or exhibiting a first disease, a second evaluation group consisting of multiple patients suffering from or exhibiting a second disease, and a third evaluation group consisting of multiple patients suffering from or exhibiting the same disease as the first or second evaluation group, and at least two data from multiple test result data, data obtained in medical treatment, and data relating to the attributes of the patients, using the two or more evaluation groups set up by the evaluation group setting process as search targets. The processor performs the following: a condition setting process for setting multiple search conditions; a search execution process for applying the multiple search conditions set by the condition setting process to each of the two or more evaluation groups to be searched; a test result evaluation preparation process for identifying evaluation data from the search result data extracted as a result of the search process and the process in a specified medical cycle; and a search result analysis process for performing an accuracy evaluation of the search results on the evaluation data identified by the test result evaluation preparation process, and presenting the optimal search condition from among the multiple search conditions in a selectable manner based on the results of the accuracy evaluation. The processor acquires the medical data from the medical system.

[0011] According to a typical embodiment of this disclosure, a data analysis device or the like automatically and comprehensively searches for patients who are highly likely to have a specific disease using multiple test and treatment-related information, patient attribute-related information, and result values, enabling the creation of search conditions and patient extraction that achieve a desired extraction accuracy. Other issues, configurations, and effects will be shown in the embodiments for carrying out the invention.

[0012] Figure 1A is a diagram showing an example of the configuration of a medical facility where the data processing device of the first embodiment is installed. Figure 1B is a diagram showing an example of data included in the medical data of the first embodiment. Figure 2 is a block diagram showing an example of the functional configuration of the data processing device of the first embodiment. Figure 3 is a diagram showing an example of the functional configuration of the test result evaluation preparation unit of the first embodiment. Figure 4 is a diagram showing an example of the functional configuration of the search result analysis unit of the first embodiment. Figure 5 is a diagram showing an example of test result data of the first embodiment. Figure 6 is a diagram showing an example of drug prescription data of the first embodiment. Figure 7 is a diagram showing an example of disease name data of the first embodiment. Figure 8 is a diagram explaining the definition of the reference range in the test items of the specimen test of the first embodiment and the test values ​​that the disease group can take. Figure 9 is a flowchart showing an example of data analysis processing performed by the calculation unit of the first embodiment. Figure 10 is a diagram showing a display screen, which is an example of the display of a display connected to the output unit of the first embodiment. Figure 11 is a diagram showing a display screen, which is an example of the display of a display connected to the output unit of the first embodiment. Figure 12 is a diagram showing an example of the functional configuration of the data processing device of the second embodiment. Figure 13 shows an example of the ROC curve and composite evaluation plot under multiple conditions in the third embodiment. Figure 14 shows an example of the composite evaluation plot displayed on the display screen of a display connected to the output unit of the third embodiment. Figure 15 shows a display screen, which is an example of the display on a display connected to the output unit of the third embodiment. Figure 16 shows an example of the configuration of a medical facility where a modified data processing device is installed.

[0013] The embodiments of this disclosure will be described in detail below with reference to the drawings. In the drawings, the same parts are generally denoted by the same reference numerals, and repeated descriptions are omitted. In the drawings, the representation of components may not show their actual location, size, shape, extent, etc., in order to facilitate understanding of the invention.

[0014] [First Embodiment] <Overview of Data Processing Device> First, an overview of the processing of the data processing device of the first embodiment will be described. The data processing device sets and extracts data from evaluation groups and patient data included in those evaluation groups that are in a state suitable for evaluation. Next, the data processing device automatically searches for search conditions based on multiple pieces of information related to examinations and medical treatments, patient attribute information, and result values ​​that result in the desired extraction accuracy for the set and extracted data. Then, the data processing device creates appropriate search conditions and extracts patients. The data processing device will be described in more detail below.

[0015] <Analysis of Medical Data> Next, we will explain the analysis of medical data. The medical data accumulated within medical facilities is time-series data that differs for each patient, including the type of disease they have, whether they are hospitalized or outpatient, the status of medical procedures performed (such as treatment or observation), and whether or not it is a health checkup. This medical data is managed by various systems, such as clinical laboratory systems, pathology laboratory systems, drug prescription systems, and DPC (Diagnosis Procedure Combination) systems.

[0016] Ideally, when analyzing medical data, all medical data should be structured. This is because it facilitates the extraction and analysis of data for each individual patient. However, medical data also includes a large amount of unstructured data, such as information entered in natural language in the SOAP section of electronic medical records, and diagnostic images taken by imaging equipment. Therefore, to understand when, what physical condition, what tests and treatments a patient received, and how their condition progressed, including recurrences and the onset of other diseases, it is necessary to integrate and interpret the data managed by each system.

[0017] For example, one type of medical data analysis involves determining a diagnostic threshold—a test value that indicates the presence of a specific disease or condition. In this analysis, a case-control study is typically used for each hospital or department, examining the distribution of test values ​​in diseased and non-disease groups for each highly specific test item, and setting an appropriate threshold based on sensitivity and specificity. When conducting such a case-control study, the data processing device first acquires the necessary data from each system, depending on the patient's medical status and the test items used. Then, it integrates and / or interprets the data to identify the data needed to determine which test items to examine for the distribution of test values ​​in diseased and non-disease groups. Subsequently, the data processing device uses the identified data to examine the distribution of test values ​​in diseased and non-disease groups and set an appropriate diagnostic threshold. It should be noted that the medical data may be considered after acquisition from each system.

[0018] <Hardware Configuration of Data Processing Device> Next, the data processing device will be described. In this first embodiment, the data processing device is a data processing device 110 installed within the medical facility 100. Figure 1A is a diagram showing an example of the configuration of the medical facility 100 in which the data processing device 110 is installed. In this first embodiment, the data processing device 110 will be described as being installed within the medical facility 100, but the data processing device 110 may also be installed outside the medical facility 100. In this case, the computer inside the medical facility 100 should be configured to be connected to a network that can communicate with the data processing device 110 which is set up on a cloud outside the medical facility 100. Alternatively, the data to be used may be retrieved from inside the medical facility 100 and stored in the data processing device 110 installed outside the medical facility 100 via a network or a storage medium such as ROM (Read Only Memory), RAM (Random Access Memory), HDD (Hard Disk Drive), SSD (Solid State Drive), and flash memory.

[0019] As shown in Figure 1A, the medical facility 100 is equipped with, for example, a data processing device 110, a storage device 120, an analyzer and medical equipment 102, and an in-facility system 103. The data processing device 110, the storage device 120, and the in-facility system 103 are each connected to each other via an in-facility network 101. The analyzer and medical equipment 102 are connected to the in-facility system 103.

[0020] The data processing device 110 includes an arithmetic unit 111, a storage unit 112, a communication unit 113, and an input / output unit 114. The arithmetic unit 111, the storage unit 112, the communication unit 113, and the input / output unit 114 are communicated together by a bus 115.

[0021] The arithmetic unit 111 (processor) controls the entire data processing device 110. The storage unit 112 is the work area for the arithmetic unit 111. The storage unit 112 is also a non-temporary or temporary recording medium that stores various types of programs 112a and data. The storage unit 112 is, for example, ROM, RAM, HDD, SSD, and flash memory. The input / output unit 114 receives input and output from the input unit 116 and the output unit 117. Data is input to the input unit 116. Devices connected to the input unit 116 are, for example, a keyboard, mouse, touch panel, numeric keypad, and scanner. The output unit 117 outputs data. Devices connected to the output unit 117 are, for example, a display and printer. The communication unit 113 is connected to the facility network 101.

[0022] The storage device 120 is a storage device composed of, for example, multiple HDDs (Hard Disk Drives) and multiple SSDs (Solid State Drives). The storage device 120 stores medical data 121 and hospital information data 122.

[0023] Medical data 121 is data relating to medical treatment performed within the medical facility 100. Figure 1B shows an example of the data included in medical data 121. Medical data 121 includes, for example, disease name data 121a, specimen test data 121b, prescription data 121c, health checkup data 121d, medical history data 121e, claims data 121f, physiological test data 121g, pathology data 121h, radiographic image data 121i, and electronic medical record data 121j. The data included in medical data 121 is not limited to these.

[0024] Disease name data 121a is data indicating the name of the disease. Specimen test data 121b is data related to the results of specimen tests. Prescription data 121c is data related to the prescribed contents. Health checkup data 121d is data related to the results of health checkups. Medical history data 121e is data related to past medical history. Claim data 121f is data related to claims. Physiological test data 121g is data related to physiological tests. Pathology data 121h is data related to pathology. Radiological image data 121i is data related to images obtained by radiation exposure. Electronic medical record data 121j is data related to information entered into the electronic medical record.

[0025] Each piece of data included in the medical data 121 can be used to collect desired data for a particular patient, for example, by using an identification number that identifies the patient. For example, by collecting drug prescription data 503 (see Figure 6) related to the prescription of drugs described later, and disease name data 504 (see Figure 7) related to patients who have disease A (first disease), test result data 502 (see Figure 5) related to the identification number of a patient suffering from a specific disease A (first disease) can be collected.

[0026] Hospital information data 122 is data relating to information that is not classified as medical procedures performed within the medical facility 100. For example, it includes logistics procurement information, appointment booking information, and drug inventory information. However, the data included in hospital information data 122 is not limited to these.

[0027] The in-facility systems 103 include, for example, a clinical testing system, a pathology testing system, a drug prescription system, and a DPC system. Each of these systems is connected to an analyzer and a medical device 102. Each system acquires data measured by the connected analyzer, diagnostic device, and medical device 102. In this first embodiment, the data acquired by each system is output from each system to a storage device 120 and stored in the storage device 120 as medical data 121. In this first embodiment, the case where medical data 121 is stored in the storage device 120 is described, but the data processing device 110 may collect the necessary data from the in-facility systems 103 at the necessary timing. Furthermore, there may be multiple storage devices 120 within the medical facility 100, and medical data may be stored in a distributed manner. The multiple storage devices 120 may include backup storage devices.

[0028] <Functional Configuration of Data Processing Device 110> Next, the functional configuration of the data processing device 110 will be described. Figure 2 is a block diagram showing an example of the functional configuration of the data processing device 110. As shown in Figure 2, the calculation unit 111 of the data processing device 110 includes a condition search unit 200 and an evaluation group setting unit 201 as functional blocks. The condition search unit 200 has a condition setting unit 202, a search execution unit 203, an inspection result evaluation preparation unit 204, and a search result analysis unit 205. In Figure 2, the storage unit 112 and the input / output unit 114 are also shown, but the communication unit 113 is not shown.

[0029] Figure 3 shows an example of the functional configuration of the inspection result evaluation preparation unit 204. As shown in Figure 3, the inspection result evaluation preparation unit 204 has a state setting unit 301 and an inspection result organization unit 302. Figure 4 shows an example of the functional configuration of the search result analysis unit 205. As shown in Figure 4, the search result analysis unit 205 has a condition evaluation unit 401 and a condition presentation unit 402.

[0030] <Data Analysis Processing> The following explanation will use the data analysis processing performed by the data processing device 110 as an example to describe how to create appropriate search conditions and extract patients when searching for patients with disease A during diagnosis.

[0031] The evaluation group setting unit 201 sets up two or more evaluation groups, including at least the first evaluation group, from among the first evaluation group consisting of multiple patients suffering from or exhibiting the first disease, the second evaluation group consisting of multiple patients suffering from or exhibiting the second disease, and the third evaluation group consisting of multiple patients suffering from or not exhibiting the symptoms of any disease. The first and second diseases are different diseases. Patients suffering from or not exhibiting the symptoms of any disease are considered healthy. Patients who do not suffer from the first disease may be considered healthy and included in the third evaluation group. Evaluation groups may be set up that include not only healthy individuals but also considered patients. Each evaluation group may be selected by the user from a pre-created list, or the user may create one on the spot by inputting conditions. Evaluation groups may also be set up based on symptoms rather than diseases.

[0032] For example, when creating an evaluation group N, which is a group of patients suffering from disease A (first disease), the evaluation group setting unit 201 designates multiple patients who have a description of disease A in the medical data 121 stored in the storage device 120 as evaluation group N.

[0033] Since medical data 121 is data of patients who visited a medical institution for some reason such as poor health, medical data 121 is basically data of patients suffering from or exhibiting a disease. For this reason, the evaluation group setting unit 201 may create a healthy group consisting of presumed healthy individuals, for example, by designating patients who do not have disease A as evaluation group M. Also, when the evaluation group setting unit 201 sets an evaluation group targeting presumed healthy individuals, it may designate patients in medical departments where the probability of visiting when experiencing symptoms of disease A is low as evaluation group M. Furthermore, the evaluation group setting unit 201 may also use the exclusion from evaluation group M patients who have visited medical departments where many patients suffering from other diseases that are likely to occur concurrently with disease A visit.

[0034] Information regarding each of the created evaluation groups N and M is stored, for example, in the storage unit 112. In the case of evaluation group M, the information regarding each created evaluation group includes, at a minimum, the patient identification number, the disease name listed in the medical data 121, the date and time the disease name was recorded, and information indicating the medical department in which the disease name was recorded, for patients included in evaluation group M. If evaluation group M consists of healthy individuals or deemed healthy individuals, the disease name listed in the medical data 121 may be left blank. In the case of evaluation group N, the information includes, at a minimum, the patient identification number, the disease name listed in the medical data 121, the date and time the disease name was recorded, and data including the medical department in which the disease name was recorded, for patients included in evaluation group N. This data is obtained from the medical data 121 in the storage device 120 and stored in the storage unit 112.

[0035] The evaluation group may be created using data that includes information such as the patient identification number, the disease name recorded in the medical data, the date and time the disease name was recorded, and the clinical department in which the disease name was recorded, as shown in the disease name data 504 (see Figure 7) included in the medical data 121.

[0036] However, evaluation group N is strictly speaking a group of patients who are likely to have disease A. Therefore, the accurate determination of who is included in evaluation group N is left to the physician. Similarly, evaluation group M is strictly speaking a group of patients who are likely to be healthy. Therefore, similarly, the accurate determination of who is included in evaluation group M is left to the physician.

[0037] Furthermore, the data for each evaluation group N and M may be stored in the storage unit 112 as described above, or in the storage device 120. Also, for example, if each evaluation group N and M has been created in advance, the data for these evaluation groups N and M may also be stored in the storage unit 112 or in the storage device 120. In this first embodiment, the explanation is given in the case where an evaluation group N consisting of multiple patients suffering from disease A and an evaluation group M consisting of multiple patients who are not considered to have disease A are set up, but an evaluation group consisting of multiple patients suffering from disease A and an evaluation group consisting of multiple patients suffering from disease B (second disease) different from disease A may also be set up. Also, an evaluation group consisting of multiple patients suffering from disease A and an evaluation group consisting of multiple patients who do not have any disease or exhibit any pathological condition may also be set up. Furthermore, in the creation of any evaluation group, the data may not be limited to patients only, but may also be limited to examinees only, or may include both patients and examinees.

[0038] The condition setting unit 202 sets multiple search conditions that combine at least two data from multiple test result data, data obtained from medical treatment, and data relating to patient attributes, using two or more evaluation groups (for example, evaluation groups N and M) set by the evaluation group setting process of the evaluation group setting unit 201 as search targets. Here, "at least two data" includes not only the case where there are two or more different types of data from multiple test result data, data obtained from medical treatment, and data relating to patient attributes, but also the case where there are two or more data from multiple test result data. Furthermore, "at least two data" includes the case where, if there are multiple data obtained from medical treatment, there are two or more of these data, and if there are multiple data relating to patient attributes, there are two or more of these data.

[0039] For example, the condition setting unit 202 sets search conditions for the purpose of extracting patients with disease A from evaluation group N, which consists of patients with disease A, and evaluation group M, which is a group of deemed healthy individuals. As a more specific example, the condition setting unit 202 sets search conditions for characteristic test items X and Y when a patient has disease A, such as the test result for test item X being 1.0 or higher and the test result for test item Y being less than 10.0. These setting values ​​may be determined, for example, by referring to guidelines for disease A, or based on the experience of users such as doctors and medical technologists (hereinafter referred to as "users").

[0040] Furthermore, the condition setting unit 202 may, for example, set the search range for test item X to ±0.1 with an increment of 10, and the search range for test item Y to ±1.0 with an increment of 10, in order to find appropriate search conditions. In addition, the condition setting unit 202 may prepare 100 patterns of search conditions by combining patterns of the two test items X and Y. Furthermore, the condition setting unit 202 may set the search range and the increment of the search range for each test item to 10% of the value of the search condition set initially, or the user may set them each time an analysis is performed. In addition, the condition setting unit 202 may generate 100 patterns of search conditions randomly within the search range given to test item X and test item Y, rather than using an increment. Furthermore, the condition setting unit 202 may set the patient attributes using data such as age, gender, and medical history.

[0041] The search execution unit 203 applies one or more search conditions set by the condition setting process of the condition setting unit 202 and performs a search for each of the one or more evaluation groups to be searched. In this embodiment, a search is performed for evaluation group M and evaluation group N, respectively. The search result data obtained for each search condition (for example, search result data 501 (see Figure 5)) is stored in, for example, the storage unit 112. The search result data may also be stored in the storage device 120.

[0042] Next, the processing of the test result evaluation preparation unit 204 will be described. The test result evaluation preparation unit 204 identifies data for evaluation from the search result data generated as a result of the search processing, the process in the specified medical cycle, and the information related to the evaluation group set by the evaluation group setting unit 201. As shown in Figure 3, the test result evaluation preparation unit 204 includes a state setting unit 301 and a test result organizing unit 302. For this reason, the processing will be described in the order of the state setting unit 301 and then the test result organizing unit 302.

[0043] The state setting unit 301 specifies a process in the medical cycle for the search result data generated by the search execution unit 203. Here, the medical cycle encompasses a series of medical processes such as the first consultation for a certain disease, the state before therapeutic intervention, and the state during treatment. In the first embodiment, a case of realizing creation of appropriate search conditions and patient extraction when searching for patients with disease A at the time of diagnosis will be described as an example, so the state setting unit 301 sets the time of diagnosis as the medical cycle.

[0044] As described above, the medical data 121 in the medical facility 100 stored in the storage device 120 is used for analysis. To collect the medical data 121 from the in-facility systems 103, the medical data 121 is collected from each of the in-facility systems 103 respectively. Therefore, in order to analyze the medical data 121, a step of integrating and interpreting each piece of medical data 121 is required. In this case, the test result organizing unit 302 understands the relationship among the search result data generated as a result of the search processing, the process in the specified medical cycle, and the information related to the evaluation group set by the evaluation group setting unit 201. Here, the search result data includes test results in various patient states performed over a long period of time at the medical facility 100, such as at the time of the first consultation, during disease follow-up, during treatment follow-up, and post-treatment follow-up. A specific example of the search result data will be described with reference to Figure 5 to be described later.

[0045] The test result compilation unit 302 uses the patient's medical cycle process set in the status setting unit 301, the patient set based on the evaluation group information set in the evaluation group setting unit 201, and the search result data 501 (see Figure 5) generated by the search execution unit 203 to determine which of the search result data should be used as appropriate evaluation data to search for patients with disease A at the time of diagnosis.

[0046] Alternatively, the evaluation group setting unit 201 may acquire drug prescription data 503 (see Figure 6) extracted from prescription data 121c from the storage device 120, and use the drug prescription data 503 generated from prescription data 121c included in the medical data 121, in addition to the disease name (disease name data 504: see Figure 7), to identify the evaluation data. For example, by using whether or not drug A2, which is characteristically prescribed in the case of disease A, is prescribed, it becomes possible to extract patients with a high probability of truly suffering from disease A, thereby reducing the possibility of patients who do not suffer from disease A being included in evaluation group N.

[0047] In addition, when collating the disease name data 504 and the drug prescription data 503, there may be a discrepancy in dates. The disease name data 504 is input with a disease name by a physician during the course of medical care, for example, when Disease A is suspected, when a test is performed to check whether the patient has Disease A, or when the patient is diagnosed with Disease A. For this reason, the disease name data 504 may contain a mixture of entries input as suspected disease names and entries input as confirmed disease names. After the disease name is input by the physician, the prescription of Drug A2 is performed if the patient truly has Disease A, or when the patient's condition worsens after observation. Although Disease A is input as a disease name, if the patient does not actually suffer from Disease A, the evaluation group setting unit 201 can exclude disease names input as suspected disease names, erroneously input disease names, and the like from the disease name data 504 of patients including Disease A, by collating with the drug prescription data 503. However, cases where the symptoms do not worsen and no drug treatment is administered will also be excluded. The identification of search result data corresponding to the process in the medical care cycle of the target patient specified by the state setting unit 301 can also be achieved by collating the disease name data 504 and the drug prescription data 503. Furthermore, in the disease name data 504, a plurality of disease names may be generated for a single patient, for example, when the disease name of Disease A is also input in another medical department.

[0048] Next, data used by the test result organizing unit 302 will be described with reference to FIGS. 5 to 7. FIG. 5 is a diagram illustrating an example of test result data 502 of the evaluation group N having Disease A. FIG. 6 is a diagram illustrating an example of drug prescription data 503 for patients included in the evaluation group N. FIG. 7 is a diagram illustrating an example of disease name data 504 related to patients including Disease A.

[0049] As shown in Figure 5, the test result data 502 associates the patient's identification number "P_No" with the test number "R_No", the date the test was performed "Date", the result value of test item X "Lab_X", the result value of test item Y "Lab_Y", etc. As shown in Figure 6, the drug prescription data 503 associates the patient's identification number "P_No" with the drug name "Drug_name", the date the prescription was issued "Date", the prescribed amount "Amount", the prescription period "Duration", etc. As shown in Figure 7, the disease name data 504 for patients with disease A associates the patient's identification number "P_No" with the date the disease was entered "Date", the disease name "Disease_Name", the department in which the disease was entered "Department", etc.

[0050] Referring to the test result data 502, drug prescription data 503, and disease name data 504 shown in Figures 5 to 7, it can be seen that patient P001, whose patient identification number P_No is "P001", underwent an examination on January 1, 2024, and the test results showed that the test result for test item X was 1.0 or higher and the test result for test item Y was less than 10, and that the patient was diagnosed with disease A in clinical department A1, disease A was entered as the disease name, and a one-month supply of drug A2 was prescribed.

[0051] Furthermore, on the same day, patient P001 was diagnosed by department Z1 with disease Z, which is different from disease A, and disease Z was entered as the disease name. However, no medication related to disease Z was prescribed. Subsequently, drug A2 was prescribed for two months, with the dosage being changed approximately every month. During that time, the results of the second test met the search criteria, but the results of the third test did not. Therefore, the results of the third test were not extracted. In other words, it can be inferred that patient P001 met the search criteria at the time of diagnosis, but at the time of the third test, the patient's condition had improved, and therefore the search criteria were no longer met. If the time of diagnosis was selected in the status setting unit 301, the results of the first test were extracted by the search, so it can be considered that the search criteria used for the search were appropriate.

[0052] On the other hand, patient P010, whose patient identification number P_No is P010, underwent an examination on January 10, 2024. The results of examination item X were 1.0 or higher and the results of examination item Y were less than 10. The patient was diagnosed with disease A in clinical department Y1, disease A was entered as the disease name, and medication A2 was prescribed three days later. Although there is a difference between the date the disease name was entered and the date the prescription was issued, the examination results on January 10, 2024, are data from before the prescription was started. Therefore, it is reasonable to consider that these examination results were extracted when the time of diagnosis was selected in the status setting unit 301.

[0053] Based on the above, in the examples shown in Figures 5 to 7, the evaluation data among the test result data 502 that will be subject to evaluation are the test results of patient P001 conducted on January 1, 2024, and the test results of patient P010 conducted on January 10, 2024.

[0054] For example, if the search criteria are set to a test result of 2.0 or higher for test item X and a test result of less than 10 for test item Y, evaluation data for patient P001 will be extracted, but evaluation data for patient P010 will not be extracted. In other words, the search criteria will not allow for the extraction of patient P010, who is considered to have a high probability of suffering from disease A. However, the examples shown in Figures 5 to 7 are merely illustrative, and in reality, there are test result data 502, drug prescription data 503, and disease name data 504 at various timings, medical departments, disease names, and prescription details.

[0055] Furthermore, the selection of the medical cycle in the state setting unit 301 may be set simultaneously with the selection of the evaluation group (evaluation group N) by the evaluation group setting unit 201, or it may be set after selecting the search result data generated by the search execution unit 203.

[0056] Furthermore, the processing related to the inspection result evaluation preparation unit 204, which has a state setting unit 301 and an inspection result organization unit 302, may be executed after or simultaneously with the processing related to the condition setting unit 202. In this case, the search execution unit 203 may perform search processing using evaluation data.

[0057] Next, the processing of the search result analysis unit 205 will be explained. The search result analysis unit 205 performs an accuracy evaluation of the search results on the evaluation data identified by the inspection result evaluation preparation processing of the inspection result evaluation preparation unit 204, and based on the results of the accuracy evaluation, presents the optimal search condition from among multiple search conditions so that it can be selected. As shown in Figure 4, the search result analysis unit 205 has a condition evaluation unit 401 and a condition presentation unit 402. Therefore, the processing of the condition evaluation unit 401 and the condition presentation unit 402 will be explained in that order.

[0058] The condition evaluation unit 401 includes at least sensitivity or specificity as an evaluation item for the evaluation data identified by the inspection result evaluation preparation processing of the inspection result evaluation preparation unit 204. In other words, the condition evaluation unit 401 performs an evaluation on the evaluation data identified by the inspection result organization unit 302 using an evaluation item that includes at least sensitivity or specificity. Through processing in the inspection result organization unit 302, the evaluation data includes only data that is appropriate for use in evaluation. That is, if a sample included in evaluation group N is included in the search results, and a sample included in evaluation group M is not included in the search results, it can be said that the search conditions are appropriate and the search results are the desired results.

[0059] Generally, for ease of reading, evaluation results are often presented in tables such as cross-tabulation tables, quadratic tables, or mixed matrices, which summarize two evaluation groups and the results of the identification process applied to those two evaluation groups. This method of summarizing evaluation results using tables may also be used for three or more evaluation groups.

[0060] According to this table, if a sample from evaluation group N is included in the search results, it will be classified as "True Positive (TP)," and if a sample from evaluation group M is not included in the search results, it will be classified as "True Negative (TN)." Generally, no matter how accurate the diagnostic test item, it is difficult to generate search conditions that result in only TP and / or only TN results.

[0061] Figure 8 is an explanatory diagram illustrating the definition of the reference range for test item X in specimen testing, and the possible test values ​​that a disease group may exhibit. In Figure 8, the vertical axis represents frequency (number of people), and the horizontal axis represents the test result value for test item X. Here, the reference range in specimen testing refers to the 95% median interval of the distribution of test values ​​measured from individuals selected from healthy individuals under certain conditions. This reference range can be shared among medical facilities if the measured values ​​are standardized. Here, standardization of measured values ​​means that there are no regional differences, and / or that evaluation criteria for external quality control are obtained. For example, for one test item, if a threshold of approximately the reference range is set, a condition can be set to extract 5% of the healthy group.

[0062] Furthermore, not all patients in a disease group will have test results exceeding the reference range. Depending on the characteristics of the disease and the characteristics of the patient, some test items may result in test results within the reference range. For this reason, no matter how strict the search conditions are or how many items are set, it is difficult to make all samples of the evaluation data TP or TN. In other words, if the accuracy evaluation index is based solely on what percentage of the total number of samples in the evaluation data the number of TPs and TNs represents, it ignores the False Negative (FN) cases where samples included in evaluation group N were not included in the search results, and the False Positive (FP) cases where samples included in evaluation group M were included in the search results. The condition evaluation unit 401 can prevent the omission of evaluation target data if the FN is low. The condition evaluation unit 401 can prevent the erroneous extraction of evaluation target data as diseased if the FP is low.

[0063] For evaluating the accuracy of search results, you may use sensitivity as shown in equation (1) below, specificity as shown in equation (2) below, accuracy as shown in equation (3) below, goodness of fit as shown in equation (4) below, or the F1 score as shown in equation (5) below. Alternatively, you may use evaluation methods used in ROC curves (Receiver operating characteristic) and AUC (Area Under the Curve), such as the Youden index, which is the sum of sensitivity and specificity minus 1. Furthermore, while ROC curves and AUC are calculated using sensitivity and specificity, you may also use PRROC curves or PR AUC calculated using sensitivity and goodness of fit.

[0064]

[0065]

[0066]

[0067]

[0068]

[0069] The condition presentation unit 402 visually presents evaluation results based on one or more search conditions, using the selected evaluation index. More specifically, based on the evaluation results performed by the condition evaluation unit 401, the condition presentation unit 402 presents the user with the option to select the most suitable search condition from among one or more search conditions. For example, the condition presentation unit 402 presents the top five evaluation results from multiple search conditions, ordered by the evaluation index selected by the user. Furthermore, the presented evaluation results do not necessarily have to be from multiple search conditions performed simultaneously. For example, the condition presentation unit 402 can present the top five evaluation results from a total of seven evaluation results, including the evaluation results from the condition evaluation unit 401 after performing a search with three search conditions on the same evaluation data, and the evaluation results from the condition evaluation unit 401 after performing a second search with four conditional expressions, ordered by the evaluation index selected by the user, ordered by the user. In addition, while the condition presentation unit 402 presents the top five evaluation results from multiple search conditions, ordered by the evaluation index selected by the user, there is no limit to the number of conditions, and the user may be allowed to specify the number.

[0070] Furthermore, the evaluation data identified by the test result compilation unit 302 is stored in the storage unit 112 along with the data used to generate the evaluation group and data indicating the process of the set medical cycle. This evaluation data may also be stored in the storage device 120. In this first embodiment, an example was shown in which disease name data 504, which includes disease A, and drug prescription data 503 are obtained from medical data 121 stored in the storage device 120. However, the data processing device 110 may be configured to obtain the data without connecting to the facility network 101 within the medical facility 100 by using a storage medium such as a DVD (Digital Versatile Disc) or a USB (Universal Serial Bus) memory.

[0071] <Example of Processing Procedure> Next, the processing procedure for performing data analysis by the data processing device 110 will be described. Figure 9 is a flowchart showing an example of data analysis processing performed by the calculation unit 111 of the data processing device 110. Details of the processing in each step ST110 to ST170 are as previously described.

[0072] As shown in Figure 9, the calculation unit 111 sets evaluation groups based on the target disease and pathological condition (ST110: Evaluation Group Setting Unit 201). Next, the calculation unit 111 sets search conditions using information related to examination items and / or patient attributes (ST120: Condition Setting Unit 202). Then, the calculation unit 111 performs a search for each evaluation group (ST130: Search Execution Unit 203). Furthermore, the calculation unit 111 sets the patient's condition to be evaluated for the search results (ST140: Condition Setting Unit 301). Then, the calculation unit 111 identifies search result data that matches the set condition and generates evaluation data (ST150: Examination Result Organization Unit 302). Next, the calculation unit 111 evaluates the search results from the evaluation data using evaluation indices such as sensitivity and specificity (ST160: Condition Evaluation Unit 401). Finally, the calculation unit 111 presents the evaluation results (ST170: Condition Presentation Unit 402).

[0073] As explained above, the data processing device 110 makes it possible to automatically and comprehensively search for patients who are highly likely to have a specific disease using information related to multiple tests and medical treatments, information related to patient attributes, and search conditions based on the result values, thereby creating search conditions and extracting patients that achieve the desired extraction accuracy.

[0074] <Example of display screen> Next, an example of a display screen shown on a display, which is an example of a device connected to the output unit 117 of the data processing device 110, will be explained with reference to Figures 10 and 11.

[0075] <<Example of the first display screen>> Figure 10 shows a display screen 800, which is an example of the display on a display connected to the output unit 117. The display screen 800 includes a mode selection field 801, a field for displaying configured evaluation groups 802, a field for selecting already created evaluation groups 803, a button 804 for transitioning to the detailed settings screen for evaluation groups, and an OK button 805.

[0076] The user operates the input unit 116 to select "Group Set" in the mode selection field 801 on the display screen 800. Next, the user presses the dropdown menu in the created group selection field 803 and selects the desired group from the created group selection fields. One or more groups may be selected. Alternatively, if no evaluation is required, the system may allow a process that only performs a search, such as skipping the evaluation. Alternatively, instead of using the created group selection field 803, the user presses the button 804 to transition to the detailed settings screen for the group, where they configure the target disease or condition, patient status, medical department, and drug prescription status on a separate screen, and then create the group. Once the input is complete, the user presses the OK button 805 to finish the settings. This reflects the information of the configured group in the configured group display field 802. Furthermore, the display screen 800 may display functions for setting not only two evaluation groups, but also three or more evaluation groups, or even just one evaluation group.

[0077] According to the display screen 800 shown in Figure 10, the data processing device 110 can set evaluation groups that will be candidate patients for the search. Furthermore, the data processing device 110 can also freely set the number of evaluation groups.

[0078] <<Example of a second display screen>> Figure 11 shows a display screen 810, which is an example of the display on a display connected to the output unit 117. The display screen 810 includes a mode selection field 801, a field 802 for displaying the set evaluation group, a search condition setting screen 811, and an OK button 812.

[0079] The user operates the input unit 116 to select a condition setting in the mode selection field 801 on the display screen 810. The user directly inputs conditions related to the examination items and patient attributes in the search condition setting screen 811. Once input is complete, the user presses the OK button 812 to save the settings and exit, at which point the search is executed.

[0080] The data processing device 110 may have a function that, when the OK button 812 is pressed and the search conditions are saved, transitions from the display screen 810 to another display screen and instructs the user to execute the search on the transitioned display screen. Alternatively, the search conditions may be prepared in advance in a file in CSV or TSV format, and the user can set the search conditions in the data processing device 110 by selecting this file.

[0081] According to the display screen 810 shown in Figure 11, the data processing device 110 can set multiple search conditions that combine multiple pieces of information related to examination items and patient attributes.

[0082] [Second Embodiment] The second embodiment describes a case in which the setting of search conditions described in the first embodiment is carried out by machine learning or the like.

[0083] Figure 12 shows an example of the functional configuration of the data processing device 110 in the second embodiment. As shown in Figure 12, the condition setting unit 202 of the condition search unit 200 includes a group recognition model learning unit 1001, an item selection unit 1002, and a condition generation unit 1003. The data processing device 110 is configured to set search conditions by data analysis using machine learning or the like through the condition setting unit 202.

[0084] The group identification model learning unit 1001 generates an identification model that identifies multiple evaluation groups, using evaluation data generated by the inspection result evaluation preparation process of the inspection result evaluation preparation unit 204, for multiple evaluation groups set by the evaluation group setting process of the evaluation group setting unit 201.

[0085] The item selection unit 1002 selects the search items to be used as search conditions from the features used to generate the discrimination model generated by the learning process of the group discrimination model learning unit 1001, based on their contribution to the prediction result. For example, the item selection unit 1002 uses SHAP values ​​to calculate the degree to which each feature contributed to the prediction of the discrimination model generated by the group discrimination model learning unit 1001, and selects the top five features that contributed most to the prediction.

[0086] The condition generation unit 1003 generates search conditions based on an analysis of the range of values ​​that contribute most to the prediction result among the features used in model generation, for the inspection items selected by the item selection process of the item selection unit 1002. For example, for the five features selected by the item selection unit 1002, the condition generation unit 1003 analyzes whether the range of values ​​of each feature contributes to the prediction, using the change in SHAP value when the value of only one of the five features is changed. Furthermore, based on the analysis results, the condition generation unit 1003 generates search conditions using a predetermined threshold, such as half of the maximum SHAP value within a single feature.

[0087] Here, the item selection unit 1002 and the condition generation unit 1003 may use, in addition to the method using SHAP values, partial dependence, feature importance using regression coefficients in linear learning methods such as logistic regression, or feature importance including the Gini coefficient in learning methods using decision trees such as random forests.

[0088] Furthermore, the selection of features that contribute significantly to the prediction may be carried out using correlation coefficients such as Pearson's product-moment correlation coefficient or Spearman's rank correlation coefficient, statistical testing methods such as the chi-squared test, information quantities in information theory such as Kullback-Leibler divergence and self-information, Bayesian estimation, recursive feature reduction, stepwise methods, statistical causal inference, and the like.

[0089] Although the condition generation unit 1003 described a method of changing only one of the five features, it is also possible to change the values ​​of multiple features. Furthermore, the condition generation unit 1003 selected the top five features that contributed most to the prediction, but there is no limit to the number of features as long as there is one or more. In addition, for example, features that show a large change in SHAP value when their value is changed, but have a low correlation coefficient with the target variable, may be used for stratification to generate multiple search conditions. For example, if the SHAP value changes significantly at age 30, but the correlation coefficient between age and the target variable is low, the condition generation can be performed separately for those under 30 and those 30 and over. This makes it possible to generate more complex search conditions that are more in line with the actual situation of diseases and pathological conditions.

[0090] Furthermore, the condition generation unit 1003 may generate multiple search conditions by dividing the value set by the generated search conditions into multiple increments. For example, the condition generation unit 1003 may create 10 patterns of search conditions by dividing the value around the set value into increments of 10 in units related to the measurement accuracy of the inspection item. Moreover, the condition generation unit 1003 may apply the mechanism for generating 10 patterns to all inspection items used in the search conditions. The condition generation unit 1003 may set the search range and its increment width to 10% of the value of the initially set search condition, or it may be set by the user each time. The condition generation unit 1003 may generate 100 patterns of search conditions randomly within the range given to the multiple inspection items used in the conditions, rather than using an increment width.

[0091] In addition, after the data processing device 110 sets search conditions through data analysis using machine learning or the like, the user may operate the input unit 116 to modify the values ​​of the search conditions or add new items to the search conditions.

[0092] As described above, in the data processing device 110 of this second embodiment, when setting search conditions using the condition setting unit 202 described above, it is possible to automatically set the search conditions using machine learning or the like.

[0093] [Third Embodiment] The third embodiment describes a case in which, during processing by the search result analysis unit 205 described in the first embodiment, the user can visually determine which search condition is the best.

[0094] The condition presentation unit 402 allows the user to select the optimal search condition from among multiple search conditions based on the evaluation results performed by the condition evaluation unit 401. For example, the condition presentation unit 402 visually presents the evaluation results based on multiple search conditions, using the evaluation indicator selected by the user.

[0095] For example, if the evaluation metric selected by the user is the Youden index, the data processing device 110 can visually represent and present the characteristics of the search results to the user using a graph that utilizes sensitivity and specificity, similar to an ROC curve.

[0096] Figure 13 shows an example of the ROC curve L1 and a composite evaluation plot under multiple conditions. As shown in Figure 13, the ROC curve L1 is generated as a single curve by changing the probability value considered to be a disease, such as the sensitivity and specificity when the threshold for considering a sample as a disease is set to 0.1, and the sensitivity and specificity when the threshold for considering a sample as a disease is set to 0.5, and connecting the plot points of the obtained results. Alternatively, in the accuracy evaluation when using the search conditions described above, the ROC curve L1 is generated by taking FP and TP as both axes, arranging the probability values ​​of being positive in descending order, extending a line in the direction of the TP axis if a sample that should be judged as positive is judged as positive, and extending a line in the direction of the FP axis if a sample that should be judged as positive is judged as negative.

[0097] Therefore, for example, similar to the ROC curve L1, the condition presentation unit 402 can take sensitivity and specificity as both axes, plot the created plot points on a graph for each search condition, and create a composite evaluation plot for multiple conditions. Figure 14 shows an example of a composite evaluation plot displayed on the display screen of the display connected to the output unit 117. By displaying the composite evaluation plot in this way, the data processing device 110 can visualize and display which search condition is superior in the same way as the ROC curve L1.

[0098] Furthermore, the data processing device 110 can also use the Youden index or similar tools to determine the best search conditions. In addition, the data processing device 110 allows the user to visually understand the characteristics of search conditions that do not have the best sensitivity and specificity, making it possible for the user to select search conditions with their desired extraction accuracy.

[0099] As shown in Figure 14, the composite evaluation plot of sensitivity and specificity under multiple conditions for search conditions T1, T2, T3, and T4 allows for an at-a-glance assessment of the relative strengths and weaknesses of each search condition. For search condition T1, specificity is high but sensitivity is low. For search condition T2, sensitivity is high but specificity is low. For search conditions T3 and T4, sensitivity and specificity are between the values ​​of search conditions T1 and T2.

[0100] In this embodiment, when the Youden Index is applied as an evaluation metric, it can be seen that the user can visually determine that search condition T3, located in the upper left of the composite evaluation plot across multiple conditions, is the best search condition. However, to create the composite evaluation plot across multiple conditions, instead of a combination of sensitivity and specificity, sensitivity and goodness of fit may be used, or TP and FP may be used, or a combination of other metrics may be used.

[0101] <Example of display screen> Figure 15 shows a display screen 820, which is an example of the display on the display connected to the output unit 117. The display screen 820 includes a mode selection field 801, a field 802 for displaying the set evaluation group, a search condition display field 821, an OK button 822, and a search result analysis display field 823.

[0102] On the display screen 820, the user operates the input unit 116 to select a search condition in the mode selection field 801. After selecting which search results to display, the selected search condition is reflected in the pre-configured evaluation group display field 802. The composite evaluation plot for multiple conditions under the search condition displayed in the search condition display field 821 is displayed in the search result analysis display field 823. The search condition display field 821 displays the first to fourth conditions, and the fourth condition is selected. The search result analysis display field 823 displays the search results based on the fourth condition. The user can confirm the conditions by operating the arrow 823a to select a search result and then pressing the OK button 822.

[0103] Here, the display area 823 for search result analysis may display the accuracy evaluation results, such as sensitivity, under the conditions displayed in the search condition display area 821. Alternatively, the accuracy evaluation results, such as sensitivity, under the conditions displayed in the search condition display area 821 may be displayed in the search result analysis display area 823 based on a function that allows selecting which evaluation indicators to use for the vertical and horizontal axes of the graph.

[0104] Furthermore, the data processing device 110 may display the accuracy evaluation results, such as sensitivity, under the conditions displayed in the search condition display field 821, in the search condition display field 821 without using the search result analysis display field 823. In addition, the data processing device 110 can also sort and display the search conditions in descending order of sensitivity by clicking on the column named "Sensitivity" in the search condition display field 821, for example, where sensitivity, which is one of the analysis results of the search results, is listed. These display functions are also available even when there is only one search condition and only one set evaluation group.

[0105] As described above, the data processing device 110 of this third embodiment can display the data in a way that allows the user to visually select the optimal search condition from among a plurality of search conditions.

[0106] [Modified Version] Next, a modified version of the first embodiment will be described. The modified version differs from the first embodiment in that it is configured to acquire medical data and drug prescription data from the in-facility system 103.

[0107] Figure 16 shows an example of the configuration of a medical facility 100 in which a data processing device 110 is installed. Compared to Figure 1, medical data 121 is not stored in the storage device 120. On the other hand, in Figure 16, an in-facility system 103 (medical system) connected to medical devices 102 including an analyzer and a diagnostic device for analyzing specimens stores system data 103a, such as data acquired from the medical devices 102 including the analyzer and the diagnostic device. The data processing system is composed of the data processing device 110 and the in-facility system 103. In a modified example, each medical system such as a clinical testing system, a pathology testing system, a drug prescription system, and a DPC (Diagnosis Procedure Combination) system each stores system data 103a. The system data 103a includes, for example, medical data 121. Therefore, the calculation unit 111 of the data processing device 110 is configured to acquire medical data 121 from the in-facility system 103.

[0108] The data processing device 110 is configured to acquire necessary medical data 121 from the facility system 103 using, for example, the patient's identification number, when performing data analysis processing. Alternatively, after acquiring the necessary medical data 121 from the facility system 103, the data analysis processing is performed using the patient's identification number, etc. Even with this configuration, the data processing device 110 can perform data analysis processing similar to that of the first embodiment and achieve similar effects.

[0109] Although embodiments of this disclosure have been described in detail above, the present invention is not limited to the embodiments described above and can be modified in various ways without departing from the gist of the invention. Each embodiment can be modified by adding, deleting, or replacing components, except for essential components. Unless otherwise specified, each component may be singular or plural. Combinations of each embodiment and its variations are also possible. Each of the above-described configurations, functions, and processing units may be implemented in part or in whole by hardware, such as by designing an integrated circuit, or by software, such as by a processor interpreting and executing a program. Data and information such as programs, tables, and files that realize each function can be stored in a recording device such as memory, a hard disk, or an SSD, or on a recording medium such as an IC card, an SD card, or a DVD.

[0110] 110 Data Processing Unit 111 Calculation Unit 111a Processor 112 Storage Unit 112a Program 113 Communication Unit 114 Input / Output Unit 115 Bus 121 Medical Data 122 Hospital Information Data 201 Evaluation Group Setting Unit 202 Condition Setting Unit 203 Search Execution Unit 204 Examination Result Evaluation Preparation Unit 205 Search Result Analysis Unit 301 State Setting Unit 302 Examination Result Organization Unit 401 Condition Evaluation Unit 402 Condition Presentation Unit 501 Search Result Data 502 Examination Result Data 503 Drug Prescription Data 504 Disease Name Data 800, 810, 820 Display Screen 1001 Group Identification Model Learning Unit 1002 Item Selection Unit 1003 Condition Generation Unit

Claims

1. A data processing device that analyzes medical data for each examination item, comprising a processor that executes a program and a storage device that stores the program, wherein the processor performs: an evaluation group setting process that sets two or more evaluation groups, including at least the first evaluation group, from a first evaluation group consisting of multiple patients suffering from or exhibiting a first disease, a second evaluation group consisting of multiple patients suffering from or exhibiting a second disease, and a third evaluation group consisting of multiple patients not suffering from or exhibiting the same disease as the first or second evaluation group; a condition setting process that sets multiple search conditions combining at least two or more data from multiple examination result data, data obtained in medical treatment, and data relating to the patient's attributes, using the two or more evaluation groups set by the evaluation group setting process as search targets; a search execution process that applies the multiple search conditions set by the condition setting process to perform a search on each of the two or more evaluation groups that are the search targets; and an examination result evaluation preparation process that identifies evaluation data from the search result data extracted as a result of the search process and the process in a specified medical cycle. A data processing device that performs a search result analysis process on the evaluation data identified by the inspection result evaluation preparation process, performs an accuracy evaluation of the search results, and presents the optimal search condition from among a plurality of search conditions based on the results of the accuracy evaluation.

2. A data processing device according to claim 1, wherein the condition setting process includes: a learning process that generates an identification model for identifying a plurality of evaluation groups using the evaluation data generated by the inspection result evaluation preparation process for a plurality of evaluation groups set by the evaluation group setting process; an item selection process that selects the inspection items to be used in the search conditions from the identification model generated by the learning process based on at least one of the contribution to the prediction result and the importance of the features used to generate the model; and a condition generation process that generates the search conditions based on an analysis of the regional ranges in which at least one of the contribution to the prediction result and the importance of the features used to generate the model is high for the inspection items selected by the item selection process.

3. A data processing device according to claim 1, wherein the evaluation items of the evaluation data identified by the inspection result evaluation preparation process include at least sensitivity or specificity, and the processor performs an accuracy evaluation using the inspection items including at least sensitivity or specificity in the search result analysis process.

4. A data processing device according to claim 3, wherein the search result analysis process includes a condition presentation process that visually presents the evaluation results based on a plurality of search conditions based on a selected evaluation index.

5. A data processing device according to claim 4, wherein the condition suggestion process includes a plot creation process that takes sensitivity and specificity as both axes, plots the plot points created for each search condition on a graph, and creates a composite evaluation plot for a plurality of search conditions.

6. A data processing device according to claim 1, further comprising a storage device for storing the medical data, wherein the processor acquires the medical data from the storage device.

7. A data processing device according to claim 6, wherein the processor obtains drug prescription data relating to the prescription of a drug from the medical data, and in the test result evaluation preparation process, identifies the evaluation data from the processes in the medical cycle and the medical data, as well as the drug prescription data.

8. A data processing device according to claim 1, wherein the medical cycle includes a series of processes in medical care, such as the initial consultation for a disease, the state before treatment intervention, and the state during treatment.

9. A data processing method for a data processing device that analyzes medical data for each examination item, comprising a processor that executes a program and a storage device that stores the program, the method comprising: the processor setting two or more evaluation groups, including at least the first evaluation group, from a first evaluation group consisting of a plurality of patients suffering from or exhibiting a first disease, a second evaluation group consisting of a plurality of patients suffering from or exhibiting a second disease, and a third evaluation group consisting of a plurality of patients not suffering from or exhibiting the same disease as the first or second evaluation group; the processor setting multiple search conditions that combine at least two or more data from a plurality of examination result data, data obtained in medical treatment, and data relating to the patient's attributes, with the two or more evaluation groups set as search targets; the processor performing a search for each of the two or more evaluation groups that are the search targets by applying the multiple search conditions set; and the processor identifying evaluation data from the search result data extracted as a result of the search process and the process in a specified medical cycle. A data processing method for a data processing device, comprising the steps of: the processor performing an accuracy evaluation of the search results with respect to the identified evaluation data, and presenting the optimal search condition from among a plurality of search conditions based on the results of the accuracy evaluation.

10. A data processing system including a data processing device that analyzes medical data for each test item, and a medical device that includes an analyzer and a diagnostic device for analyzing specimens, and a medical system that stores medical data acquired by the medical device including the analyzer and the diagnostic device, wherein the processor includes: an evaluation group setting process that sets two or more evaluation groups, including at least the first evaluation group, from a first evaluation group consisting of a plurality of patients suffering from or exhibiting a first disease, a second evaluation group consisting of a plurality of patients suffering from or exhibiting a second disease, and a third evaluation group consisting of a plurality of patients suffering from or exhibiting the same disease as the first or second evaluation group, A data processing system that performs the following: a test result evaluation preparation process that identifies evaluation data from the search result data extracted as a result of the search process and the process in a specified medical cycle; a search result analysis process that performs an accuracy evaluation of the search results on the evaluation data identified by the test result evaluation preparation process and presents the optimal search condition from among a plurality of search conditions in a selectable manner based on the results of the accuracy evaluation, wherein the processor acquires the medical data from the medical system.