Automatic verification system for bioequivalence test data
The automated verification system for bioequivalence test data has solved the problems of inconsistent data formats and unquantified correlations, achieved standardized data integration and anomaly identification, ensured the integrity and compliance of test data, and improved the reliability and timeliness of test results.
Patent Information
- Application Number
- CN202511937136.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-22
- Publication Date
- 2026-03-24
AI Technical Summary
The current bioequivalence test data collection lacks a unified format and standardized process. The different formats of multi-source data lead to low integration efficiency. Data correlations are not screened through quantitative calculations, making it difficult to trace and verify the data logic. Data verification relies on manual or simple rule judgments, resulting in poor timeliness of updates and easily leading to unreliable test results.
By collecting BE trial data from clinical data management unit, testing instruments, and laboratory information management unit, standardizing the format, constructing a knowledge graph, defining the relationships and mapping rules between entities, calculating the association strength, generating verification result warning reports and pop-up warnings, the system achieves automated data verification and dynamic updates.
It has achieved standardized integration and traceability of bioequivalence test data, ensuring that data anomalies are identified and handled in a timely manner, guaranteeing the integrity and reliability of test data, adapting to compliance standard requirements in a timely manner, and improving the accuracy and compliance of test results.
Smart Images
Figure CN121725960A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of data service technology, specifically to an automated verification system for bioequivalence test data. Background Technology
[0002] Bioequivalence studies are a crucial step in ensuring the quality of generic drugs and the safety of clinical alternatives. The data involved includes multiple dimensions such as clinical subject information, sample test results, and detailed trial protocols. This data is stored in different management systems. As the scale of trials expands and the volume of data grows, the difficulty of data integration and management continues to increase. At the same time, regulatory agencies have put forward more stringent requirements for the authenticity, completeness, and compliance of trial data. The accuracy and standardization of data management have become important factors affecting the validity of trial results.
[0003] However, in existing technologies, the data collection of bioequivalence (BE) experiments lacks a unified standardized process. The diverse formats of multi-source data lead to low integration efficiency, data conflicts or omissions are prone to occur, and the correlation between data is not screened through quantitative calculations, so the reliability of the correlation cannot be guaranteed, making it difficult to achieve traceability and verification of data logic. Moreover, data verification mostly relies on manual or simple rule judgments, and it is impossible to identify deep data anomalies through quantitative indices. After the test compliance standards are updated, the relevant data management logic needs to be manually adjusted, resulting in poor update timeliness and risks such as unreliable test results and non-compliant data due to data deviations or compliance delays. Summary of the Invention
[0004] The purpose of this invention is to overcome the shortcomings of existing technologies and provide an automated verification system for bioequivalence trial data. This invention collects BE trial data from the clinical data management unit, testing instruments, and laboratory information management unit respectively, and forms a standardized data pool through format standardization processing, achieving standardized integration of various data. It defines core entities and their corresponding attributes, clarifies the relationships and mapping rules between entities, calculates the association strength between different entities, selects reliable entity pairs for inclusion in a knowledge graph and storage, and establishes a traceable semantic association system between data. Then, it extracts the data to be verified, calculates an anomaly judgment index based on the entity association strength, generates a verification result warning report with pop-up alerts, and simultaneously pushes anomaly warning emails to data administrators, ensuring that data anomalies are identified and handled in a timely manner, and guaranteeing the integrity and reliability of the trial data.
[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: an automated verification system for bioequivalence test data, the system comprising: Data acquisition module: Collects BE trial data from the clinical data management unit, testing instruments, and laboratory information management unit respectively, and performs format standardization processing on the collected BE trial data to form a standardized data pool with unified structure and standardized format; Knowledge graph construction module: Based on a standardized data pool, it defines the core entities and attributes of the knowledge graph, the relationships and mapping rules between entities, calculates the relationship strength between different entities using the BE data knowledge graph entity relationship strength formula, filters entities based on the relationship strength, and stores entity attributes, relationships and relationship strengths in the graph database to form a knowledge graph; Semantic reasoning verification module: Extracts data to be verified from the standardized data pool, performs basic format verification on key identifiers and removes data with format errors, calculates the anomaly judgment index of the data to be verified using the data semantic reasoning anomaly judgment index formula, makes anomaly judgment based on the anomaly judgment index, and generates a verification result warning report. Anomaly warning module: Based on the verification result warning report, a real-time pop-up warning window will appear on the operation interface, and the verification result warning report will be stored in the local database. At the same time, anomaly warning emails will be pushed to the data administrator. Collaborative Update Module: Provides automatic update functionality for the knowledge graph, dynamically updates the knowledge graph, calculates the update validity coefficient using the dynamic update validity formula of the knowledge graph, and sends a knowledge graph update completion notification through the operation interface after the update is completed.
[0006] Furthermore, in the data acquisition module, the BE trial data collected from the clinical data management unit includes: subject name, gender, age, height, weight, trial group, enrollment time, exit time, medication time, medication dosage, route of administration, and adverse reaction records after medication; the BE trial data collected from the testing instrument includes: sample collection time, sample type, collection site, collection operator, sample preservation conditions, original test values, number of test repetitions, testing time, and testing operator information; the BE trial data collected from the laboratory information management unit includes: trial name, trial purpose, trial period, subject inclusion criteria, subject exclusion criteria, testing item operation specifications, testing item judgment criteria, testing item reference range, regulatory agency name, trial compliance standard document name, specific content of the trial compliance standard document, and effective date of the trial compliance standard document.
[0007] Furthermore, the knowledge graph construction module defines the core entities and attributes of the knowledge graph, the relationships between entities, and the mapping rules. Specifically, based on a standardized data pool, the core entities of the knowledge graph are determined to be subjects, samples, testing projects, experimental protocols, and experimental compliance standards. For each core entity, corresponding attributes are extracted from three categories of data sources in the standardized data pool, and each entity is assigned a unique identifier attribute: subject number, sample barcode, project number, experimental number, and compliance standard number. Then, based on the business logic relationships of each data in the standardized data pool, the relationships between entities are defined. Based on the binding relationship of the unique identifiers of the entities, mapping rules are defined.
[0008] Furthermore, in the knowledge graph construction module, the formula for the entity association strength of the BE data knowledge graph is: ,in, For the first Class entity and the first The association strength of class entities, with values ranging from [0, 1]. For the first Class entity and the first The attribute matching degree of a class entity is determined by calculating the attribute overlap degree of entities in the standardized data pool. For the first Class entity and the first The validity of the association history of class entities is determined by statistical analysis of the first historical experiment. Class entity and the first The percentage of correct associations among class entities is determined. These are weighting coefficients, dynamically adjusted based on the entity type. Will Entity pairs that are identified as reliable related entity pairs are included in the knowledge graph. Entity pairs are identified as abnormal associations and removed. Then, the entity attributes, association relationships, and association strengths are stored in a graph database to form a knowledge graph of BE data.
[0009] Furthermore, in the semantic reasoning verification module, data to be verified containing key identifiers such as subject ID, sample barcode, project ID, trial ID, and compliance standard ID is extracted from the standardized data pool. Basic format verification is performed on the encoding format and numerical range of the key identifiers, and data with format errors is removed. Then, based on the entity association strength in the knowledge graph, combined with the association relationships and mapping rules between entities in the knowledge graph, the anomaly judgment index of the data to be verified is calculated using the data semantic reasoning anomaly judgment index formula. ,when If an anomaly is detected, a verification result warning report is generated, which records the anomaly type, anomaly data identifier, anomaly determination index and calculation process, anomaly time, and data source.
[0010] Furthermore, in the semantic reasoning verification module, the formula for the data semantic reasoning anomaly determination index is: ,in, For the first The anomaly detection index for the data to be verified, with a value range of [0, 1]. For the first Class entity and the first The strength of association between class entities, The discrepancy between the data to be verified and the reasoning results from the knowledge graph is determined by calculating the key values of the data to be verified and the key values of the reasoning from the knowledge graph. The verification threshold is set according to the scenario of the data to be verified, based on the different verification scenarios of the BE test.
[0011] Furthermore, the anomaly warning module displays a real-time pop-up warning on the operation interface based on the verification result warning report, specifically showing: BE test data anomaly warning title, abnormal data source, anomaly type, abnormal data identifier, anomaly judgment index and calculation process, and anomaly time; and stores the complete information in the verification result warning report to the local database.
[0012] Furthermore, the automatic update function in the collaborative update module specifically involves: connecting to the BE (Bio) test compliance standard database, periodically retrieving the latest test compliance standard content via API, comparing the retrieved content with the existing test compliance standards in the knowledge graph, and performing update operations after confirming the differences. This includes modifying the content of existing test compliance standards in the knowledge graph, adding test compliance standard entities that meet the test requirements, marking expired test compliance standards in the knowledge graph and removing their association with the test plan. An update log is generated synchronously during the update operation to record update information. During the update process, the update effectiveness coefficient is calculated using the dynamic update effectiveness formula of the knowledge graph. ,when Once the update is deemed valid, the knowledge graph update operation is executed, applying the content to be updated to the knowledge graph to complete the update. At the same time, a knowledge graph update completion notification is sent to the data administrator through the operation interface.
[0013] Furthermore, in the collaborative update module, the formula for the effectiveness of dynamic updates of the knowledge graph is: ,in, To update the validity coefficient, the value range is [0, 1]. To automatically update the data integrity rate, the percentage of valid data is determined by calculating the total amount of test compliance standard data retrieved via API. To determine the matching degree between the newly added test compliance standards and the existing content of the knowledge graph, the semantic similarity and structural matching degree between the entities and attributes of the newly added standards and the existing content of the knowledge graph are used. To determine the fit of internal relationships within the updated knowledge graph, the ratio of logically conflict-free relationships to the total number of relationships in the updated knowledge graph is used. , The weighting coefficient is set by the business priority of BE test data management.
[0014] Compared with existing technologies, this automated verification system for bioequivalence test data has the following advantages: I. This invention collects BE trial data from the clinical data management unit, testing instruments, and laboratory information management unit respectively, and forms a standardized data pool through format standardization processing to achieve standardized integration of various data. It defines core entities and their corresponding attributes, clarifies the relationships and mapping rules between entities, and selects reliable entity pairs for inclusion in the knowledge graph and storage by calculating the association strength between different entities, thus establishing a traceable semantic association system between data. Then, it extracts the data to be verified, calculates the anomaly judgment index based on the entity association strength, and generates a verification result warning report with pop-up warning. At the same time, it pushes anomaly warning emails to the data administrator to ensure that data anomalies are identified and handled in a timely manner, and to ensure the integrity and reliability of the trial data.
[0015] Second, this invention connects to the BE (Bioequivalence) test compliance standard database, retrieves the latest test compliance standard content through an API interface, compares the differences with the existing knowledge graph, performs an update operation, generates an update log synchronously, calculates the update validity coefficient, and applies the content to be updated to the knowledge graph after determining that the update is valid. After the update is completed, an update completion notification is sent to the administrator through the operation interface, realizing the dynamic adaptation between the knowledge graph and the test compliance standards, ensuring that the data verification logic conforms to the latest compliance requirements, and guaranteeing the compliance and timeliness of test data management.
[0016] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description
[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.
[0018] Figure 1 A flowchart of an automated verification system for bioequivalence test data; Figure 2 A framework diagram of an automated verification system for bioequivalence test data; Figure 3 This is a flowchart of the semantic reasoning verification module in an automated verification system for bioequivalence test data. Detailed Implementation
[0019] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below. Example 1:
[0020] In the scenario of verifying bioequivalence (BE) trial data for oral antihypertensive drug tablets, BE trial data is collected from the clinical data management unit, testing instruments, and laboratory information management unit. Specifically: BE trial data collected from the clinical data management unit includes subject name, gender, age, height, weight, trial group, enrollment time, exit time, medication time, dosage, route of administration, and adverse reaction records after medication; BE trial data collected from the testing instruments includes sample collection time, sample type, collection site, collection operator, sample preservation conditions, original test values, number of test repetitions, test time, and testing operator information; BE trial data collected from the laboratory information management unit includes trial name, trial purpose, trial period, subject inclusion criteria, subject exclusion criteria, test item operation specifications, test item judgment criteria, test item reference range, regulatory agency name, trial compliance standard document name, specific content of the trial compliance standard document, and effective date of the trial compliance standard document. All collected BE trial data undergoes format standardization processing, unifying key elements such as encoding format, time format, and numerical units, ultimately forming a standardized data pool with a unified structure and standardized format.
[0021] Based on a standardized data pool, a knowledge graph adapted to this bioequivalence trial is constructed. First, the core entities and attributes, inter-entity relationships, and mapping rules of the knowledge graph are defined. The core entities are identified as subjects, samples, testing items, trial protocols, and trial compliance standards. For each core entity, corresponding attributes are extracted from three categories of data sources in the standardized data pool: For the subject entity, attributes such as name, gender, age, height, weight, trial group, enrollment time, and exit time are extracted from data collected by the clinical data management unit; for the sample entity, attributes such as collection time, sample type, collection site, collection operator, and sample preservation conditions are extracted from data collected by the testing instrument; for the testing item entity, attributes such as testing standard, testing instrument model, testing reagent batch, and testing duration are extracted from data collected by the laboratory information management unit; for the trial protocol entity, attributes such as trial name, trial institution name, trial manager, subject inclusion criteria, subject exclusion criteria, and sample collection plan are extracted; and for the trial compliance standard entity, attributes such as applicable trial type, data verification requirements, standard effective time, and standard update time are extracted. Each entity is assigned a unique identifier attribute: subject number, sample barcode, item number, trial number, and compliance standard number. Then, based on the data in the standardized data pool… The business logic relationships between various data are defined to establish the relationships between entities: Based on the business logic of sample ownership in the sample collection record, a subject-sample relationship is defined; based on the business logic of the sample needing to undergo specified testing items in the test protocol, a sample-test item relationship is defined; based on the business logic of the test needing to cover specified testing items in the test protocol, a test protocol-test item relationship is defined; based on the business logic of the test needing to comply with specified compliance standards in the test compliance standard document, a test protocol-test compliance standard relationship is defined. Mapping rules are then defined based on the binding relationship of unique entity identifiers: Based on the binding field between subject number and sample barcode in the sample collection record, a subject number-sample barcode mapping rule is defined; based on the corresponding list of sample barcodes and test item numbers in the test protocol, a sample barcode-test item number mapping rule is defined; based on the association field between test number and test item list in the test protocol, a test number-test item list mapping rule is defined; based on the corresponding directory of test number and applicable compliance standard number in the test compliance standard document, a test number-applicable compliance standard number mapping rule is defined. Finally, the entity association strength between different entities is calculated using the BE data knowledge graph entity association strength formula. The BE data knowledge graph entity association strength formula is: ,in, For the first Class entity and the first The association strength of class entities, with values ranging from [0, 1]. For the first Class entity and the first The attribute matching degree of a class entity is determined by calculating the attribute overlap degree of entities in the standardized data pool. For the first Class entity and the first The validity of the association history of class entities is determined by statistical analysis of the first historical experiment. Class entity and the first The percentage of correct associations among class entities is determined. The weighting coefficient is dynamically adjusted based on the entity type; Entity pairs that are identified as reliable related entity pairs are included in the knowledge graph. Entity pairs are identified as abnormal associations and removed. Then, the entity attributes, association relationships, and association strengths are stored in a graph database to form a knowledge graph of the BE data, such as... Figure 1 As shown.
[0022] Data to be verified, containing key identifiers such as subject ID, sample barcode, project ID, trial ID, and compliance standard ID, is extracted from a standardized data pool. Basic format verification is performed on the encoding format and numerical range of these key identifiers, and data with incorrect formats is removed to ensure the accuracy of subsequent verification. For data that passes the basic format verification, an anomaly index is calculated based on the entity association strength in the knowledge graph using the data semantic reasoning anomaly index formula. The data semantic reasoning anomaly index formula is as follows: ,in, For the first The anomaly detection index for the data to be verified, with a value range of [0, 1]. For the first Class entity and the first The strength of association between class entities, The discrepancy between the data to be verified and the reasoning results from the knowledge graph is determined by calculating the key values of the data to be verified and the key values of the reasoning from the knowledge graph. The verification threshold for the scenario to which the data to be verified belongs is set according to the different verification scenarios of the BE experiment; anomaly detection is performed based on the anomaly detection index, when... When it is judged as abnormal, Data is judged to be normal at any time. After all the data to be verified is judged, a verification result warning report is generated. The verification result warning report records in detail the key information such as the anomaly type, the anomaly data identifier, the anomaly judgment index and calculation process, the anomaly time, and the data source, providing a clear basis for subsequent anomaly handling.
[0023] Based on the generated verification result warning report, a real-time pop-up warning appears on the operation interface. The pop-up displays the title of the BE test data anomaly warning, the source of the abnormal data, the type of abnormality, the identifier of the abnormal data, the anomaly judgment index and calculation process, and the time of the abnormality, allowing users to intuitively understand the abnormal situation. At the same time, the complete information in the verification result warning report is stored in the local database to ensure the traceability and security of the data, and anomaly warning emails are pushed to the data administrator to remind the data administrator to promptly verify and process the abnormal data.
[0024] The system provides an automatic update function for the knowledge graph, connecting to the test compliance standard database for BE (Business Application) testing. It retrieves the latest test compliance standards via API at preset fixed intervals. After retrieval, it compares the retrieved content with the existing test compliance standards in the knowledge graph. Once the differences are confirmed, it performs an update operation, including modifying the content of existing test compliance standards in the knowledge graph, adding test compliance standard entities that meet test requirements, marking expired test compliance standards in the knowledge graph and removing their association with the test plan. An update log is generated synchronously during the update operation, recording detailed update information. Furthermore, during the update process, the update effectiveness coefficient is calculated using the knowledge graph dynamic update effectiveness formula. The knowledge graph dynamic update effectiveness formula is: ,in, To update the validity coefficient, the value range is [0, 1]. To automatically update the data integrity rate, the percentage of valid data is determined by calculating the total amount of test compliance standard data retrieved via API. To determine the matching degree between the newly added test compliance standards and the existing content of the knowledge graph, the semantic similarity and structural matching degree between the entities and attributes of the newly added standards and the existing content of the knowledge graph are used. To determine the fit of internal relationships within the updated knowledge graph, the ratio of logically conflict-free relationships to the total number of relationships in the updated knowledge graph is used. , As a weighting coefficient, it is set by the business priority of BE trial data management; when Once the update is deemed valid, the knowledge graph update operation is executed, applying the content to be updated to the knowledge graph to complete the update. After all updates are completed, a knowledge graph update completion notification is sent to the data administrator through the operation interface, allowing the data administrator to be informed of the update status in a timely manner.
[0025] In summary, in the scenario of verifying bioequivalence (BE) test data for oral antihypertensive drug tablets, a standardized data pool is formed by collecting BE test data from three types of data sources and standardizing the format. The standardized data pool defines core entities, attributes, relationships, and mapping rules. After filtering using the entity association strength formula of the BE data knowledge graph, a knowledge graph is constructed. Then, the data to be verified is extracted to complete basic format verification and anomaly detection, generating a verification result warning report. Warnings are then implemented through pop-ups, local storage, and email push notifications. Finally, the knowledge graph is dynamically updated and notifications are sent by connecting to a compliance standard database, ensuring the standardization and efficiency of test data verification throughout the entire process. Example 2:
[0026] In the scenario of verifying bioequivalence (BE) trial data for injectable antibiotics, BE trial data is collected from the clinical data management unit, testing instruments, and laboratory information management unit. Specifically: BE trial data provided by the clinical data management unit includes subject name, gender, age, height, weight, trial group, enrollment time, exit time, medication time, dosage, route of administration, and adverse reaction records; BE trial data provided by the testing instruments includes sample collection time, sample type, collection site, collection operator, sample preservation conditions, raw test values, number of test repetitions, testing time, and operator information; BE trial data provided by the laboratory information management unit includes trial name, trial purpose, trial duration, subject inclusion criteria, subject exclusion criteria, testing procedure specifications, testing criteria, testing reference range, regulatory agency name, trial compliance standard document name, specific content of the trial compliance standard document, and effective date of the trial compliance standard document. These collected BE trial data undergo format standardization processing, unifying the encoding format, time format, and numerical units to eliminate data format differences, ultimately forming a standardized data pool with a unified structure and standardized format, such as... Figure 2 As shown.
[0027] The knowledge graph construction was carried out based on a standardized data pool. First, the core entities and attributes, relationships, and mapping rules between entities were defined. The core entities were identified as subjects, samples, testing items, trial protocols, and trial compliance standards. Attributes corresponding to each core entity were extracted from three categories of data sources in the standardized data pool: For the subject entity, attributes such as name, gender, age, height, weight, trial group, enrollment time, and exit time were extracted from data collected by the clinical data management unit; for the sample entity, attributes such as collection time, sample type, collection site, collection operator, and sample preservation conditions were extracted from data collected by the testing instrument; for the testing item entity, attributes such as testing standard, testing instrument model, testing reagent batch, and testing duration were extracted from data collected by the laboratory information management unit; for the trial protocol entity, attributes such as trial name, trial institution name, trial manager, subject inclusion criteria, subject exclusion criteria, and sample collection plan were extracted; and for the trial compliance standard entity, attributes such as applicable trial type, data verification requirements, standard effective time, and standard update time were extracted. Simultaneously, unique identifier attributes were assigned to each entity, namely subject number, sample barcode, item number, trial number, and compliance standard number. Furthermore, the attributes were determined based on the business logic of each data point in the standardized data pool. Business logic associations are defined to establish relationships between entities: Subject-sample association is defined based on the business logic of sample attribution in the sample collection record; sample-test item association is defined based on the business logic of the required tests in the test protocol; test protocol-test item association is defined based on the business logic of the test covering the required tests in the test protocol; test protocol-test compliance standard association is defined based on the business logic of the test complying with the required compliance standards in the test compliance standard document. Mapping rules are defined based on the binding relationship of unique entity identifiers: Subject number-sample barcode mapping rules are defined based on the binding field between subject number and sample barcode in the sample collection record; sample barcode-test item number mapping rules are defined based on the corresponding list of sample barcodes and test item numbers in the test protocol; test number-test item list mapping rules are defined based on the association field between test number and test item list in the test protocol; test number-applicable compliance standard number mapping rules are defined based on the corresponding directory of test number and applicable compliance standard number in the test compliance standard document. Finally, the association strength between different entities is calculated using the BE data knowledge graph entity association strength formula, which is: ;Will Entity pairs that are identified as reliable related entity pairs are included in the knowledge graph. Entity pairs are identified as abnormal associations and removed. Then, the entity attributes, association relationships, and association strengths are stored in a graph database to form a knowledge graph of BE data.
[0028] Data to be verified, containing key identifiers such as subject ID, sample barcode, project ID, trial ID, and compliance standard ID, is extracted from the standardized data pool. First, a basic format verification is performed on the encoding format and numerical range of these key identifiers. The encoding is carefully checked to ensure it conforms to preset specifications and the values are within reasonable ranges. Data with incorrect formats is directly discarded to avoid erroneous data affecting the verification results. Figure 3 As shown, for the data to be verified that has passed the basic format validation, the anomaly index of each data point is calculated based on the entity association strength in the knowledge graph using the data semantic reasoning anomaly index formula. The data semantic reasoning anomaly index formula is as follows: Anomaly detection is performed based on the anomaly detection index. When it is judged as abnormal, Data is judged to be normal at any time; after all data verification is completed, a verification result warning report is generated. The verification result warning report fully records the anomaly type, anomaly data identifier, anomaly judgment index and calculation process, anomaly time, and data source, providing comprehensive support for the subsequent verification and processing of anomaly data.
[0029] Based on the verification result warning report, a real-time pop-up warning appears on the operation interface. The pop-up clearly displays the title of the BE test data anomaly warning, the source of the abnormal data, the type of anomaly, the identifier of the abnormal data, the anomaly judgment index and calculation process, and the time of the anomaly, so as to quickly obtain the core information of the anomaly. At the same time, the complete information in the verification result warning report is securely stored in the local database to ensure that the data is not lost and is traceable. An anomaly warning email is also pushed to the data administrator to promptly remind the data administrator to pay attention to the abnormal data and carry out verification and processing work as soon as possible to avoid the abnormal data from adversely affecting the test process.
[0030] The system provides automatic update functionality for the knowledge graph, continuously connecting to the BE (Business Application) test compliance standard database. It retrieves the latest test compliance standards via API at pre-set intervals, compares the retrieved content with existing test compliance standards in the knowledge graph, and executes update operations upon confirmation. These operations include modifying existing test compliance standards in the knowledge graph, adding test compliance standard entities that meet test requirements, marking expired test compliance standards in the knowledge graph and removing their association with the test plan. Update logs are generated synchronously during the update process, recording detailed update information. Furthermore, the update effectiveness coefficient is calculated using the knowledge graph dynamic update effectiveness formula, which is: ;when Once the update is deemed valid, a knowledge graph update operation is performed, applying all the content to be updated to the knowledge graph to complete the dynamic update of the knowledge graph. After the update is completed, a knowledge graph update completion notification is sent to the data administrator through the operation interface to ensure that the data administrator is aware of the latest status of the knowledge graph in a timely manner.
[0031] In summary, in the scenario of bioequivalence trial data verification for injectable antibiotics, BE trial data is collected from the clinical data management unit, testing instrument and laboratory information management unit and processed for format standardization. Based on the standardized data pool, core entities and attributes such as subjects and samples, as well as their relationships and mapping rules, are identified and a knowledge graph is formed after filtering by association strength. Data to be verified containing key identifiers is extracted from the standardized data pool, basic format verification and anomaly detection are performed, and a verification result warning report is generated. Anomaly information is pushed through pop-up windows in the operation interface, local database storage, and emails to the data administrator. The latest trial compliance standards are retrieved regularly, and the knowledge graph is dynamically updated after being determined to be valid by the dynamic update validity formula of the knowledge graph. An update completion notification is sent to the data administrator simultaneously, comprehensively ensuring the compliance and verification accuracy of the trial data.
[0032] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.
Claims
1. An automated verification system for bioequivalence test data, characterized in that, The system includes: Data acquisition module: Collects BE trial data from the clinical data management unit, testing instruments, and laboratory information management unit respectively, and performs format standardization processing on the collected BE trial data to form a standardized data pool; Knowledge graph construction module: Based on a standardized data pool, it defines the core entities and attributes of the knowledge graph, the relationships and mapping rules between entities, calculates the relationship strength between different entities using the BE data knowledge graph entity relationship strength formula, filters entities based on the relationship strength, and stores entity attributes, relationships and relationship strengths in the graph database to form a knowledge graph; Semantic reasoning verification module: Extracts data to be verified from the standardized data pool, performs basic format verification on key identifiers and removes data with format errors, calculates the anomaly judgment index of the data to be verified using the data semantic reasoning anomaly judgment index formula, makes anomaly judgment based on the anomaly judgment index, and generates a verification result warning report. Anomaly warning module: Based on the verification result warning report, a real-time pop-up warning window will appear on the operation interface, and the verification result warning report will be stored in the local database. At the same time, anomaly warning emails will be pushed to the data administrator. Collaborative Update Module: Provides automatic update functionality for the knowledge graph, dynamically updates the knowledge graph, calculates the update validity coefficient using the dynamic update validity formula of the knowledge graph, and sends a knowledge graph update completion notification through the operation interface after the update is completed.
2. The automated verification system for bioequivalence test data according to claim 1, characterized in that, In the data acquisition module, the BE trial data collected from the clinical data management unit includes: subject name, gender, age, height, weight, trial group, enrollment time, exit time, medication time, medication dosage, route of administration, and adverse reaction records after medication; the BE trial data collected from the testing instrument includes: sample collection time, sample type, collection site, collection operator, sample preservation conditions, original test values, number of test repetitions, testing time, and testing operator information; the BE trial data collected from the laboratory information management unit includes: trial name, trial purpose, trial period, subject inclusion criteria, subject exclusion criteria, testing item operation specifications, testing item judgment criteria, testing item reference range, regulatory agency name, trial compliance standard document name, specific content of the trial compliance standard document, and effective date of the trial compliance standard document.
3. The automated verification system for bioequivalence test data according to claim 1, characterized in that, The knowledge graph construction module defines the core entities and attributes of the knowledge graph, as well as the relationships and mapping rules between entities. Specifically, based on a standardized data pool, the core entities of the knowledge graph are determined to be subjects, samples, testing projects, experimental protocols, and experimental compliance standards. For each core entity, corresponding attributes are extracted from three categories of data sources in the standardized data pool, and each entity is assigned a unique identifier attribute: subject number, sample barcode, project number, experimental number, and compliance standard number. Then, based on the business logic relationships of each data in the standardized data pool, the relationships between entities are defined. Based on the binding relationships of the unique identifiers of entities, mapping rules are defined.
4. The automated verification system for bioequivalence test data according to claim 1, characterized in that, In the knowledge graph construction module, the formula for the entity association strength of the BE data knowledge graph is: ,in, For the first Class entity and the first The strength of association between class entities, For the first Class entity and the first Attribute matching degree of class entities, For the first Class entity and the first Validity of association history of class entities These are the weighting coefficients; Will Entity pairs that are identified as reliable related entity pairs are included in the knowledge graph. Entity pairs are identified as abnormal associations and removed. Then, the entity attributes, association relationships, and association strengths are stored in a graph database to form a knowledge graph of BE data.
5. The automated verification system for bioequivalence test data according to claim 1, characterized in that, In the semantic reasoning verification module, data to be verified containing key identifiers such as subject ID, sample barcode, project ID, trial ID, and compliance standard ID is extracted from the standardized data pool. Basic format verification is performed on the encoding format and numerical range of the key identifiers, and data with format errors is removed. Then, based on the entity association strength in the knowledge graph, combined with the association relationships and mapping rules between entities in the knowledge graph, the anomaly judgment index of the data to be verified is calculated using the data semantic reasoning anomaly judgment index formula. ,when If an anomaly is detected, a verification result warning report is generated, which records the anomaly type, anomaly data identifier, anomaly determination index and calculation process, anomaly time, and data source.
6. The automated verification system for bioequivalence test data according to claim 5, characterized in that, In the semantic reasoning verification module, the formula for the data semantic reasoning anomaly determination index is: ,in, For the first Anomaly detection index for each piece of data to be verified. For the first Class entity and the first The strength of association between class entities, To determine the consistency deviation between the data to be verified and the reasoning results from the knowledge graph, This is the verification threshold for the scenario to which the data to be verified belongs.
7. The automated verification system for bioequivalence test data according to claim 1, characterized in that, In the aforementioned anomaly warning module, based on the verification result warning report, a real-time pop-up warning window appears on the operation interface, specifically displaying: BE test data anomaly warning title, abnormal data source, anomaly type, abnormal data identifier, anomaly judgment index and calculation process, and anomaly time; and storing the complete information in the verification result warning report to the local database.
8. The automated verification system for bioequivalence test data according to claim 1, characterized in that, The automatic update function in the collaborative update module is as follows: it connects to the test compliance standard database of BE test, periodically retrieves the latest test compliance standard content through the API interface, compares the retrieved content with the existing test compliance standards in the knowledge graph, and performs update operations after confirming the differences. This includes modifying the content of existing test compliance standards in the knowledge graph, adding test compliance standard entities that meet the test requirements, marking expired test compliance standards in the knowledge graph and removing their association with the test plan. An update log is generated synchronously during the update operation to record the update information. During the update process, the update effectiveness coefficient is calculated using a dynamic update effectiveness formula based on the knowledge graph. ,when Once the update is deemed valid, the knowledge graph update operation is executed, applying the content to be updated to the knowledge graph to complete the update. At the same time, a knowledge graph update completion notification is sent to the data administrator through the operation interface.
9. The automated verification system for bioequivalence test data according to claim 8, characterized in that, In the collaborative update module, the formula for the effectiveness of dynamic updates of the knowledge graph is: ,in, To update the validity coefficient, To automatically update the data integrity rate, To ensure the new test compliance standards match the existing content of the knowledge graph, To improve the fit of internal associations within the updated knowledge graph, , These are the weighting coefficients.