Medical data multi-source fusion verification and correction method and system

Abstract

The invention belongs to the technical field of medical data processing, and discloses a medical data multi-source fusion verification and correction method and system, and the method comprises the steps: obtaining multi-source data; constructing a dynamic baseline model of the emergency department, the ICU and the common ward, responding to and verifying the abnormal condition of the patient data, and performing early warning; establishing a term library, carrying out multi-source fusion on the patient data obtained from multiple sources, and carrying out data correction; and generating a multi-source data quality evaluation table of different departments based on the whole-process data, and feeding back and updating the term library and fusing the correction weight based on the quality evaluation table. Through a real-time monitoring and intelligent early warning mechanism, quality problems in the data can be found and corrected in time, human errors are reduced, and the data management cost is greatly reduced. Through multi-source data fusion and consistency verification, the system can ensure seamless connection of data from different data sources, and cross-system and cross-platform data consistency and cooperative work are realized, so that the accuracy of medical decision and the reliability of data are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of medical data processing, and specifically relates to a method and system for multi-source fusion verification and correction of medical data. Background Art

[0002] With the development of informatization in the medical industry, various medical data have been widely collected and stored, and the types and sources of data are increasing continuously. The accuracy, consistency, and integrity of medical data directly affect patients' diagnosis and treatment decisions and health management. However, the current medical data management faces the following challenges: 1. Diverse data sources and complex formats: Medical data come from multiple systems such as HIS, EMR, LIMS, etc., and the data formats, structures, and storage methods are different, resulting in difficulties in data fusion.

[0003] 2. Prominent data quality problems: Due to the complex processes of medical data collection and processing, data often have problems such as missing, duplicate, incorrect, or inconsistent formats, affecting the usability of data and the accuracy of analysis results.

[0004] 3. Lack of automated data correction mechanism: Most existing systems rely on manual intervention to handle data problems, with low efficiency, high error rates, and difficulty in meeting the real-time processing requirements of large-scale data.

[0005] 4. Difficulties in cross-system data consistency verification: There is a lack of unified data standards and coordination mechanisms between multiple data sources, resulting in data conflicts and inconsistencies easily occurring during data exchange between different systems.

[0006] Therefore, there is an urgent need for a system that can monitor data quality in a multi-source data environment and improve data quality and integration efficiency through intelligent automated correction and consistency verification methods. Summary of the Invention

[0007] In view of this, the present invention provides a method and system for multi-source fusion verification and correction of medical data to solve the above problems.

[0008] To solve the above technical problems, the present invention provides a method for multi-source fusion verification and correction of medical data, including: Obtaining multi-source data, and after standardized preprocessing, assigning trust weights to each data based on different sources; Constructing dynamic baseline models for the emergency department, ICU, and general ward, establishing the association relationship between patient data and model prediction data, and verifying and warning abnormal situations of patient data through data flow response; Build a thesaurus that includes both international standard terms and regionally common terms. Match the patient data obtained from multiple sources with terms to generate standard terms, and then perform multi-source fusion. Based on the fusion results, perform data correction. Generate a multi-source data quality evaluation form for different departments based on the full-process data, including completeness rate, standardization rate, conflict resolution rate, and clinical recognition rate. Update the thesaurus and fusion correction weights based on the feedback from the quality evaluation form.

[0009] As an optional method, when performing standardization processing, assign unified label encodings to the original data through a pre-loaded lightweight medical knowledge graph.

[0010] As an optional method, Build dynamic baseline models for the emergency department, ICU, and general ward, including setting baselines that conform to the data sources for each type of patient data, which are used as the standard ranges for anomaly alerts. Establish the association between patient data and model prediction data, including: inputting the timestamp, department, personal information, and data type of the patient data, and predicting the reasonable fluctuation range of the corresponding data through an LSTM neural network.

[0011] As an optional method, verify the anomalies of patient data through data stream responses, including: After dividing the data sources by department, visualize the input patient data information using a data stream platform. Connect the data stream platform to the dynamic baselines of different departments, monitor whether there are any indicators in the patient data that exceed the preset thresholds, and call the medical knowledge base API to verify the data and logical consistency in real time, and generate warning strategies.

[0012] As an optional method, warnings include: Set the first warning strategy to give audible and visual alerts for data anomalies that directly endanger the safety of diagnosis and treatment. Set the second warning strategy to generate to-do tasks for data anomalies that affect data quality but are not urgent and send them to the quality control port. Set the third warning strategy to mark semantic risks and generate logs.

[0013] As an optional method, assign trust weights to add source authority scores to each piece of data. Match the patient data obtained from multiple sources with terms to generate standard terms for unified term expression, including: Extract the text fields in the patient data and use a pre-trained model for matching and recognition. Term matching includes exact matching and fuzzy matching. Exact matching is used to directly extract the text information in the data that matches the thesaurus. Fuzzy matching maps incorrect and / or abbreviated regionally common terms to international standard terms. Among them, if it is detected that any regionally common term accumulates a preset number of times and is not mapped, dynamically adjust the term.

[0014] As an alternative, before outputting the matching recognition results, semantic conflict recognition and dynamic priority adjustment are also included; semantic conflict recognition is used to record the conflicts of text data; The dynamic priority adjustment is carried out by the following method: Set the weight dimensions, including source authority, time decay, and operator qualification; Set the conflict resolution rules, including: if the difference between the total weight of any data source and that of other sources is greater than the preset value, it is directly adopted; if the difference between the total weight of any data source and that of other sources is less than the preset value, manual review is carried out, and the conflict type and conflict reason are recorded.

[0015] As an alternative, the correction includes: Filling in the missing data based on similar records, converting the data that does not conform to the format into the standard format, identifying the patient through the unique identifier and deleting the duplicate records, and then setting the hierarchical correction strategies for low-risk conflicts, medium-risk conflicts, and high-risk conflicts; The low-risk conflict correction is used to correct the differences in term expressions between different systems, but for the conflicts with consistent semantics; The medium-risk conflict correction is used to correct the conflicts with numerical contradictions but logical inferability; its correction weight is determined based on the weight dimensions of source authority, time decay, and operator qualification; The high-risk conflict correction is used to correct the conflicting data that directly affects the safety of diagnosis or treatment, and immediately issue a warning and freeze the internal process of the current doctor's order; After completing the correction at any level, record the snapshot of the original conflict data and the basis for the correction decision into the database.

[0016] As an alternative, the completeness rate is used to calculate the proportion of missing data, which is used to represent the degree of data completion; the standardization rate is used to calculate the proportion of actual terms that conform to international standard terms; the conflict resolution rate is used to represent the proportion of automatically resolved conflicts without human participation in the total conflicts, and the clinical recognition rate is used to sample and obtain the adoption proportion of doctors for the correction results.

[0017] On the other hand, the present invention also provides a multi-source fusion verification and correction system for medical data, including: A data preprocessing module, which is used to obtain multi-source data, and after standard preprocessing, assign a trust weight to each data based on different sources; A data monitoring module, which is used to construct dynamic baseline models for the emergency department, ICU, and general ward, establish the association relationship between patient data and model prediction data, and verify the abnormal conditions of patient data through data stream response; A knowledge base and fusion correction module, which is used to establish a terminology database for international standard terms and regionally common terms, perform term matching on patient data obtained from multiple sources to generate standard terms, then conduct multi-source fusion, and perform data correction based on the fusion results; A feedback optimization module, which is used to generate a multi-source data quality evaluation form including completeness rate, standardization rate, conflict resolution rate, and clinical recognition rate for different departments based on the full-process data, and update the terminology database and the fusion correction system based on the quality evaluation form.

[0018] The beneficial effects of the present invention are as follows: Through the real-time monitoring and intelligent early warning mechanism, the present invention can timely detect and correct quality problems in the data, reduce human errors and significantly reduce the data management cost. At the same time, the automated verification and correction functions improve the data processing efficiency, making data management more efficient. Through multi-source data fusion and consistency verification, the system can ensure seamless docking of data from different data sources, achieve cross-system and cross-platform data consistency and collaborative work, thereby improving the accuracy of medical decisions and the reliability of data. Description of the Drawings

[0019] Figure 1 It is a schematic flowchart of a method for multi-source fusion verification and correction of medical data provided by an embodiment of the present invention. Detailed Embodiments

[0020] In order to enable those skilled in the art to better understand the technical solutions of the present invention, the present invention will be further described in detail below in conjunction with the specific embodiments.

[0021] Please refer to Figure 1 , this embodiment provides a method for multi-source fusion verification and correction of medical data, including: obtaining multi-source data, assigning trust weights to each data based on different sources after standardization preprocessing; constructing dynamic baseline models for the emergency department, ICU, and general ward, establishing the association relationship between patient data and model prediction data, and verifying and warning abnormal conditions of patient data through data flow response; establishing a terminology database for international standard terms and regionally common terms, performing term matching on patient data obtained from multiple sources to generate standard terms, then conducting multi-source fusion, and performing data correction based on the fusion results; generating a multi-source data quality evaluation form including completeness rate, standardization rate, conflict resolution rate, and clinical recognition rate for different departments based on the full-process data, and updating the terminology database and the fusion correction weights based on the quality evaluation form.

[0022] In this embodiment, data is first obtained from multiple medical data sources such as hospital information systems, electronic medical record systems, laboratory information management systems, etc. The data of each data source is usually stored in different formats and structures. For example, a hospital information system may store basic patient information, from which this embodiment obtains patient registration and basic file information; an electronic medical record system stores medical records, from which this embodiment obtains diagnostic records, doctor's orders, etc.; a laboratory information management system stores laboratory test data, from which this embodiment obtains test reports, imaging data, etc. In addition, real-time blood pressure / oxygen saturation of patients can also be obtained through devices connected to the Internet of Things such as vital sign monitors.

[0023] After collecting the data, data preprocessing is first performed to unify the data format, clean the data, and remove invalid information. Then, in an optional scenario, a medical knowledge graph constructed by the International Classification of Diseases standard code ICD-10 and the Systematized Nomenclature of Medicine - Clinical Terms is pre-loaded to assign a unified label encoding to the original data. For example, myocardial infarction is automatically mapped to the encoding of myocardial infarction, so that the data initially maintains the same standard. Then weights are assigned based on the source. This embodiment sets from three dimensions: source authority, time decay, and operator qualification. Optionally, among source authorities, the weight of a tertiary hospital is greater than that of a community hospital; in terms of time decay, the weight of recently calibrated is greater than that of not recently calibrated; among operator qualifications, the weight of a doctor is greater than that of a nurse. The specific settings are not limited in this embodiment.

[0024] Next, this embodiment begins to establish a dynamic quality assessment standard to solve the problem that traditional fixed thresholds cannot adapt to the variability of medical scenarios.

[0025] Baseline alarm is one of the most important functions of most database monitoring software currently. The implementation of a dynamic baseline is to solve the problem of the combined fluctuation range of certain indicators during different business loads during the day and night. Considering the specificity of different departments, this embodiment constructs a dynamic baseline model based on the differences between the emergency department, ICU, and general ward, establishes the correlation between patient data and model prediction data, and conducts it by verifying the abnormal situation of patient data through data flow response and giving early warnings. This is because generally, the reasonable range of blood pressure in the ICU is: systolic blood pressure 70 - 160 mmHg, while the reasonable range of blood pressure in the general ward is: 90 - 140 mmHg. Conflicts are likely to occur during data fusion.

[0026] Subsequently, data such as historical data timestamps, departments, patient ages, and test items can be input, and the LSTM neural network is used to capture diurnal / seasonal patterns to output a reasonable range of predicted values. This embodiment does not limit the specific data types and data stream software. Any common data stream software on the market can be selected for monitoring, and the numerical anomalies and logical conflicts are verified by responding to the data stream. Optionally, the monitoring types are divided by department, and the above-generated dynamic baseline is docked to monitor whether the indicators exceed the threshold. Optionally, considering that in some cases, the blood glucose data may conflict with the diabetes history due to fluctuations, the existing knowledge graph API can also be called to verify the logical consistency of the data in real time.

[0027] Among them, this embodiment sets up a conflict warning classification mechanism, including setting the first warning strategy to give an audible and visual prompt for data anomalies such as blood potassium > 6.5 mmol / L that directly endanger the safety of diagnosis and treatment, setting the second warning strategy to generate a to-do task for data anomalies such as the lack of fields such as gender, medical history, and medical records, which affect data quality but are not urgent, and sending it to the quality control port, and setting the third warning strategy to mark and generate a log for terms that are not strictly unified but can be inferred logically and content with semantic risks.

[0028] The fusion process of this embodiment includes: Assigning trust weights to add source authority scores to each piece of data, and performing term matching on patient data obtained from multiple sources to generate standard terms for unified term expression, which includes: extracting text fields in patient data and using a pre-trained model for both exact matching and fuzzy matching. Exact matching is used to directly extract text information in the data that matches the term library; fuzzy matching maps common terms in error and / or abbreviated areas to international standard terms (such as writing "esophagus" as "Si Dao", handling spelling mistakes or dialects such as "chronic bronchitis" mapped to "chronic obstructive pulmonary disease J44.9", "liver cancer" mapped to "C22.0 malignant liver tumor", "chemotherapy" mapped to "Z51.1 neoplasm chemotherapy"); among them, if it is detected that any common term in the area accumulates a preset number of times and is not mapped, the term is dynamically adjusted. This can overcome the problem that conventional text matching cannot solve the expression differences.

[0029] Before outputting the matching recognition result, it also includes semantic conflict recognition and dynamic priority adjustment; semantic conflict recognition is used to record the conflicts in text data; the dynamic priority adjustment is adjusted by the following method: Set weight dimensions, including source authority, time decay, and operator qualification; set conflict resolution rules, including: if the total weight score of any data source differs from that of other sources by more than a preset value, it is directly adopted; if the total weight score of any data source differs from that of other sources by less than a preset value, manual review is performed, and the conflict type and conflict reason are recorded.

[0030] After data fusion is completed, this embodiment starts the correction: First, fill in the missing data based on similar records, convert the data that does not conform to the format into the standard format, identify patients through unique identifiers and delete duplicate records to ensure that the data is complete and in the same dimension. Then set the hierarchical correction strategies for low-risk conflicts, medium-risk conflicts, and high-risk conflicts: Based on the terminology standardization conflict, set the low-risk conflict correction to correct the conflicts with different term expressions but the same semantics between different systems; Based on the problem that there are differences in records of different systems but the rationality can be inferred through associated data, set the medium-risk conflict correction to correct the conflicts with numerical contradictions but logical inferability; its correction weight is determined based on the weight dimensions of source authority, time decay, and operator qualification; Based on the key indicator conflict, set the high-risk conflict correction to correct the conflicting data that directly affects the safety of diagnosis or treatment, immediately issue a warning and freeze the internal process of the current medical order. This high-risk conflict is because the allergy history and allergens recorded in different systems are inconsistent due to patient expression or doctor recording problems, which is likely to lead to unexpected situations in later treatment. After completing the correction at any level, record the snapshot of the original conflict data and the basis for the correction decision in the database. Through the above correction, this embodiment significantly improves the clinical usability of the solution by reducing technical complexity, stratifying strategies, and providing transparent traceability. Finally, this embodiment also sets up a multi-source data quality evaluation form for different departments, including the completeness rate, standardization rate, conflict resolution rate, and clinical recognition rate. The completeness rate is used to calculate the proportion of missing data and represents the degree of data completion; the standardization rate is used to calculate the proportion of actual terms that conform to international standard terms; the conflict resolution rate is used to represent the proportion of automatically resolved conflicts without manual participation in the total conflicts, and the clinical recognition rate is used to sample the proportion of doctors' adoption of the correction results. It is convenient to generate reports for the unit's reference and can also feedback to drive the iterative update of the terminology library. For example, when a local term such as "chronic bronchitis" is manually corrected by a doctor ≥5 times, it is automatically upgraded to a standard term alias. In this way, through feedback and new data, continuously optimize the machine learning model and improve the accuracy of data quality monitoring, correction, and consistency verification. For example, optimize the outlier detection model through historical data to improve the system's recognition ability for new data.

[0031] On the other hand, this embodiment also provides a multi-source fusion verification and correction system for medical data, including: A data preprocessing module for obtaining multi-source data, performing standardized preprocessing, and assigning trust weights to each data based on different sources; The data monitoring module is used to build a dynamic baseline model for the emergency department, ICU and general ward, establish the correlation between patient data and model prediction data, and verify the abnormal conditions of patient data through data flow response. The knowledge base and fusion correction module is used to establish a terminology library for international standard terms and regional common terms, perform term matching on patient data obtained from multiple sources to generate standard terms and then conduct multi-source fusion, and perform data correction based on the fusion result. The feedback optimization module is used to generate a multi-source data quality evaluation form for different departments including completeness rate, standardization rate, conflict resolution rate and clinical recognition rate based on the whole-process data, and update the terminology library and fusion correction system based on the quality evaluation form.

[0032] Through the above solution, in this embodiment, through the real-time monitoring and intelligent early warning mechanism, the quality problems in the data can be discovered and corrected in time, reducing human errors and greatly reducing the data management cost. At the same time, the automated verification and correction functions improve the data processing efficiency, making data management more efficient. Through multi-source data fusion and consistency verification, it can ensure the seamless docking of data from different data sources, realize cross-system and cross-platform data consistency and collaborative work, thereby improving the accuracy of medical decision-making and the reliability of data. It also continuously optimizes the data quality management process through machine learning algorithms, enabling the system to adapt to complex medical data management requirements and gradually improving its intelligent level. With the continuous use of the system, the data quality monitoring and correction rules can be continuously optimized and adjusted to improve the processing ability of new data.

[0033] The above is only the preferred embodiment of the present invention. It should be noted that the above preferred embodiment should not be regarded as a limitation of the present invention. The protection scope of the present invention should be subject to the scope defined by the claims. For those of ordinary skill in the art of this technology, without departing from the spirit and scope of the present invention, several improvements and refinements can still be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A medical data multi-source fusion verification and correction method, characterized in that: include: Obtain multi-source data, standardize and pre-process it, and assign trust weights to each data based on different sources; Build dynamic baseline models for the emergency department, ICU, and general wards, establish the relationship between patient data and model prediction data, and verify abnormalities in patient data through data stream response and issue warnings; Establish a terminology database of international standard terms and regional common terms, match the patient data obtained from multiple sources to generate standard terms, then perform multi-source fusion, and perform data correction based on the fusion results; Based on the full-process data, a multi-source data quality evaluation table including completeness rate, standardization rate, conflict resolution rate and clinical recognition rate is generated for different departments, and the terminology library and fusion correction weight are updated based on the feedback of the quality evaluation table.

2. A medical data multi-source fusion verification and correction method according to claim 1, characterized in that: During the standardization process, the raw data is given a unified label encoding through the pre-loaded lightweight medical knowledge graph.

3. A medical data multi-source fusion verification and correction method according to claim 1, characterized in that: The construction of the dynamic baseline model for the emergency department, ICU and general ward includes setting a baseline that conforms to the data source for each type of patient data, which is used as a standard range for abnormal alarms; The establishment of the association relationship between patient data and model prediction data includes: inputting the timestamp, department, personal information and data type of the patient data, and predicting the reasonable fluctuation range of the corresponding data through the LSTM neural network.

4. A medical data multi-source fusion verification and correction method according to claim 3, characterized in that: The abnormal situation of verifying the patient data by data stream response includes: After dividing the data sources by department, the data flow platform is used to visualize the input patient data information; The data flow platform is connected to the dynamic baselines of different departments to monitor whether there are indicators in the patient data exceeding the preset threshold, and the medical knowledge base API is called to check the data and logic consistency in real time to generate an early warning strategy.

5. A medical data multi-source fusion verification and correction method according to claim 1, characterized in that: The abnormal situation of checking patient data through data stream response and issuing an early warning includes the following early warning strategies: The first warning strategy is set to provide sound and light prompts for data anomalies that directly endanger the safety of diagnosis and treatment. The second warning strategy is set to generate to-do tasks for data anomalies that affect data quality but are not urgent and send them to the quality control port. The third warning strategy is set to mark semantic risks and generate logs.

6. A medical data multi-source fusion verification and correction method according to claim 2, characterized in that: The assigning of trust weights is used to add a source authority score to each piece of data, and the term matching of patient data obtained from multiple sources to generate standard terms for unified term expression includes: Extract text fields from patient data and use pre-trained models for matching and identification; the term matching includes exact matching and fuzzy matching; the exact matching is used to directly extract text information matching the term library from the data; the fuzzy matching maps erroneous and / or abbreviated regional common terms to international standard terms; wherein, if it is detected that any regional common term has accumulated a preset number of occurrences and has not been mapped, the term is dynamically adjusted.

7. A medical data multi-source fusion verification and correction method according to claim 6, characterized in that: Before outputting the matching recognition result, it also includes semantic conflict recognition and dynamic priority adjustment; the semantic conflict recognition is used to record the conflict of text data; The dynamic priority adjustment is performed by the following method: Set weight dimensions, including source authority, time decay, and operator qualifications; Set conflict resolution rules, including: if the difference between the total weight score of any data source and other sources is greater than the preset value, it will be directly adopted; If the difference between the total weight score of any data source and other sources is less than the preset value, a manual review will be conducted and the conflict type and cause will be recorded.

8. A medical data multi-source fusion verification and correction method according to claim 1, characterized in that: The amendments include: Fill missing data based on similar records, convert non-compliant data to standard format, identify patients by unique identifiers and delete duplicate records, and then set up a hierarchical correction strategy for low-risk conflicts, medium-risk conflicts, and high-risk conflicts; The low-risk conflict correction is used to correct conflicts between different systems that have different terminology expressions but consistent semantics; The medium-risk conflict correction is used to correct conflicts that are numerically contradictory but logically inferable; its correction weight is determined based on the weight dimensions of source authority, time decay and operator qualifications; The high-risk conflict correction is used to correct the contradictory data that directly affects the safety of diagnosis or treatment, immediately issue an early warning and freeze the internal process of the current doctor's order; After completing any level of correction, the original conflict data snapshot and the basis for the correction decision are recorded in the database.

9. A medical data multi-source fusion verification and correction method according to claim 1, characterized in that: The completeness rate is used to calculate the proportion of missing data, which is used to indicate the degree of data completeness; the standardization rate is used to calculate the proportion of actual terms that conform to international standard terms; the conflict resolution rate is used to indicate the proportion of automatic conflict resolution without human involvement in the total conflicts; and the clinical recognition rate is used to sample the proportion of doctors adopting the revised results.

10. A medical data multi-source fusion verification and correction system, characterized in that: include: The data preprocessing module is used to obtain multi-source data, standardize preprocessing, and assign trust weights to each data based on different sources; The data monitoring module is used to build dynamic baseline models for the emergency department, ICU, and general wards, establish the relationship between patient data and model prediction data, and verify abnormalities in patient data through data stream response; The knowledge base and fusion correction module is used to establish a terminology library of international standard terms and regional common terms, match the patient data obtained from multiple sources to generate standard terms, then perform multi-source fusion, and perform data correction based on the fusion results; The feedback optimization module is used to generate a multi-source data quality evaluation table for different departments based on the full-process data, including the completeness rate, standardization rate, conflict resolution rate and clinical approval rate, and update the terminology library and fusion correction system based on the feedback of the quality evaluation table.

Images