Disease-Specific Database Generation Device and Method for Data Fabric Infrastructure

Abstract

A method of operating an apparatus operated by at least one processor, the method including constructing a medical data library that generates a table for each data item by applying disease-specific fabric conditions including data items extractable from a clinical data warehouse and medical conditions including disease-specific patient groups for disease-specific research; obtaining a database generation request for research on a specific disease; determining specific data items related to research on the specific disease from the medical data library; applying the fabric conditions for the specific disease to generate a table for each of the specific data items using the limited patient group data for the specific disease; and generating a database for research on the specific disease using the tables of the specific data items.

Description

Technical Field

[0001] The present disclosure relates to a medical data library.

Background Art

[0002] A Clinical Data Warehouse (CDW) stores a vast amount of data such as a Hospital Information System (HIS), Electronic Medical Records (EMR), and an Order Communication System (OCS). Therefore, researchers can extract desired medical data from the CDW and proceed with medical research.

[0003] Real-World Evidence (RWE) is clinical evidence derived by processing and analyzing Real-World Data (RWD) generated and collected in various ways in the actual environment. RWE can solve ethical problems that occur in Randomized Controlled Trials (RCTs) that recruit patients, can greatly save time and costs, and can be widely used.

[0004] For RWE analysis, researchers must construct a research objective database in a way that designs the research setup, extracts a patient group that meets the conditions from the CDW, and continuously adds necessary data. Such a method takes time to construct the database and there is a possibility that necessary data is missing. In addition, an existing research objective database cannot be reused for other research, and the data extraction work should be repeated each time according to the research objective.

Summary of the Invention

Problems to be Solved by the Invention

[0005] The present disclosure relates to an apparatus and method for generating a disease-specific database based on a data fabric infrastructure.

[0006] The present disclosure relates to an apparatus and method for constructing a medical data library capable of selecting data items based on a data fabric infrastructure, and combining the data of the selected items to generate a disease-specific database for the study of a specific disease.

Means for Solving the Problem

[0007] A method of operating an apparatus operated by at least one processor according to an embodiment, including applying disease-specific fabric conditions including medical conditions that can extract disease-specific patient groups and data items extractable from a clinical data warehouse for disease-specific research to construct a medical data library that generates a table for each data item, obtaining a database generation request for the study of a specific disease, determining specific data items related to the study of the specific disease from the medical data library, applying the fabric conditions of the specific disease, and using the limited patient group data of the specific disease to generate a table for each of the specific data items, and generating a database for the study of the specific disease using the tables of the specific data items.

[0008] The step of determining the specific data items may include providing selectable data items to a user interface for the study of the specific disease, and determining the data items confirmed from the user interface as the specific data items.

[0009] The step of providing to the user interface may manage items basically selected from the medical data library by disease or research purpose, and provide basic items related to the database generation request to the user interface.

[0010] In the step of generating each table of the specific data items, when the table of an arbitrary specific data item is to be generated from unstructured data, the table of the arbitrary specific data item can be generated by extracting table record values from the unstructured data through an artificial intelligence model.

[0011] In the step of generating each table of the specific data items, the table generation order of the specific data items can be determined, and the tables can be generated in order.

[0012] The data items of the medical data library can be classified into any one of a core part, an extended part, or an analysis part. The core part can include data items related to essential clinical information basically used in research. The extended part can include data items related to clinical information selectively used according to diseases or research purposes. The analysis part can include data items including derived information required for research analysis using clinical information.

[0013] In the step of obtaining the database generation request, the disease name of the specific disease can be received as input.

[0014] After obtaining the database generation request, the operation method can further include the step of extracting and providing the number of patients with the specific disease from the clinical data warehouse.

[0015] A method for operating a device operated by at least one processor according to another embodiment, the method comprising: obtaining a table specification of target items included in a medical data library; classifying the target items into any one of a core part, an extension part, or an analysis part of the medical data library based on classification criteria; and generating table information and a structure of the target items into structured query language codes, and registering the target items in the medical data library. The medical data library can be constructed to include data items extractable from a clinical data warehouse for disease-specific research, and to apply disease-specific fabric conditions including medical conditions capable of extracting disease-specific patient groups to generate a table for each data item.

[0016] In the classifying step, if the target item is essential clinical information basically used in research, it can be classified into the core part. If the target item is clinical information selectively used according to a disease or research purpose, it can be classified into the extension part. If the target item is derived information necessary for research analysis using clinical information, it can be classified into the analysis part.

[0017] The operating method can further include setting a table generation order based on the table information of the target items.

[0018] If the table of the target items is to be generated from unstructured data, the operating method can further include adding an artificial intelligence model to the table generation process of the target items to convert the unstructured data into a structured table.

[0019] The operation method may further include obtaining a request for generating a database for the study of a specific disease, determining specific data items related to the study of the specific disease from the medical data library, applying the fabric conditions of the specific disease, and generating respective tables of the specific data items using the limited patient group data of the specific disease, and generating a database for the study of the specific disease using the tables of the specific data items.

[0020] An apparatus according to an embodiment includes a memory and a processor that executes instruction words stored in the memory. By executing the instruction words, the processor applies fabric conditions for each disease, including medical conditions that can extract data items extractable from a clinical data warehouse for disease-specific research and can extract disease-specific patient groups, constructs a medical data library that generates a table for each data item, applies the fabric conditions of a specific disease, generates respective tables of the data items selected from the medical data library using the limited patient group data of the specific disease, and is embodied to provide a database including the generated tables for the study of the specific disease.

[0021] The processor can be embodied to obtain a request for generating a database for the study of the specific disease, determine specific data items related to the study of the specific disease from the medical data library, apply the fabric conditions of the specific disease, generate respective tables of the specific data items using the limited patient group data of the specific disease, and generate a database for the study of the specific disease using the tables of the specific data items.

[0022] The processor can be embodied to provide data items selectable from the medical data library for the study of the specific disease to a user interface, and determine the determined data items from the user interface as the specific data items.

[0023] The processor can be embodied to manage items that are basically selected from the medical data library according to disease type or research purpose, and provide basic items related to the database generation request to the user interface. When a table of arbitrary specific data items is to be generated from unstructured data, the processor can be embodied to generate a table of the arbitrary specific data items by extracting table record values from the unstructured data through an artificial intelligence model.

[0024] The processor can be embodied to determine the order of generating tables of the specific data items and generate tables according to the order.

[0025] The data items in the medical data library can be classified into any one of a core part, an extended part, or an analysis part. The core part can include data items related to essential clinical information basically used in research. The extended part can include data items related to clinical information selectively used according to a disease or research purpose. The analysis part can include data items including derived information necessary for research analysis using clinical information.

Advantages of the Invention

[0026] According to the embodiment, a database for real-world evidence (RWE) analysis can be quickly and precisely configured.

[0027] According to the embodiment, a research-purpose database constructed depending on researcher ability can be generated with consistent quality.

[0028] According to the embodiment, a disease-specific database suitable for a research purpose can be quickly generated by freely combining data items constituting the medical data library.

[0029] According to an embodiment, a plurality of researchers can use data items in a medical data library to generate a desired database, and a disease-specific database that has already been generated can be reused.

Brief Description of the Drawings

[0030]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Modes for Carrying Out the Invention

[0031] Hereinafter, with reference to the attached drawings, the embodiments of the present disclosure will be described in detail so that those having ordinary knowledge in the technical field to which the present invention belongs can easily implement it. However, the present disclosure can be embodied in various different forms and is not limited to the embodiments described in this specification. And, in order to clearly explain the present invention in the drawings, parts not related to the explanation are omitted, and similar reference numerals are given to similar parts throughout the specification.

[0032] When a certain part of the entire specification states that a certain component "includes", this means that, unless otherwise stated to the contrary, it does not exclude other components and can further include other components. In addition, terms such as "… part", "… device", "module", etc. described in the specification mean a unit that processes at least one function or operation, and this can be embodied by hardware, software, or a combination of hardware and software.

[0033] The device can include one or more processors, a memory for loading a computer program executed by the processors, a storage device for storing the computer program and various types of data, and a communication interface. Additionally, the device can further include various components. The processor is a device that controls the operation of the device and can be various forms of processors that process the instruction words included in the computer program. For example, it can be configured to include at least one of a CPU (Central Processing Unit), MPU (Micro Processor Unit), MCU (Micro Controller Unit), GPU (Graphic Processing Unit), or any well-known form of processor in the technical field of the present disclosure. The memory stores various types of data, instructions, and / or information. The memory can load the computer program from the storage device so that the instruction words described for executing the operation of the present disclosure can be processed by the processor. The memory can be, for example, a ROM (read only memory), RAM (random access memory), etc. The storage device can non-temporarily store the computer program and various types of data. The storage device can be configured to include non-volatile memories such as ROM (Read Only Memory), EPROM (Erasable Programmable ROM), EEPROM (Electrically Erasable Programmable ROM), flash memory, a hard disk, a removable disk, or any well-known form of computer-readable recording medium in the technical field to which the present disclosure belongs. The communication interface can be a wired / wireless communication module that supports wired / wireless communication. The computer program includes instruction words (instructions) to be executed by the processor and is stored in a non-transitory computer readable storage medium, and the instruction words cause the processor to perform the operations of the present disclosure.

[0034] FIG. 1 is a drawing for explaining a disease-specific database providing apparatus of a data fabric infrastructure according to an embodiment, and FIG. 2 is a drawing for explaining a method of generating a table of unstructured data of an artificial intelligence infrastructure according to an embodiment.

[0035] Referring to FIG. 1, an apparatus 10 operated by at least one processor constructs a medical data library 100 that can select data items (simply referred to as items) in a data fabric infrastructure, and combines the data of the selected items for the study of a specific disease to generate disease-specific databases (Databases, DBs) 200: 200-1, 200-2, 200-3,....

[0036] The apparatus 10 designs a table specification of each item extracted from a Clinical Data Warehouse (CDW) 1, and generates code for generating a table having the designed information and structure. The information and structure of the table can be generated in a Structured Query Language (for example, SQL). The apparatus 10 stores medical conditions that can extract patient groups by disease, and fabric conditions including variables that need to be set according to research purposes. For example, the fabric conditions for lung cancer disease can include conditions for selecting patients who have received a diagnosis including all sub-codes of C33 and C34 of ICD-10 for lung cancer disease classification code, patients who have received medical treatment in a medical oncology department or a cancer hospital, patients who have undergone a chest X-ray examination or a chest computed tomography (CT) before the first diagnosis date, and the like. The fabric conditions for each disease can be set variously according to the clinical guidelines for each disease.

[0037] After that, according to the request for generating a research-purpose DB for a specific disease, the apparatus 10 can apply the fabric conditions for the specific disease to generate tables of items selected from the medical data library 100, and generate a table for each item only with the patient group data of the specific disease. The apparatus 10 can convert medical data according to the fabric conditions to generate a table for each selected item. According to the fabric conditions, the code for converting medical data into a table can be generated in Structured Query Language (e.g., SQL). Here, for the items used in generating the research-purpose DB for a specific disease, depending on the specific disease and research purpose, the apparatus 10 can select at least some items by default, recommend selectable items to the user, and select the items desired by the user.

[0038] The apparatus 10 can generate a research-purpose DB for a specific disease including the tables generated by applying the fabric conditions, and provide it to the requester.

[0039] For example, when the apparatus 10 generates a database for the "study on treatment patterns for pancreatic cancer patients admitted to the intensive care unit", when generating a table for each selected item, by applying the fabric condition of extracting the patient group who "enters the intensive care unit" and receives treatment, a table for each item can be generated only with the patient group data of pancreatic cancer patients who entered the intensive care unit and received treatment.

[0040] For example, when the apparatus 10 generates a database for the "survival analysis study for asthma and lung cancer patients", when generating a table for each selected item, by applying the fabric condition of extracting the patient group related to "asthma" and "lung cancer", a table for each item can be generated only with the patient group data of "asthma" and "lung cancer".

[0041] The medical data library 100 can be composed of selectable items according to specific diseases and research purposes, and these items can be classified into a core part, an extension part, and an analysis part. The items that make up the medical data library 100 can be added, and through this, the next user can reuse the added items. The data of each item is converted into a table with a designed information and structure. Each item can further include selectable detailed items, and the selected detailed items can form the columns of the table.

[0042] The core part can include items that contain essential clinical information basically used for research analysis. The items classified into the core part can include information generated in the processes commonly carried out for patients for medical treatment and treatment when they visit the hospital. For example, the items classified into the core part can include patient information, demographics, hospital visit history, hospital visit, diagnosis, drug prescription, surgery / procedure, diagnostic test results, imaging test results, etc. The patient information item can be composed of detailed items such as age, gender, nationality, blood type, cause of death, etc.

[0043] The extended part can include items that selectively use clinical information according to diseases, application conditions, and research purposes. For example, the items classified in the extended part can be classified and managed into categories such as diseases, procedures, test results, continuous observations of patients, and others, and each category can be further subdivided. For example, cancer diseases can further include clinical items specific to cancer tumors such as lung cancer and pancreatic cancer. For example, cancer-related items can include cancer registration basics, cancer registered tumors, anti-cancer drug administration protocols, anti-cancer Flowsheet, radiation therapy, etc. Procedure-related items can include surgical details, anesthesia details, blood transfusion, medication, etc. Test-related items can include pathological reading results, immunohistochemical test results, molecular genetic test results, electrocardiogram test results, etc. Patient observation items can be relevant clinical information in the emergency room or intensive care unit, including emergency room visits, intensive care unit visits, drug administration, consciousness status, clinical observations, etc.

[0044] The analysis part can include items that contain derived information necessary for research analysis using the data of the core part items and the extended part items. For example, the items classified in the analysis part can include treatment pathways, initial treatments, lines of therapy, event information, etc.

[0045] When the table specification of a new item included in the medical data library 100 is input, the device 10 can determine whether it is mandatory, specialized, or for what purpose it can be used based on the information of the new item, and classify the new item into any one of the core part, the extended part, or the analysis part. Alternatively, the device 10 can receive an input from the user about the part where the new item is located. The device 10 can generate the table information and structure of the new item into a structured query language (e.g., SQL) through the data definition word generation function. The device 10 can set the table generation order of the new item so that the table of the necessary items at the time of generating the table of the new item is generated first. For example, the table generation order can be determined such that after generating the tables of the items included in the core part and the extended part, the table of the items included in the analysis part is generated. When there is a hierarchical relationship between items, the generation order can be determined first.

[0046] For example, the device 10 can receive an input of a new item included in the medical data library 100 and the table specification of the patient information item. Since the patient information item is common data for all patients visiting the hospital, the device 10 can classify the patient information item into the core part. The device 10 can develop a code composed of a structured query language to manage the information (e.g., age, gender, nationality, blood type, cause of death, etc. that make up the columns of the table) included in the table specification of the patient information item. Since the patient number of the patient information is used as a foreign key (FK) for all tables, the device 10 can set the table generation order of the patient information item as number 1.

[0047] For example, the device 10 can receive a new item included in the medical data library 100, namely, an anti-cancer drug administration protocol item and a table specification. Since the anti-cancer drug administration protocol item is data generated when a patient visiting the hospital is diagnosed with a cancer disease and receives an anti-cancer drug prescription, the device 10 can classify the anti-cancer drug administration protocol item into the extended part. The device 10 can develop a code composed of a structured query language to manage the information included in the table specification of the anti-cancer drug administration protocol item (for example, the number of anti-cancer cycles, start date, end date, regimen, dosage, etc. that constitute the columns of the table). Since the table of the anti-cancer drug administration protocol item uses the patient number, hospital visit number, and prescription number as foreign keys (FKs), the device 10 can set the table generation order so that the anti-cancer drug tour protocol item is generated after the patient information table, hospital visit table, and prescription table.

[0048] For example, the device 10 can receive a new item included in the medical data library 100, namely, a treatment pathway item and a table specification. Since the treatment pathway item is an item that mixes the prescription items of the core drugs and the surgery / procedure items for analyzing the process of treating a disease, and the radiation prescription item, anti-cancer drug administration protocol item, and surgery details item in the extended part and applies them to the analysis code, the device 10 can classify the treatment pathway item into the analysis item. The device 10 can develop a code composed of a structured query language to manage the information included in the table specification of the treatment pathway item (for example, the type of treatment, treatment start / end date, order, treatment result, etc. that constitute the columns of the table). The device 10 can set the table generation order so that the table of the treatment pathway item is generated after the tables of the core part and the extended part.

[0049] The device 10 selects items for the research purpose of a specific disease from the medical data library 100 according to the request for generating a research purpose DB of the specific disease, and generates tables of the selected items. At this time, the device 10 applies the fabric conditions for extracting the patient group of the specific disease, and generates a table for each item only with the patient group data of the specific disease.

[0050] The items in the medical data library 100 can be set to be selected according to the disease and research purpose. For example, the core patient information items can be set to be selected by default when generating the research purpose DB of a specific disease. The anti-cancer drug administration protocol items in the extended part are set to be selected by default when generating a database related to cancer diseases, and can be manually selected by the user when generating a database related to other diseases. The treatment pathway items in the analysis part can be set to be selected by default when the research purpose is treatment process analysis.

[0051] The device 10 can generate a table of patient information items only with the patient group data of a specific disease by applying the fabric conditions for extracting the patient group of the specific disease. The device 10 can generate a table of anti-cancer drug administration protocol items only with the patient group data of cancer diseases by applying the fabric conditions for extracting the patient group of cancer diseases. The device 10 can generate a table of treatment pathway items only with the patient group data of a specific disease by applying the fabric conditions for extracting the patient group of the specific disease.

[0052] Referring to FIG. 2, the apparatus 10 can convert the unstructured data 310 stored in the CDW1 into a structured table 320 using an artificial intelligence (AI) model 300. For example, test results such as pathological reading results, immunohistochemical test results, and molecular genetic test results are stored in the form of a report. Therefore, the apparatus 10 can generate the records of the table by extracting the column (field) values that make up the table from the report using an AI model 300 such as a large language model (LLM) or a text recognition model.

[0053] For example, the apparatus 10 can apply fabric conditions for extracting a patient group with a specific disease to obtain immunohistochemical test reports of the patient group with the specific disease. The apparatus 10 can use the AI model 300 to extract information (e.g., test type pathology number, diagnosis name, surgical pathology diagnosis name, biomarker, negative / positive code, test result content, test result symbol characters, test result numerical value, test result unit) that makes up the table from the immunohistochemical test report and generate a table of immunohistochemical test result items.

[0054] FIG. 3 is a drawing for explaining functions implemented in a disease-specific database providing apparatus based on a data fabric according to an embodiment.

[0055] Referring to FIG. 3, the apparatus 10 can include a data item classifier 410, a data definition word generator 420, and a table generation order setter 430 for constructing a medical data library 100. Here, the data item classifier 410, the data definition word generator 420, and the table generation order setter 430 are divided for explaining typical operations of the apparatus 10.

[0056] When the table specification of data items (such as patient information items, anti-cancer drug administration protocol items, treatment pathway items, etc.) included in the medical data library 100 is input to the data item classifier 410, it can determine the availability of essential data, the availability of specialized data, the purpose of utilization, etc. based on the information of the item, and classify the item into any one of the core part, the extended part, or the analysis part. On the other hand, the items classified into the extended part can be classified and managed into categories such as diseases, procedures, test results, patient observations (continuous obs), and others, and each category can be further subdivided. In that case, the data item classifier 410 can classify the item into at least one category.

[0057] The data definition word generator 420 can generate the table information and structure of the item into a structured query language (such as SQL). Information is extracted from the CDW1 by the code generated in the structured query language, and the table of the item can be generated based on the extracted information. When generating the table, fabric conditions are applied to extract the information for constructing the table from the data of the patient group with a specific disease.

[0058] The table generation order setter 430 can set the table generation order of the item so that the table of the necessary item is generated first when generating the table of the item based on the table information of the item. For example, the table generation order can be determined so that the table of the item included in the analysis part is generated after the tables of the items included in the core part and the extended part are generated. When there is a hierarchical relationship between items, the generation order can be determined first.

[0059] The device 10 can generate a disease-specific DB 200 by utilizing the constructed medical data library 100, and thus can include a data item selector 440, a fabric condition applicator 450, an AI-based unstructured data structurer 460, and a table generator 470. Here, the data item selector 440, the fabric condition applicator 450, the AI-based unstructured data structurer 460, and the table generator 470 are classified for explaining the typical operations of the device 10.

[0060] The data item selector 440 can select items for the research purpose of a specific disease from the medical data library 100 according to the generation request of a research purpose DB for a specific disease. Therefore, the data item selector 440 can set the items that are basically selected from the medical data library 100 by disease type and / or research purpose, and can select the items for constructing the DB to meet the user request. For example, the core patient information items can be set to be basically selected when generating a research purpose DB for a specific disease. The anti-cancer drug administration protocol items can be set to be basically selected when generating a database related to cancer diseases, and can be manually selected by the user when generating a database related to other diseases. The treatment pathway items can be set to be basically selected when the research purpose is the analysis of the treatment process.

[0061] The fabric condition applicator 50 stores medical conditions that can extract patient groups by disease type, and fabric conditions including variables that need to be set by research purpose, and can apply the fabric conditions of a specific disease to table generation.

[0062] The AI-based unstructured data structurer 460 can use the AI model 300 to convert the unstructured data stored in the CDW1 into a structured table.

[0063] The table generator 470 can limit the extraction range of patient group data for a specific disease using the fabric conditions for the specific disease, and generate tables of selected items using the patient group data for the specific disease. The table generator 470 can generate tables by converting the patient group data for the specific disease to fit the table structure of each item. The table generator 470 can apply conversion rules to the data using data conversion code written in Structured Query Language to generate the desired tables. For example, the conversion rules can include rules for generating patient group data for a specific disease, using the actually performed data excluding abnormal data and excluding cancellation information. Note that the conversion rules can include rules for generating a combined table of prescriptions, results, and code master tables so that they can be immediately used for analysis, generating a combined table of the same type of information in one table, and standardizing and generating unstructured data through the AI model 300.

[0064] The disease-specific DB 200 can be constructed including the tables generated from the table generator 470.

[0065] The device 10 can generate a table of patient information items using only the patient group data for a specific disease by applying the fabric conditions for extracting the patient group for the specific disease. The device 10 can generate a table of anti-cancer drug administration protocol items using only the patient group data for the cancer disease by applying the fabric conditions for extracting the patient group for the cancer disease. The device 10 can generate a table of treatment pathway items using only the patient group data for a specific disease by applying the fabric conditions for extracting the patient group for the specific disease.

[0066] FIG. 4 is a drawing for explaining, as an example, the generation of a disease-specific database according to an embodiment.

[0067] Referring to FIG. 4, the apparatus 10 can construct a medical data library 100A composed of selectable items according to specific diseases and research purposes. The medical data library 100A includes items classified into a core part, an extension part, and an analysis part. The table of each item is generated using the patient group data of a specific disease limited by fabric conditions. The code for generating a table from CDW1 can be generated in SQL.

[0068] The items of the medical data library 100A can be selected according to the research purposes of specific diseases. For example, the items of the core part can be automatically selected, the items of the extension part can select the items related to specific diseases and research purposes, and the items of the analysis part can select the items related to research purposes.

[0069] For example, assume that the device 10 receives a DB generation request for the "Research on Treatment Patterns for Pancreatic Cancer Patients Who Visited the Intensive Care Unit". The device 10 can select all the items in the core part of the medical data library 100A. Among the items in the extended part, the device 10 can select cancer disease-related items (such as cancer registration information, tumor detailed information, NGS test results, clinical immunology test results, radiation prescription information, anti-cancer drug protocol, pathological reading results, etc.), intensive care unit-related items (such as hospital visit history information, intensive care unit admission information, continuous drug administration information, etc.). Among the items in the analysis part, the device 10 can select items related to the treatment pattern research (such as Line of Therapy, Treatment Pathway, initial treatment, etc.). When generating the tables for each selected item, the device 10 can apply the fabric condition for extracting the group of patients who "entered the intensive care unit" and received treatment and the fabric condition for extracting the group of patients related to "pancreatic cancer" to generate the tables for each item only with the patient group data of pancreatic cancer patients who entered the intensive care unit. At that time, the device 10 can confirm the table generation order of the selected items and generate the tables of the items according to the table generation order. The device 10 can generate the requested disease-specific DB 200-4 using the generated tables.

[0070] For example, assume that device 10 receives a "Survival Analysis Study for Asthma and Lung Cancer Patients". Device 10 can select all the items in the core part of the medical data library 100A. Among the items in the extended part, device 10 can select cancer disease-related items (such as cancer registration information, tumor detailed information, NGS test results, clinical immunology test results, etc.) and asthma disease-related items (such as skin reaction test results, allergy test results, lung function test results, etc.). Among the items in the analysis part, device 10 can select items related to the survival analysis study (such as Event information, CCI score, etc.). When generating a table for each selected item, device 10 can apply a fabric condition to extract patient groups related to "lung cancer" and "asthma", and generate a table for each item only with the patient group data of "lung cancer" and "asthma". At that time, device 10 can confirm the table generation order of the selected items and generate the tables of the items according to the table generation order. Device 10 can generate the requested disease-specific DB200-5 using the generated tables.

[0071] Figure 5 is a flowchart for explaining a method of constructing a medical data library according to an embodiment.

[0072] Referring to Figure 5, device 10 obtains a table specification of target items included in the medical data library 100 in S110.

[0073] Device 10 classifies the target items into any one of the core part, extended part, or analysis part of the medical data library 100 based on classification criteria in S120. The classification criteria can include the availability of essential data for DB generation, the availability of specialized data, the purpose of utilization, etc. If the target item is essential clinical information basically used in the study, it can be classified into the core part. If the target item is clinical information selectively used according to the disease, application conditions, and research purpose, it can be classified into the extended part. If the target item includes derived information necessary for research analysis using clinical information, it can be classified into the analysis part.

[0074] Device 10 generates the table information and structure of the target items into Structured Query Language code and registers the target items in the medical data library 100 at S130. Information is extracted from CDW1 according to the code generated in Structured Query Language, and the table records of the items can be generated based on the extracted information.

[0075] Device 10 sets the table generation order based on the table information of the target items at S140. The table generation order can be determined based on the foreign keys (FKs) to be concatenated during the generation of the target item tables. For example, the table generation order can be determined such that the tables of the items included in the core part and the extended part are generated first, and then the tables of the items included in the analysis part are generated. When there is a hierarchical relationship between the items, the generation order can be determined first.

[0076] On the other hand, when the table of a specific item (for example, immunoassay results) should be generated from the unstructured data stored in report form, device 10 adds the AI model 300 for converting the unstructured data into a structured table to the table generation process of the specific item at S150. The AI model 300 can be implemented with an LLM, a text recognition model, etc.

[0077] The items registered in the medical data library 100 are provided in a selectable form for the study of specific diseases, and can be implemented with a data fabric architecture in which the tables of the selected items are generated using only the patient group data according to the disease-specific fabric conditions.

[0078] FIG. 6 is a flowchart for explaining a disease-specific database generation method according to an embodiment.

[0079] Referring to FIG. 6, the apparatus 10 constructs a medical data library S210 that includes data items extractable from the CDW1 for disease-specific research and applies fabric conditions including medical conditions that can extract disease-specific patient groups to generate a table of each data item.

[0080] The apparatus 10 obtains a request for generating a DB for a specific disease S220. The apparatus 10 can obtain a request including a disease name and a data search period. The apparatus 10 can extract and provide the number of patients with a specific disease from the CDW1. The apparatus 10 can provide patient characteristic statistics such as age, gender, nationality, etc., so that the user can confirm whether the number of patients is appropriate for the research.

[0081] The apparatus 10 determines specific data items related to the research of a specific disease from the medical data library 100 S230. The apparatus 10 can provide a user interface that allows the user to determine the necessary items for the disease-specific DB. The apparatus 10 can manage the items initially (default) selected from the medical data library 100 by disease and / or research purpose, and provide the basic items related to the request to the user interface. The apparatus 10 can provide selectable data items to the user interface for the research of a specific disease, and determine the data items determined by the user at the user interface as the specific data items. The apparatus 10 can recommend to the user the necessary items for the disease-specific DB based on the roles of the items in the medical data library 100 and the relationships between the items, and assist the user in selecting items.

[0082] The device 10 performs S240 which applies the fabric conditions of a specific disease and uses the data of a group of patients with the specific disease limited thereto to generate respective tables of specific data items. The fabric conditions may further include variables that need to be set according to research purposes. When the tables of specific items are to be generated from unstructured data, the device 10 can use the AI model 300 to convert the unstructured data into structured tables. The device 10 can generate a table of specific items by extracting table record values from unstructured data through the AI model 300. At this time, the device 10 can determine the order in which the tables of the items are generated and generate the tables according to the order.

[0083] The device 10 performs S250 which generates a disease-specific DB using the tables of the specific data items.

[0084] As described above, according to the embodiment, a database for real-world evidence (RWE) analysis can be quickly and precisely configured.

[0085] According to the embodiment, a research-purpose database constructed depending on the capabilities of researchers can be generated with consistent quality.

[0086] According to the embodiment, a disease-specific database that meets research purposes can be quickly generated by freely combining the data items that make up the medical data library.

[0087] According to the embodiment, the data items of the medical data library can be used by multiple researchers to generate a desired database, and the already generated disease-specific database can be reused.

[0088] The embodiments of the present disclosure described above are not implemented only by the device and method, but can also be implemented through a program that realizes functions corresponding to the configurations of the embodiments of the present disclosure or a recording medium on which the program is recorded.

[0089] Although the embodiments of the present disclosure have been described in detail above, the scope of the rights of the present disclosure is not limited thereto, and various modifications and improvements by those skilled in the art using the basic concept of the present disclosure defined in the following claims also belong to the scope of the rights of the present disclosure.

Claims

1. A method for operating a device operated by at least one processor, comprising: Applying disease-specific fabric conditions including data items extractable from a clinical data warehouse and medical conditions including disease-specific patient groups that can be extracted for disease-specific research to construct a medical data library that generates a table for each data item; Obtaining a request for generating a database for research on a specific disease; Determining specific data items related to the research on the specific disease from the medical data library; Applying the fabric conditions of the specific disease and using the limited patient group data of the specific disease to generate a table for each of the specific data items; and Generating a database for the research on the specific disease using the tables of the specific data items. The operating method is characterized by including the above steps.

2. The step of determining the specific data items includes: Providing selectable data items to a user interface for the research on the specific disease; and Determining the data items confirmed from the user interface as the specific data items. The operating method according to claim 1 is characterized by including the above steps.

3. The step of providing to the user interface includes: Managing items basically selected from the medical data library by disease or research purpose, and providing basic items related to the database generation request to the user interface. The operating method according to claim 2 is characterized by including the above steps.

4. The step of generating a table for each of the specific data items includes: When a table for an arbitrary specific data item is to be generated from unstructured data, extracting table record values from the unstructured data through an artificial intelligence model to generate the table for the arbitrary specific data item. The operating method according to claim 1 is characterized by including the above steps.

5. The step of generating a table for each of the specific data items includes: Determining the order of generating tables for the specific data items and generating tables according to the order. The operating method according to claim 1 is characterized by including the above steps.

6. The data items in the medical data library are classified into any one of a core part, an extended part, or an analysis part. The core part includes data items related to essential clinical information basically used in research. The extended part includes data items related to clinical information that are selectively used depending on the disease or research purpose. The operation method according to claim 1, wherein the analysis part includes data items including derived information necessary for research analysis using clinical information.

7. The step of obtaining the database generation request The operation method according to claim 1, characterized in that it receives the disease name of the specific disease.

8. After obtaining the database generation request, the step of extracting and providing the number of patients with the specific disease from the clinical data warehouse The operation method according to claim 1, further comprising the above.

9. An operation method of a device operated by at least one processor, The step of obtaining the table specification of the target item included in the medical data library Classifying the target item into any one of the core part, extended part, or analysis part of the medical data library based on classification criteria, and Generating the table information and structure of the target item into a structured query language code, and registering the target item in the medical data library. The medical data library includes data items extractable from a clinical data warehouse for disease-specific research, and is constructed to generate a table for each data item by applying disease-specific fabric conditions including medical conditions capable of extracting disease-specific patient groups. The operation method is characterized by this.

10. The classifying step When the target item is essential clinical information basically used in research, classify it into the core part. When the target item is clinical information selectively used depending on the disease or research purpose, classify it into the extended part. The operation method according to claim 9, characterized in that when the target item is derived information necessary for research analysis using clinical information, it is classified into the analysis part.

11. The step of setting the table generation order based on the table information of the target item The operation method according to claim 9, further comprising the above.

12. When the table of the target item is to be generated from unstructured data, the step of adding an artificial intelligence model to the table generation process of the target item in order to convert the unstructured data into a structured table The operation method according to claim 9, further comprising the above.

13. The step of obtaining a request for generating a database for the study of a specific disease, The step of determining specific data items related to the study of the specific disease from the medical data library, The step of applying the fabric conditions of the specific disease and generating respective tables of the specific data items using the limited patient group data of the specific disease, and The step of generating a database for the study of the specific disease using the tables of the specific data items, The operation method according to claim 9, further comprising the above.

14. A memory, and A processor that executes instruction words stored in the memory, By executing the instruction words, the processor Constructs a medical data library that includes data items extractable from a clinical data warehouse for disease-specific research and applies disease-specific fabric conditions including medical conditions that can extract disease-specific patient groups to generate a table for each data item, An apparatus characterized in that it is embodied to apply the fabric conditions of a specific disease, generate respective tables of data items selected from the medical data library using the limited patient group data of the specific disease, and provide a database including the generated tables for the study of the specific disease.

15. The processor Obtains a request for generating a database for the study of the specific disease, Determines specific data items related to the study of the specific disease from the medical data library, Applies the fabric conditions of the specific disease and generates respective tables of the specific data items using the limited patient group data of the specific disease, The apparatus according to claim 14, characterized in that it is embodied to generate a database for the study of the specific disease using the tables of the specific data items.

16. The processor Provides data items selectable from the medical data library to a user interface for the study of the specific disease, The apparatus according to claim 14, characterized in that it is embodied to determine the confirmed data items from the user interface as the specific data items.

17. The processor The device according to claim 16, wherein items basically (default) selected from the medical data library are managed according to disease type or research purpose, and basic items related to the database generation request are provided to the user interface.

18. The processor is configured to generate a table of any specific data items by extracting table record values from the unstructured data through an artificial intelligence model when the table of the any specific data items is to be generated from the unstructured data, as claimed in claim 14.

19. The processor is configured to determine the order of generating tables of the specific data items and generate tables according to the order, as claimed in claim 14.

20. The data items of the medical data library are classified into any one of a core part, an extended part, or an analysis part, the core part includes data items related to essential clinical information basically used in research, the extended part includes data items related to clinical information selectively used according to diseases or research purposes, the analysis part includes data items including derived information necessary for research analysis using clinical information, as claimed in claim 14.

Images