Medical clinical data standardization conversion method and device based on SDTM standard

Abstract

The invention provides a medical clinical data standardization conversion method and device based on an SDTM standard, and relates to the technical field of medical clinical data processing. The method comprises the following steps: acquiring medical clinical data from different sources; analyzing the medical clinical data into two-dimensional metadata and a data entity; the two-dimensional metadata serve as a retrieval condition, a text generation model is used for interacting with a knowledge base to conduct retrieval enhancement generation, the mapping relation between the two-dimensional metadata and domains and variables in the SDTM standard is obtained, and the knowledge base is a multi-dimensional knowledge base which is constructed based on the SDTM standard and faces medical clinical data standardization processing; converting the two-dimensional metadata and the data entity according to the mapping relation to obtain a preliminary data set meeting the SDTM standard; and performing quality detection processing meeting the SDTM standard on the preliminary data set to obtain a standard data set.

Description

Technical Field

[0001] The present application relates to the technical field of medical clinical data processing, and in particular to a method and device for standardizing and converting medical clinical data based on the SDTM standard. Background Art

[0002] Medical clinical data comes from a wide range of sources (including hospital information systems and pharmaceutical companies' electronic data collection systems), is complex (including structured medical records and case report forms), and comes in a variety of formats (for example, databases and application programming interfaces). These data's quality deficiencies and non-standardization not only impact the reliability of clinical trial results but also complicate regulatory review, thereby delaying new drug approvals and medical device launches. Cross-border collaborations for multinational pharmaceutical companies, biomedical research institutions, and clinical research organizations present challenges such as differing data standards, inconsistent formats, and privacy compliance requirements. This creates an urgent need for effective integration of heterogeneous data from multiple sources and consistent cross-regional management.

[0003] To address the standardization issues of medical clinical data, the Clinical Data Interchange Standards Consortium (CDISC) has developed a series of international standards, including the Study Data Tabulation Model (SDTM), requiring that medical clinical data be classified according to specific domains and converted into the SDTM standardized format to support regulatory submission.

[0004] However, existing solutions for converting medical clinical data into the SDTM standard format rely on manually configuring mapping tables and implementing SDTM standard format conversion through program writing. This has the problems of low automation, high time consumption, high cost, and inability to process multi-source heterogeneous data. Summary of the Invention

[0005] In view of the above problems, the embodiments of the present application provide a method and apparatus for standardizing and converting medical clinical data based on the SDTM standard, so as to overcome the above problems or at least partially solve the above problems.

[0006] In a first aspect of the embodiments of the present application, a method for standardizing and converting medical clinical data based on the SDTM standard is disclosed, the method comprising: Obtain medical clinical data from different sources; Parsing the medical clinical data into two-dimensional metadata and data entities, wherein the two-dimensional metadata represents data attribute information and feature information described in a two-dimensional table structure, and the data entity represents the actually recorded medical clinical data; Using the two-dimensional metadata as a retrieval condition, a text generation model is used to interact with a knowledge base to perform retrieval enhancement generation, thereby obtaining a mapping relationship between the two-dimensional metadata and domains and variables in the research data tabulation model SDTM standard. The knowledge base is a multidimensional knowledge base for medical clinical data standardization processing, which is constructed based on the SDTM standard; Converting the two-dimensional metadata and the data entity according to the mapping relationship to obtain a preliminary data set that meets the SDTM standard; The preliminary data set is subjected to quality inspection to ensure that it meets the SDTM standard, thereby obtaining a standard data set.

[0007] Optionally, obtain medical clinical data from various sources, including: Obtain medical clinical data from different sources through data access or data import methods; Among them, medical clinical data from different sources include: daily patient medical record data from the hospital information system, electronic case report forms from the electronic data collection system, and other medical clinical data from sources other than the hospital information system and the electronic data collection system.

[0008] Optionally, parsing the medical clinical data into two-dimensional metadata and data entities includes: Parsing the medical clinical data into metadata and data entities, wherein the metadata is used to describe data attribute information and feature information; According to the semantic information of the medical clinical data and the provisions of the SDTM standard, the metadata is converted into a standardized two-dimensional table structure to obtain the two-dimensional metadata.

[0009] Optionally, the knowledge base includes: a rule base with data processing rules and data mapping rules, a standard base integrating SDTM standards, and a term base with professional terms in the medical field; The two-dimensional metadata is used as a retrieval condition, and a text generation model is used to interact with a knowledge base to perform retrieval enhancement generation, thereby obtaining a mapping relationship between the two-dimensional metadata and domains and variables in the SDTM standard, including: Using the two-dimensional metadata as a search condition, the text generation model is used to search the rule library, the standard library, and the term library respectively to obtain a search result consisting of target rules, target domain knowledge, and target terms; Integrating the search results into system prompt words, and placing the two-dimensional metadata into user prompt words; The mapping relationship is generated by utilizing the text generation model based on the system prompt words and the user prompt words.

[0010] Optionally, the two-dimensional metadata and the data entity are converted according to the mapping relationship to obtain a preliminary data set that meets the Research Data Tabulation Model (SDTM) standard, including: converting the two-dimensional metadata into standard metadata according to the mapping relationship, wherein the standard metadata includes a domain identifier, a variable label, a data type, and a controlled terminology; A preliminary data set meeting the SDTM standard is generated according to the standard metadata and the data entity.

[0011] Optionally, the preliminary data set is subjected to quality inspection to meet the SDTM standard to obtain a standard data set, including: Testing the preliminary data set according to a target detection strategy, wherein the target detection strategy includes: data set integrity check, data format and type verification, controlled terminology correction, logical contradiction check, and domain rule verification; In the case of a detection trigger error, the error message and the corresponding two-dimensional metadata are sent to the text generation model, and the execution step is returned: the two-dimensional metadata is used as a search condition, and the text generation model is used to interact with the knowledge base to perform search enhancement generation to obtain a mapping relationship between the two-dimensional metadata and the domains and variables in the standard; In the case that the detection does not trigger an error, the preliminary data set currently being detected is used as the standard data set.

[0012] Optionally, the method further includes: When the number of times the preliminary data set is tested according to the target detection strategy reaches a number threshold, it is determined that the medical clinical data has quality problems, and the quality problems are output to prompt that manual verification processing is required.

[0013] Optionally, the method further includes: Obtaining a generation rule for a derived variable described in natural language, wherein the derived variable represents a variable related to the clinical trial protocol and calculated based on the recorded variables; Using the text generation model, generating data processing code for configuring the derived variables according to the generation rules of the derived variables; The data processing code is executed according to the standard data set to generate derived variable data, and the derived variable data is inserted into the standard data set to obtain a final standard data set.

[0014] Optionally, manually reviewing the final standard data set; After confirming that the final standard data set meets regulatory requirements, the final standard data set and a target format file are output, where the target format file records standard metadata of the final standard data set.

[0015] A second aspect of the embodiments of the present application discloses a medical clinical data standardization conversion device based on the SDTM standard, the device comprising: Acquisition module, used to obtain medical clinical data from different sources; a parsing module, configured to parse the medical clinical data into two-dimensional metadata and data entities, wherein the two-dimensional metadata represents data attribute information and feature information described in a two-dimensional table structure, and the data entities represent the actually recorded medical clinical data; An interactive module, configured to use the two-dimensional metadata as a retrieval condition, utilize a text generation model to interact with a knowledge base for retrieval enhancement generation, and obtain a mapping relationship between the two-dimensional metadata and domains and variables in the SDTM standard. The knowledge base is a multidimensional knowledge base for standardized processing of medical clinical data, constructed based on the SDTM standard; a conversion module, configured to convert the two-dimensional metadata and the data entity according to the mapping relationship to obtain a preliminary data set that meets the SDTM standard; The detection module is used to perform quality detection processing on the preliminary data set to ensure that the data set meets the SDTM standard, so as to obtain a standard data set.

[0016] The third aspect of the embodiments of the present application discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the medical clinical data standardization conversion method based on the SDTM standard described in the first aspect of the embodiments of the present application are implemented.

[0017] The fourth aspect of the embodiments of the present application discloses a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the medical clinical data standardization conversion method based on the SDTM standard described in the first aspect of the embodiments of the present application are implemented.

[0018] The fifth aspect of the embodiments of the present application discloses a computer program product, including a computer program, which, when executed by a processor, implements the steps of the medical clinical data standardization conversion method based on the SDTM standard described in the first aspect of the embodiments of the present application.

[0019] The embodiments of the present application include the following advantages: In the embodiment of the present application, the existing limitations of single data are broken through, and medical clinical data from different sources are parsed into two-dimensional metadata and data entities. The two-dimensional metadata is used as a retrieval condition, and a text generation model is used to interact with the knowledge base for retrieval enhancement generation, and a mapping relationship between the two-dimensional metadata and the domains and variables in the SDTM standard is obtained. The automatic and accurate mapping of the two-dimensional metadata and the domains and variables in the SDTM standard is achieved, which significantly reduces the professional threshold of manual configuration and improves standardization efficiency. Thus, the two-dimensional metadata and data entities are converted according to the mapping relationship to obtain a preliminary data set that meets the SDTM standard, and the preliminary data set is subjected to quality inspection processing that meets the SDTM standard to ensure that a high-quality standard data set is obtained. In this way, low-cost, efficient and accurate standardized intelligent conversion of medical clinical data is achieved, reducing dependence on manual labor and high professional requirements. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0021] Figure 1 This is a flowchart of the steps of a medical clinical data standardization conversion method based on the SDTM standard provided in an embodiment of the present application; Figure 2 Schematic diagram of a medical clinical data standardization conversion system based on the SDTM standard provided in an embodiment of the present application; Figure 3 This is a schematic diagram of the structure of a medical clinical data standardization conversion device based on the SDTM standard provided in an embodiment of the present application; Figure 4 This is a structural diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0022] To make the above-mentioned purposes, features, and advantages of this application more clearly understood, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of this application.

[0023] In order to better understand the technical solution of this application, the technical concepts involved in this application are first explained.

[0024] Existing solutions for converting medical clinical data into the SDTM standard format involve manually determining the mapping between source and target variables using Excel tables (Excel refers to spreadsheet software), then writing a SAS (Statistical Analysis System) program to convert the original medical clinical data into the SDTM format; or using an Excel mapping table to clearly define the conversion process from source variables to target variables, and then writing a SAS macro program to implement the specific conversion.

[0025] However, these solutions suffer from the following issues: 1) They have a low level of automation, relying on manual Excel spreadsheets for variable mapping and rule configuration, which is inefficient and prone to human error. 2) They are cost-effective and user-friendly, and require a high level of expertise, requiring in-depth familiarity with the SDTM standard and proficiency in writing SAS programs. 3) They are unable to handle multi-source heterogeneous data. Existing solutions struggle to effectively integrate heterogeneous data from diverse data sources such as HIS (Hospital Information System) and EDC (Electronic Data Capture), and are therefore only adaptable to specific systems. 4) They lack dynamic configuration and flexible scalability, relying heavily on the SAS ecosystem. Its closed architecture limits collaboration with open-source tools like Python, necessitating the development of customized macros for each trial and lacking reusable standardized modules. 5) Derived variable generation is complex. Derived variables are diverse and complex, requiring extensive manual coding of SAS code based on complex statistical logic and clinical rules. This process is time-consuming and prone to data quality compromise due to programming errors.

[0026] To overcome the limitations of related technologies, embodiments of the present application provide a standardized conversion method for medical clinical data based on the SDTM standard. This method intelligently converts medical clinical data from various sources (e.g., HIS, EDC, etc.) into an SDTM standard dataset. Leveraging a text generation model, this method converts the metadata in the medical clinical data into a two-dimensional form based on semantic information and maps it to the domains and variables defined by the SDTM standard. The medical clinical data is then gradually converted based on the metadata mapping relationships. The preliminary SDTM dataset is then subjected to SDTM standard-compliant quality checks (e.g., implementing multi-level quality control measures including range checks, logical checks, and consistency verification) to ensure that the output data conforms to the technical specifications of the SDTM standard, resulting in an SDTM standard dataset.

[0027] In some embodiments, the application is further integrated with the clinical trial protocol and, through natural language dialogue, generates data processing code to intelligently configure derived variables, ultimately outputting a SDTM-compliant dataset and Define-XML metadata (Define-XML is a source data description file format in CDISC, used to define dataset variables, controlled terminology, and specific source data). This application will help improve the review efficiency of regulatory agencies while supporting subsequent statistical analysis, cross-study data integration, and electronic submission, thereby improving the processing efficiency and quality of clinical research data, reducing labor costs and technical barriers, and providing technical support for accelerating new drug development and medical research.

[0028] The following is a detailed description of the medical clinical data standardization conversion method based on the SDTM standard in conjunction with the accompanying drawings.

[0029] Reference Figure 1 As shown, Figure 1 This is a flowchart of a method for standardizing and converting medical clinical data based on the SDTM standard provided in the embodiment of the present application. Figure 1 As shown, the medical clinical data standardization conversion method based on the SDTM standard may include steps S110 to S150: Step S110: Acquire medical clinical data from different sources.

[0030] Medical clinical data is usually structured or semi-structured data, and medical clinical data from different sources usually have different data structures. The embodiments of the present application break through the limitations of existing single data and obtain medical clinical data from different sources for conversion, that is, convert multi-source heterogeneous medical clinical data.

[0031] In an optional embodiment, obtaining medical clinical data from different sources includes: obtaining medical clinical data from different sources through a data access method or a data import method; wherein the medical clinical data from different sources include: daily patient medical record data from a hospital information system, electronic case report forms from an electronic data acquisition system, and other medical clinical data from sources other than the hospital information system and the electronic data acquisition system.

[0032] In an embodiment of the present application, daily medical records of patients from a hospital information system include structured or semi-structured data generated by medical activities such as outpatient visits, hospitalizations, and tests and examinations. Although these data have clinical value, their original design purpose is not to directly serve clinical trial research. Therefore, when a new clinical trial project needs to reuse historical data from a hospital information system (i.e., daily medical records of patients), the embodiment of the present application converts them into a standard data set (i.e., an SDTM standard data set) based on the specific requirements of the current trial plan to ensure that they comply with the inclusion and exclusion criteria, observation indicator definitions, and data collection specifications of the clinical trial.

[0033] The electronic case report form from the electronic data collection system strictly follows the data structure requirements of the clinical trial protocol during the database design phase. Each field is pre-defined with clear clinical semantics and collection specifications, so that the database structure has a data logic framework that is relatively consistent with the trial protocol.

[0034] Other medical clinical data (i.e., other structured data) can also be converted into SDTM standard datasets through the method implemented in this application.

[0035] Specifically, the methods of obtaining medical clinical data are divided into data access and data import. Data access is a dynamic data interaction mode that accesses source data in real time or on a scheduled basis through an API (Application Programming Interface), database connection, or streaming transmission, without the need to directly obtain files. This method does not require the transmission and storage of physical files and can maintain the timeliness and consistency of the data. Data import directly provides structured files of medical clinical data, and obtains the corresponding medical clinical data through batch loading of structured files.

[0036] Step S120: Parsing the medical clinical data into two-dimensional metadata and data entities, wherein the two-dimensional metadata represents data attribute information and feature information described in a two-dimensional table structure, and the data entity represents the actually recorded medical clinical data.

[0037] In the embodiments of this application, two-dimensional metadata refers to metadata in a two-dimensional table structure, where rows of the two-dimensional metadata represent observations (or actual records) and columns represent variables. Each medical clinical data item is described by corresponding two-dimensional metadata, including its data attribute information and feature information. Data entities represent actual recorded medical clinical data and can be understood as actual observations. For medical clinical data from different sources, analysis is performed to identify their inherent patterns, and then the two-dimensional metadata and data entities are uniformly extracted.

[0038] In an optional embodiment, the medical clinical data is parsed into two-dimensional metadata and data entities, including: parsing the medical clinical data into metadata and data entities, the metadata being used to describe data attribute information and feature information; and converting the metadata into a standardized two-dimensional table structure according to the semantic information of the medical clinical data and the provisions of the SDTM standard to obtain the two-dimensional metadata.

[0039] Specifically, after obtaining the metadata through parsing, the metadata is organized into a standardized two-dimensional table structure through an intelligent mapping algorithm based on the semantic information of the medical clinical data and the provisions of the SDTM standard, ensuring that the fields of each single table can accurately correspond to the domains defined in the SDTM dataset during the subsequent conversion process, thereby improving the accuracy of data standardization conversion.

[0040] Step S130: Using the two-dimensional metadata as a retrieval condition, the text generation model is used to interact with the knowledge base to perform retrieval enhancement generation, and a mapping relationship between the two-dimensional metadata and the domains and variables in the research data tabulation model SDTM standard is obtained. The knowledge base is a multidimensional knowledge base for medical clinical data standardization processing built based on the SDTM standard.

[0041] Among them, the mapping relationship between two-dimensional metadata and domains and variables in the SDTM standard can also be understood as a conversion template for executing medical clinical data (i.e., two-dimensional metadata and data entities).

[0042] The text generation model can be an open source general-purpose large language model (for example, Tongyi Qianwen, DeepSeek, etc.). The text generation model can be migrated to the closed environment of a local medical data center through private deployment, avoiding the risk of leakage of sensitive medical clinical data.

[0043] The text generation model can also be fine-tuned locally. Specifically, the local fine-tuning model is a customized model that is fine-tuned for domain adaptability based on local medical corpus (such as electronic medical records, clinical trial reports, and other professional texts). Through continuous training and parameter optimization in the medical vertical field, the local fine-tuning model can more accurately understand medical terminology and grasp clinical context logic, and demonstrate stronger domain adaptability in professional scenarios such as data standardization and conversion and case report generation.

[0044] In the embodiment of the present application, based on the intelligent mapping technology between the knowledge base and the text generation model, a multidimensional knowledge base for the standardized processing of medical clinical data is constructed based on the SDTM standard, so that retrieval enhancement generation is performed through the interaction between the text generation model and the knowledge base, and the automatic and accurate mapping of two-dimensional metadata and domains and variables in the SDTM standard is achieved, which significantly reduces the professional threshold of manual configuration and improves standardization efficiency.

[0045] In an optional embodiment, the knowledge base includes: a rule base having data processing rules and data mapping rules, a standard base integrating the SDTM standard, and a term base having professional terminology in the medical field.

[0046] The rule base includes data processing rules for structured processing, such as data cleansing, logical validation, and anomaly detection. Data mapping rules can be understood as establishing standardized conversion templates for field mapping relationships between different data sources and the SDTM standard. The standard library, which integrates the SDTM standard, can be understood as a standard library that integrates the domain knowledge of the SDTM standard's technical documentation and implementation guidelines. A terminology library containing specialized medical terminology can be understood as a terminology system related to medical clinical trials.

[0047] Furthermore, the above step S130 specifically includes sub-steps S130-1 to S130-3: Step S130 - 1 : Using the two-dimensional metadata as a search condition, the text generation model is used to search the rule library, the standard library, and the term library respectively to obtain a search result consisting of target rules, target domain knowledge, and target terms.

[0048] Step S130 - 2 : Integrate the search results into the system prompt words, and place the two-dimensional metadata into the user prompt words.

[0049] Step S130 - 3 : Generate the mapping relationship based on the system prompt word and the user prompt word using the text generation model.

[0050] Specifically, using 2D metadata as retrieval conditions, the rule library is searched to obtain target rules related to the current data processing task. The standard library is searched to obtain target domain knowledge related to the current data processing task. The term library is searched to obtain target terms related to the current data processing task. The current data processing task can be determined based on the user's experimental plan and user description information. The retrieval results are integrated into the system prompt word to form a processing context, and the 2D metadata is placed in the user prompt word as a specific processing object. Thus, a mapping relationship between the 2D metadata and the domains and variables in the SDTM standard is generated through a text generation model.

[0051] Step S140: converting the two-dimensional metadata and the data entity according to the mapping relationship to obtain a preliminary data set that meets the SDTM standard.

[0052] In the embodiment of the present application, since the mapping relationship includes the precise matching relationship between the two-dimensional metadata and the domains and variables in the SDTM standard, the two-dimensional metadata and data entities can be accurately converted into a preliminary data set that meets the SDTM standard based on the mapping relationship.

[0053] In an optional embodiment, the above step S140 specifically includes sub-steps S140-1 to S140-2: Step S140 - 1 : According to the mapping relationship, convert the two-dimensional metadata into standard metadata, where the standard metadata includes a domain identifier, a variable label, a data type, and a controlled terminology.

[0054] Step S140 - 2 : Generate a preliminary data set that meets the SDTM standard based on the standard metadata and the data entity.

[0055] Among them, the domain identifier is used to identify a specific domain; the variable label is used to identify a specific variable, and the label of each variable in the SDTM dataset should be unique; the data type can include character type or numeric type; controlled terminology refers to the data stored in the SDTM standard dataset, and the variable value is presented through terminology.

[0056] In some embodiments, the standard metadata also includes: variable name, source (used to describe the source of the data), role (used to determine how the variable is used in the corresponding dataset), and other necessary information for drug regulatory agencies to communicate about the variable and its content.

[0057] In this embodiment, by parsing the mapping relationship and constructing standard metadata according to the data structure requirements specified in the SDTM standard implementation guide, the data entity and the standard metadata are then integrated to generate a preliminary data set that meets the SDTM standard.

[0058] Step S150: performing a quality inspection on the preliminary data set to ensure that the data set meets the SDTM standard, and obtaining a standard data set.

[0059] After obtaining the preliminary data set, we conduct a quality check that meets the SDTM standard to ensure that the generated standard data set meets the technical specifications of the SDTM standard and obtains a high-quality standard data set. The quality check includes multi-level quality control measures such as range checking, logic verification, and consistency verification.

[0060] In an optional embodiment, the above step S150 specifically includes sub-steps S150-1 to S150-3: Step S150 - 1 : Detect the preliminary data set according to the target detection strategy, which includes: data set integrity check, data format and type verification, controlled terminology correction, logical contradiction check, and domain rule verification.

[0061] Step S150-2: When the detection triggers an error, the error information and the corresponding two-dimensional metadata are sent to the text generation model, and the process returns to step S130: the two-dimensional metadata is used as a retrieval condition, and the text generation model is used to interact with the knowledge base to perform retrieval enhancement generation, and obtain the mapping relationship between the two-dimensional metadata and the domains and variables in the standard.

[0062] Step S150 - 3 : When the detection does not trigger an error, the preliminary data set currently being detected is used as the standard data set.

[0063] The dataset integrity check includes key domain checks and mandatory field filling checks. Key domain checks are used to confirm whether core domains such as demographics and adverse events are missing. Mandatory field filling checks, for example, ensure that identification variables such as domain names and subject identifiers must exist and be non-empty, and that key variables in specific domains must be filled.

[0064] Data format and type verification includes date and time format checking and value range checking. The date and time format check converts DTC (Date / Time of Comment, time variable) into ISO 8601 format. In SDTM, if a variable ends with DTC, it indicates the time of comment, intervention, or inspection, such as AESTDTC (Adverse Event Start Date / Time of Comment), which indicates the start time of the adverse event. The value range check refers to checking whether the numerical variable is within a reasonable range. For example, age cannot be a negative number.

[0065] Controlled terminology correction is used to identify incorrect terms and replace them with standard terms, and ensure that any added variables must be declared in Define-XML (Define-XML is not generated in the last step of this method, but is initialized from the beginning of the parsing of medical clinical data in step S120 and its content is continuously improved subsequently).

[0066] Logical contradiction checking is used to identify and correct contradictions and inconsistencies in data. For example, the end date cannot be earlier than the start date, and the subject's death date must be within the study period. Domain rule validation is used to verify data according to pre-defined domain rules. For example, the LB (Laboratory Test Results) domain unit consistency check is used.

[0067] In the embodiment of the present application, after obtaining preliminary data through the above-mentioned step S140, step S150-1 is first executed to detect the preliminary data set according to the target detection strategy. If the detection triggers an error, step S150-2 is executed to send the error information and the corresponding two-dimensional metadata to the text generation model, and then return to step S130. The text generation model again provides the mapping relationship between the two-dimensional metadata and the domains and variables in the standard. Step S140 is then executed in sequence to generate a new preliminary data set, and step S150-1 is then executed to continue detecting the newly generated preliminary data set. Until the detection does not trigger an error, step S150-2 is executed to use the currently detected preliminary data set as the standard data set.

[0068] It can be understood that according to the target detection strategy, multiple types of detection are included (i.e., data set integrity check, data format and type verification, controlled terminology correction, logical contradiction check, and domain rule verification). If multiple errors are triggered, all error information and corresponding two-dimensional metadata can be sent to the text generation model for processing after each detection is completed. Alternatively, when one error is triggered, the corresponding error information and corresponding two-dimensional metadata can be sent to the text generation model for processing.

[0069] Furthermore, the above S150 further includes sub-step S150-4: Step S150-4: When the number of times the preliminary data set is tested according to the target detection strategy reaches a threshold, it is determined that there is a quality problem with the medical clinical data, and the quality problem is output to prompt that manual verification processing is required.

[0070] In the embodiment of the present application, a number threshold can be set. When the number of times the preliminary data set is tested reaches (equal to) the number threshold, step S150-4 is executed to output a quality problem, thereby prompting that manual verification is required.

[0071] Through the above implementation process, the preliminary data set obtained based on the mapping relationship transformation is quality checked to meet the SDTM standard to ensure a high-quality standard data set; thereby achieving low-cost, efficient and accurate standardized intelligent conversion of medical clinical data, reducing dependence on manual labor and high professional requirements.

[0072] In an optional embodiment, step S160 is further included after step S150: Step S160: Obtaining a generation rule for a derived variable described in natural language, wherein the derived variable represents a variable related to the clinical trial protocol and calculated based on the recorded variables; utilizing the text generation model to generate a data processing code for configuring the derived variable according to the generation rule for the derived variable; executing the data processing code according to the standard data set to generate derived variable data, and inserting the derived variable data into the standard data set to obtain a final standard data set.

[0073] In an embodiment of the present application, natural language interaction technology is used to implement intelligent configuration of clinical trial derived variables, obtaining the generation rules of derived variables described by the user in natural language (for example, "Please create a new derived variable: calculate BMI = weight (kg) / (height (m)^2), weight and height data are in the VS domain", BMI refers to Body Mass Index, and VS refers to Vital Signs). The text generation model thus automatically converts the generation rules of the derived variables (i.e., the unstructured requirements input by the user) into program code (i.e., data processing code), and intelligent configuration of the derived variables can be achieved by executing the data processing code.

[0074] In this way, based on the natural language-driven intelligent generation technology of derived variables, combined with clinical trial plans, natural language dialogue is used to dynamically generate executable data processing code, replacing traditional manual programming and solving the problem of complex and error-prone configuration of derived variables.

[0075] In an optional embodiment, step S170 is further included after step S160: Step S170: manually reviewing the final standard dataset; after confirming that the final standard dataset meets regulatory requirements, outputting the final standard dataset and a target format file, wherein the target format file records the standard metadata of the final standard dataset.

[0076] In this embodiment of the present application, the target format file is a Define-XML metadata document. After completing the conversion of the medical clinical data through steps S110 to S160, step S170 is executed. The clinical data management personnel manually review the final standard data set and finally output the final standard data set and target format file. The final standard data set adheres to the SDTM standard domain structure and variable definitions. The target format file records standard metadata including variable labels, value ranges, controlled terminology, etc.

[0077] The technical solution implemented by this application breaks through the existing limitations of single data, parses medical clinical data from different sources into two-dimensional metadata and data entities, and uses the two-dimensional metadata as a retrieval condition. The text generation model is used to interact with the knowledge base for retrieval enhancement generation, and the mapping relationship between the two-dimensional metadata and the domains and variables in the SDTM standard is obtained. The automatic and accurate mapping of the two-dimensional metadata and the domains and variables in the SDTM standard is achieved, which significantly reduces the professional threshold for manual configuration and improves standardization efficiency. The two-dimensional metadata and data entities are converted according to the mapping relationship to obtain a preliminary data set that meets the SDTM standard, and the preliminary data set is subjected to quality inspection processing that meets the SDTM standard to ensure that a high-quality standard data set is obtained. In this way, low-cost, efficient and accurate standardized intelligent conversion of medical clinical data is achieved, reducing dependence on manual labor and high professional requirements.

[0078] like Figure 2 As shown, Figure 2 This is a schematic diagram of a medical clinical data standardization conversion system based on the SDTM standard provided in an embodiment of the present application. The medical clinical data standardization conversion method based on the SDTM standard provided in an embodiment of the present application can be implemented by a medical clinical data standardization conversion system based on the SDTM standard, which includes a preprocessing module, a derivative variable generation module, a mapping execution module, a quality management module, a text generation module interface, a user dialogue interface, a text generation model and a knowledge base.

[0079] Specifically, medical clinical data from different sources (e.g., hospital information systems, electronic data acquisition systems, other structured data) are acquired through data access methods or data import methods, and the medical clinical data from different sources are parsed into two-dimensional metadata and data entities through a preprocessing module.

[0080] The text generation model and knowledge base (rule base, standard base and term base) are connected to the system through the text generation module interface, and then the two-dimensional metadata is used as the retrieval condition. The text generation model is used to search the rule base, standard base and term base respectively to obtain the target rules, target domain knowledge and target terminology that are most relevant to the current data processing task; among which, the current data processing task can be determined based on the user's experimental plan and the user's description information.

[0081] The retrieval results consisting of target rules, target domain knowledge and target terms are integrated into the system prompt words, and the two-dimensional metadata is placed into the user prompt words; the text generation model is used to generate the mapping relationship between the two-dimensional metadata and the domains and variables in the SDTM standard based on the system prompt words and user prompt words.

[0082] Then, the mapping execution module converts the two-dimensional metadata into standard metadata according to the mapping relationship; the quality management module generates a preliminary data set that meets the SDTM standard based on the standard metadata and data entities, and performs quality inspection on the preliminary data set to meet the SDTM standard to obtain the standard data set.

[0083] Afterwards, the user dialogue interface acquires the generation rules for derived variables described in natural language. Using a text generation model, the data processing code for configuring the derived variables is generated based on the generation rules. The derived variable generation module executes the data processing code against the standard dataset, generating derived variable data. This data is then inserted into the standard dataset to produce the final standard dataset. Finally, the final standard dataset is manually reviewed and, after confirmation that it meets regulatory requirements, is output along with the target format file.

[0084] In this way, the system can convert medical clinical data from different sources into standard data sets that meet the SDTM standard. The conversion process parses the medical clinical data into two-dimensional metadata and data entities through the preprocessing module, and accurately matches them to the domains and variables defined by the SDTM standard based on the predefined knowledge base; at the same time, the integrated data quality control module ensures the accuracy and completeness of the conversion results, and finally generates a standardized data set and Define-XML that can be directly submitted to regulatory agencies, thereby significantly reducing the time cost and technical threshold of traditional manual conversion methods, and providing an automated solution for the standardized conversion of multi-source heterogeneous clinical data.

[0085] The present application also provides a medical clinical data standardization conversion device based on the SDTM standard, referring to Figure 3 As shown, Figure 3 : is a schematic diagram of a medical clinical data standardization conversion device based on the SDTM standard provided in an embodiment of the present application, the device comprising: An acquisition module 310 is used to acquire medical clinical data from different sources; A parsing module 320 is configured to parse the medical clinical data into two-dimensional metadata and data entities, wherein the two-dimensional metadata represents data attribute information and feature information described in a two-dimensional table structure, and the data entities represent the actual recorded medical clinical data; Interaction module 330 is configured to use the two-dimensional metadata as a retrieval condition and utilize a text generation model to interact with a knowledge base for retrieval enhancement generation, thereby obtaining a mapping relationship between the two-dimensional metadata and domains and variables in the SDTM standard. The knowledge base is a multidimensional knowledge base for medical clinical data standardization processing, constructed based on the SDTM standard. A conversion module 340 is configured to convert the two-dimensional metadata and the data entity according to the mapping relationship to obtain a preliminary data set that meets the SDTM standard; The detection module 350 is configured to perform a quality detection process on the preliminary data set to ensure that the data set meets the SDTM standard, thereby obtaining a standard data set.

[0086] In an optional embodiment, the acquisition module is further used to acquire medical clinical data from different sources through a data access method or a data import method; Among them, medical clinical data from different sources include: daily patient medical record data from the hospital information system, electronic case report forms from the electronic data collection system, and other medical clinical data from sources other than the hospital information system and the electronic data collection system.

[0087] In an optional embodiment, the parsing module includes: A first parsing submodule is configured to parse the medical clinical data into metadata and data entities, wherein the metadata is used to describe data attribute information and feature information; The second parsing submodule is configured to convert the metadata into a standardized two-dimensional table structure according to the semantic information of the medical clinical data and the provisions of the SDTM standard to obtain the two-dimensional metadata.

[0088] In an optional embodiment, the knowledge base includes: a rule base with data processing rules and data mapping rules, a standard base integrating the SDTM standard, and a terminology base with medical terminology; the interaction module includes: a knowledge retrieval module, configured to use the two-dimensional metadata as a retrieval condition and utilize the text generation model to search the rule library, the standard library, and the term library respectively, to obtain a retrieval result consisting of target rules, target domain knowledge, and target terminology; A prompt word module, configured to integrate the search results into system prompt words, and to place the two-dimensional metadata into user prompt words; The relationship generation module is used to generate the mapping relationship based on the system prompt word and the user prompt word by using the text generation model.

[0089] In an optional embodiment, the conversion module includes: a metadata conversion module, configured to convert the two-dimensional metadata into standard metadata according to the mapping relationship, wherein the standard metadata includes a domain identifier, a variable label, a data type, and a controlled terminology; The data generation module is used to generate a preliminary data set that meets the SDTM standard based on the standard metadata and the data entity.

[0090] In an optional embodiment, the detection module includes: A first detection submodule is configured to detect the preliminary data set according to a target detection strategy, wherein the target detection strategy includes: data set integrity check, data format and type verification, controlled terminology correction, logical contradiction check, and domain rule verification; The second detection submodule is configured to, when a detection triggers an error, send the error message and corresponding two-dimensional metadata to the text generation model, and return to the execution step of using the two-dimensional metadata as a search condition, utilizing the text generation model to interact with the knowledge base to perform search enhancement generation, and obtaining a mapping relationship between the two-dimensional metadata and domains and variables in the standard; The third detection submodule is configured to use the preliminary data set currently being detected as the standard data set when no error is triggered during the detection.

[0091] In an optional embodiment, the device further includes: The prompt module is used to determine that there are quality problems in the medical clinical data when the number of times the preliminary data set is tested according to the target detection strategy reaches a number threshold, and output the quality problem to prompt that manual verification processing is required.

[0092] In an optional embodiment, the device further includes: a rule acquisition module, configured to acquire a generation rule for a derived variable described in natural language, wherein the derived variable represents a variable related to the clinical trial protocol and calculated based on the recorded variables; a code generation module, configured to generate data processing code for configuring the derived variables according to the generation rules of the derived variables by using the text generation model; The derived data module is used to execute the data processing code according to the standard data set to generate derived variable data, and insert the derived variable data into the standard data set to obtain a final standard data set.

[0093] In an optional embodiment, the device further includes: A review module, configured to manually review the final standard data set; The output module is configured to output the final standard data set and a target format file after confirming that the final standard data set meets regulatory requirements, wherein the target format file records the standard metadata of the final standard data set.

[0094] The present application also provides an electronic device, Figure 4 , Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Figure 4 As shown, the electronic device 400 includes: a memory 410 and a processor 420. The memory 410 and the processor 420 are connected via a bus communication. A computer program is stored in the memory 410, and the computer program can be run on the processor 420 to implement the steps of the medical clinical data standardization conversion method based on the SDTM standard described in the embodiment of the present application.

[0095] An embodiment of the present application also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the medical clinical data standardization conversion method based on the SDTM standard described in the embodiment of the present application are implemented.

[0096] An embodiment of the present application also provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of the medical clinical data standardization conversion method based on the SDTM standard described in the embodiment of the present application.

[0097] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0098] The embodiments of the present application are described with reference to the flowcharts and / or block diagrams of the methods and devices according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing terminal device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0099] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing terminal device to operate in a specific manner, so that the instructions stored in the computer readable memory produce a manufactured product including an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0100] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device so that a series of operating steps are executed on the computer or other programmable terminal device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable terminal device to implement the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0101] Although preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they become aware of the basic inventive concepts. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present invention.

[0102] Finally, it should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or terminal device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article, or terminal device that includes the element.

[0103] The above is a detailed introduction to a medical clinical data standardization conversion method and device based on the SDTM standard provided by this application. Specific examples are used in this article to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for general technical personnel in this field, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on this application.

Claims

1. A medical clinical data standardization conversion method based on the SDTM standard, characterized in that: The method comprises: Obtain medical clinical data from different sources; Parsing the medical clinical data into two-dimensional metadata and data entities, wherein the two-dimensional metadata represents data attribute information and feature information described in a two-dimensional table structure, and the data entity represents the actually recorded medical clinical data; Using the two-dimensional metadata as a retrieval condition, a text generation model is used to interact with a knowledge base to perform retrieval enhancement generation, thereby obtaining a mapping relationship between the two-dimensional metadata and domains and variables in the research data tabulation model SDTM standard. The knowledge base is a multidimensional knowledge base for medical clinical data standardization processing, which is constructed based on the SDTM standard; Converting the two-dimensional metadata and the data entity according to the mapping relationship to obtain a preliminary data set that meets the SDTM standard; The preliminary data set is subjected to quality inspection to ensure that it meets the SDTM standard, thereby obtaining a standard data set.

2. The medical clinical data standardization conversion method based on the SDTM standard according to claim 1 is characterized in that: Access medical clinical data from various sources, including: Obtain medical clinical data from different sources through data access or data import methods; Among them, medical clinical data from different sources include: daily patient medical record data from the hospital information system, electronic case report forms from the electronic data collection system, and other medical clinical data from sources other than the hospital information system and the electronic data collection system.

3. The medical clinical data standardization conversion method based on the SDTM standard according to claim 1 is characterized in that: Parsing the medical clinical data into two-dimensional metadata and data entities, including: Parsing the medical clinical data into metadata and data entities, wherein the metadata is used to describe data attribute information and feature information; According to the semantic information of the medical clinical data and the provisions of the SDTM standard, the metadata is converted into a standardized two-dimensional table structure to obtain the two-dimensional metadata.

4. The medical clinical data standardization conversion method based on the SDTM standard according to any one of claims 1 to 3, characterized in that: The knowledge base includes: a rule base with data processing rules and data mapping rules, a standard base integrating SDTM standards, and a term base with professional terms in the medical field; The two-dimensional metadata is used as a retrieval condition, and a text generation model is used to interact with a knowledge base to perform retrieval enhancement generation, thereby obtaining a mapping relationship between the two-dimensional metadata and domains and variables in the research data tabulation model SDTM standard, including: Using the two-dimensional metadata as a search condition, the text generation model is used to search the rule library, the standard library, and the term library respectively to obtain a search result consisting of target rules, target domain knowledge, and target terms; Integrating the search results into system prompt words, and placing the two-dimensional metadata into user prompt words; The mapping relationship is generated by utilizing the text generation model based on the system prompt words and the user prompt words.

5. The medical clinical data standardization conversion method based on the SDTM standard according to claim 1 is characterized in that: The two-dimensional metadata and the data entity are converted according to the mapping relationship to obtain a preliminary data set that meets the SDTM standard, including: converting the two-dimensional metadata into standard metadata according to the mapping relationship, wherein the standard metadata includes a domain identifier, a variable label, a data type, and a controlled terminology; A preliminary data set meeting the SDTM standard is generated according to the standard metadata and the data entity.

6. The medical clinical data standardization conversion method based on the SDTM standard according to claim 1 or 5, characterized in that: The preliminary data set is subjected to quality inspection to meet the SDTM standard to obtain a standard data set, including: Testing the preliminary data set according to a target detection strategy, wherein the target detection strategy includes: data set integrity check, data format and type verification, controlled terminology correction, logical contradiction check, and domain rule verification; In the case of a detection trigger error, the error message and the corresponding two-dimensional metadata are sent to the text generation model, and the execution step is returned: the two-dimensional metadata is used as a search condition, and the text generation model is used to interact with the knowledge base to perform search enhancement generation to obtain a mapping relationship between the two-dimensional metadata and the domains and variables in the standard; In the case that the detection does not trigger an error, the preliminary data set currently being detected is used as the standard data set.

7. The medical clinical data standardization conversion method based on the SDTM standard according to claim 6 is characterized in that: The method further comprises: When the number of times the preliminary data set is tested according to the target detection strategy reaches a number threshold, it is determined that the medical clinical data has quality problems, and the quality problems are output to prompt that manual verification processing is required.

8. The medical clinical data standardization conversion method based on the SDTM standard according to claim 1 is characterized in that: The method further comprises: Obtaining a generation rule for a derived variable described in natural language, wherein the derived variable represents a variable related to the clinical trial protocol and calculated based on the recorded variables; Using the text generation model, generating data processing code for configuring the derived variables according to the generation rules of the derived variables; The data processing code is executed according to the standard data set to generate derived variable data, and the derived variable data is inserted into the standard data set to obtain a final standard data set.

9. The medical clinical data standardization conversion method based on the SDTM standard according to claim 8, further comprising: Manually reviewing the final standard data set; After confirming that the final standard data set meets regulatory requirements, the final standard data set and a target format file are output, where the target format file records standard metadata of the final standard data set.

10. A medical clinical data standardization conversion device based on the SDTM standard, characterized in that: include: Acquisition module, used to obtain medical clinical data from different sources; a parsing module, configured to parse the medical clinical data into two-dimensional metadata and data entities, wherein the two-dimensional metadata represents data attribute information and feature information described in a two-dimensional table structure, and the data entities represent the actually recorded medical clinical data; An interactive module, configured to use the two-dimensional metadata as a retrieval condition, utilize a text generation model to interact with a knowledge base for retrieval enhancement generation, and obtain a mapping relationship between the two-dimensional metadata and domains and variables in the SDTM standard. The knowledge base is a multidimensional knowledge base for standardized processing of medical clinical data, constructed based on the SDTM standard; a conversion module, configured to convert the two-dimensional metadata and the data entity according to the mapping relationship to obtain a preliminary data set that meets the SDTM standard; The detection module is used to perform quality detection processing on the preliminary data set to ensure that the data set meets the SDTM standard, so as to obtain a standard data set.

Images