Physical examination data exception association analysis method and system based on knowledge graph

Abstract

The invention provides a physical examination data exception association analysis method and system based on a knowledge graph, and relates to the technical field of exception association analysis, and the method comprises the steps: determining all physical examination items of a target physical examination center; aiming at a plurality of selectable physical examination indexes, executing abnormal generality association and generality confidence degree analysis among the indexes, and establishing an abnormal generality association map; receiving a physical examination report; executing abnormal generality association verification based on the abnormal generality association graph; performing abnormal distribution characteristic analysis in a preset period for the plurality of selectable physical examination indexes, and performing distribution abnormality verification on the physical examination report; and performing joint abnormity verification according to the distribution abnormity verification result and the generality verification result, then performing abnormal physical examination item positioning, and sending the abnormal physical examination item positioning to a target physical examination center for reminding. The technical problem that potential health risks are difficult to recognize in time and the accuracy of physical examination results is affected in the prior art can be solved, and the technical effect of improving the accuracy of physical examination anomaly detection is achieved.

Description

Technical Field

[0001] This application relates to the field of anomaly correlation analysis technology, and in particular to a method and system for anomaly correlation analysis of physical examination data based on knowledge graphs. Background Technology

[0002] As public health awareness continues to increase, the importance of physical examinations in early disease screening, chronic disease management, and health intervention is becoming increasingly prominent. Most physical examination centers provide hundreds or even thousands of tests to assess an individual's health status and determine the presence of health abnormalities based on these indicators. However, current technologies, when processing massive amounts of physical examination data, often rely solely on single indicators for anomaly detection, neglecting the potential correlations between indicators, particularly their weak ability to identify common patterns of abnormalities. For example, when multiple indicators are mildly abnormal simultaneously, although a single indicator may not meet the diagnostic criteria, the combined abnormality may indicate a deeper health risk, thus leading to the risk of missed or incorrect diagnoses.

[0003] In summary, existing technologies suffer from a lack of analytical capabilities to identify common abnormal correlations among health examination indicators. This makes it difficult to promptly identify potential health risks when faced with a combination of mildly abnormal indicators, further impacting the accuracy, comprehensiveness, and efficiency of early intervention in health examination results. Summary of the Invention

[0004] The purpose of this application is to provide a knowledge graph-based method and system for abnormal correlation analysis of physical examination data, in order to solve the technical problem in the prior art that the lack of analytical ability to analyze abnormal common correlations between physical examination indicators makes it difficult to identify potential health risks in a timely manner when faced with a combination of mild abnormalities in multiple indicators, which further affects the accuracy, comprehensiveness and early intervention efficiency of physical examination results.

[0005] In view of the above problems, this application provides a method and system for abnormal association analysis of physical examination data based on knowledge graph.

[0006] Firstly, this application provides a knowledge graph-based method for analyzing abnormal correlations in physical examination data, implemented through a knowledge graph-based system for analyzing abnormal correlations in physical examination data. The method includes: identifying all physical examination items at a target physical examination center and generating several optional physical examination indicators; for the several optional physical examination indicators, performing anomaly common correlation and common confidence level analysis based on historical abnormal data to establish an anomaly common correlation graph; receiving M physical examination reports generated by the target physical examination center in a first time zone; reading abnormal markers from the M physical examination reports, performing anomaly common correlation verification based on the anomaly common correlation graph, and generating common verification results; performing anomaly distribution feature analysis for the several optional physical examination indicators under a preset period, and verifying the distribution anomalies in the M physical examination reports using the preset anomaly distribution features; performing joint anomaly verification using the distribution anomaly verification results and the common verification results, locating abnormal physical examination items, and sending a notification to the target physical examination center.

[0007] Preferably, the knowledge graph-based abnormal correlation analysis method for physical examination data further includes: extracting a first optional physical examination indicator from the plurality of optional physical examination indicators; determining a first indicator abnormality space for the first optional physical examination indicator based on the historical abnormal data; identifying other indicators that have abnormal commonalities with the first optional physical examination indicator within the first indicator abnormality space based on the historical abnormal data, and calculating the commonality confidence level to generate a first abnormal commonality correlation graph; and adding the first abnormal commonality correlation graph to the abnormal commonality correlation graph.

[0008] Preferably, the knowledge graph-based physical examination data anomaly correlation analysis method further includes: the first indicator anomaly space includes a high abnormal range and a low abnormal range.

[0009] Preferably, the knowledge graph-based abnormal correlation analysis method for physical examination data further includes: segmenting the abnormal space of the first indicator according to a preset step size to generate multiple abnormal sub-regions of the first indicator; mapping the historical abnormal data to the multiple abnormal sub-regions of the first indicator according to the indicator value of the first optional physical examination indicator to generate multiple sub-region mapping data; identifying other indicators that have abnormal common correlations with the first optional physical examination indicator based on the multiple sub-region mapping data to generate multiple sub-region abnormal common correlation indicators; calculating the common abnormality ratio of the multiple sub-region abnormal common correlation indicators, generating a common confidence level, extracting sub-region abnormal common correlation indicators with a common confidence level greater than a preset coefficient, and generating the first abnormal common correlation graph.

[0010] Preferably, the knowledge graph-based abnormal correlation analysis method for physical examination data further includes: performing similarity analysis on the common abnormal correlation indicators and corresponding common confidence levels of the multiple sub-regions; if the similarity of the common abnormal correlation indicators and corresponding common confidence levels of any two or more sub-regions is greater than a preset similarity, sub-region merging is performed, and the first abnormal common correlation graph is optimized by dimensionality reduction.

[0011] Preferably, the knowledge graph-based abnormal association analysis method for physical examination data further includes: extracting the first abnormal marker, the second abnormal marker, and up to the Nth abnormal marker from the first physical examination report in the M physical examination reports; inputting the first abnormal marker, the second abnormal marker, and up to the Nth abnormal marker into the abnormal common association graph; performing matching verification in the abnormal common association graph according to the feature values ​​of the markers; determining whether the abnormal common association between the first abnormal marker, the second abnormal marker, and up to the Nth abnormal marker matches the abnormal common association graph; and generating the common verification result.

[0012] Preferably, the knowledge graph-based abnormal correlation analysis method for physical examination data further includes: analyzing the abnormal detection fluctuation range of the several optional physical examination indicators corresponding to different degrees of abnormal values ​​under the preset period; calculating the real-time abnormal detection rate of the several optional physical examination indicators for different degrees of abnormal values ​​under the preset period based on the M physical examination reports; determining whether the deviation between the real-time abnormal detection rate and the abnormal detection fluctuation range is greater than a preset deviation threshold, and if so, the distribution anomaly verification result is that an anomaly exists.

[0013] Preferably, the knowledge graph-based physical examination data anomaly association analysis method further includes: when both the distribution anomaly verification and the commonality verification results show anomalies, locating the commonality association anomaly index as an anomaly index based on the commonality verification results; determining the detection instrument and detection item corresponding to the anomaly index, and generating the abnormal physical examination item.

[0014] Preferably, the knowledge graph-based abnormal correlation analysis method for physical examination data further includes: when the target physical examination center issues an abnormal detection response for an abnormal physical examination item, the M physical examination users who read the M physical examination reports perform physical examination abnormality traceability management.

[0015] Secondly, this application also provides a knowledge graph-based system for analyzing abnormal correlations in physical examination data, used to execute the knowledge graph-based method for analyzing abnormal correlations in physical examination data as described in the first aspect, including: an optional physical examination indicator generation module, used to determine all physical examination items of the target physical examination center and generate several optional physical examination indicators; an abnormal common correlation graph establishment module, used to perform abnormal common correlation and common confidence level analysis between the indicators based on historical abnormal data for the several optional physical examination indicators, and establish an abnormal common correlation graph; and an M physical examination report generation module, used to receive reports from the target physical examination center in the first time... The system generates M physical examination reports. A commonality verification result generation module reads abnormal markers from the M physical examination reports, performs commonality association verification based on the commonality association map, and generates commonality verification results. A distribution anomaly verification module performs anomaly distribution feature analysis on several selectable physical examination indicators under a preset period, and performs distribution anomaly verification on the M physical examination reports using preset anomaly distribution features. An anomaly physical examination item location module performs joint anomaly verification using the distribution anomaly verification results and the commonality verification results, locates the anomaly physical examination item, and sends it to the target physical examination center for notification.

[0016] The technical solution provided in this application has at least the following technical effects or advantages: by realizing the technical goal of constructing an abnormal indicator association network based on knowledge graph, automatically identifying common abnormal patterns in physical examination data and performing accurate verification and tracing, the technical effects of improving the accuracy of abnormal detection in physical examinations, strengthening the linkage identification capability of complex health problems, and supporting physical examination centers to efficiently carry out personalized health interventions are achieved.

[0017] The above description is merely an overview of the technical solution of this application. To better understand the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0019] Figure 1This is a flowchart illustrating the knowledge graph-based method for abnormal correlation analysis of physical examination data in this application.

[0020] Figure 2 This is a schematic diagram of the structure of the knowledge graph-based physical examination data anomaly correlation analysis system of this application.

[0021] Figure labeling: Optional physical examination indicator generation module 11, abnormal common correlation map establishment module 12, M physical examination report generation module 13, common verification result generation module 14, distribution abnormality verification module 15, abnormal physical examination item location module 16. Detailed Implementation

[0022] This application provides a knowledge graph-based method and system for analyzing abnormal correlations in physical examination data. This addresses the technical problem in existing technologies where the lack of analytical capabilities for common abnormal correlations between physical examination indicators makes it difficult to promptly identify potential health risks when faced with combinations of mildly abnormal indicators, further impacting the accuracy, comprehensiveness, and efficiency of early intervention in physical examination results. The application aims to construct an abnormal indicator correlation network based on a knowledge graph, automatically identify common abnormal patterns in physical examination data, and perform precise verification and tracing. This results in improved accuracy in detecting abnormalities in physical examinations, enhanced ability to identify complex health problems, and support for physical examination centers to efficiently conduct personalized health interventions.

[0023] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. It should be understood that this application is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application. It should also be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, not all of them.

[0024] Example 1, please refer to the appendix. Figure 1 This application provides a knowledge graph-based method for abnormal correlation analysis of physical examination data, which is applied to a knowledge graph-based physical examination data abnormal correlation analysis system, and specifically includes the following steps: S1: Determine all medical examination items of the target medical examination center and generate several optional medical examination indicators.

[0025] Specifically, the target health checkup center is the one from which the health checkup data correlation analysis will be conducted. First, all health checkup items currently offered by the target center are identified, i.e., the various tests used to examine health, such as complete blood count, liver function tests, and electrocardiograms. Next, several optional health checkup indicators are selected from all these items as research subjects, such as blood glucose concentration, blood pressure, and triglyceride levels. For example, if the target center offers fifty checkup services, then the indicators corresponding to all offered checkup items can be used to construct several optional indicators. Twenty of these checkup items are key numerical values ​​related to metabolism and cardiovascular function; these twenty items can then be defined as optional health checkup indicators. In subsequent anomaly analysis, these indicators can be focused on those more likely to cause problems or are more representative, thereby improving the efficiency and effectiveness of the analysis.

[0026] S2: For the several optional physical examination indicators, perform anomaly common correlation and common confidence level analysis between indicators based on historical abnormal data, and establish anomaly common correlation map.

[0027] Specifically, historical anomaly data utilizes a large number of real physical examination records from a past period that have been marked as abnormal. For several selectable physical examination indicators, the study identifies patterns where an abnormality in one indicator is frequently accompanied by abnormalities in one or more other indicators. For example, elevated blood sugar is often accompanied by elevated blood pressure, revealing common associations. A common confidence level analysis is then performed on these commonly associated physical examination indicators to analyze the frequency and stability of their simultaneous occurrence in the historical anomaly data, thus establishing an anomaly common association map.

[0028] S3: Receive M medical examination reports generated by the target medical examination center in the first time zone.

[0029] Specifically, the first time zone refers to a preset time range, such as a week, a month, or a quarter, used to limit the data to be analyzed to belong to the same time window, ensuring the timeliness consistency of statistical and trend analysis. The system receives M medical examination reports generated by the target medical examination center within the first time zone. Here, M refers to the specific number of reports, which can be 10, 100, or 1000, measured in reports. Each medical examination report represents the set of test results for all medical examination items for an individual in a given medical examination, including normal and abnormal markers, testing time, report number, and other information.

[0030] S4: Read the abnormal markers in the M physical examination reports, perform abnormal common association verification based on the abnormal common association map, and generate common verification results.

[0031] Specifically, the process involves reading abnormal markers from M medical examination reports from the target medical examination center, extracting the abnormal indicators (such as high blood sugar, abnormal liver function, or abnormal electrocardiogram) from each report, and marking them in each user's medical examination report. Subsequently, anomaly commonality association verification is performed based on an anomaly commonality association map. This involves analyzing the patterns of frequently occurring abnormalities based on the extracted abnormal markers within the map, comparing whether the abnormal markers conform to existing common patterns, and thus determining whether there is a reasonable commonality association between the abnormal markers. Finally, a commonality verification result is generated, indicating whether the abnormal markers belong to common combinations.

[0032] S5: Perform abnormal distribution feature analysis on the several optional physical examination indicators under a preset period, and verify the distribution anomalies of the M physical examination reports using the preset abnormal distribution features.

[0033] Specifically, for several selectable physical examination indicators, an abnormal distribution characteristic analysis is performed under a preset period. The frequency and distribution pattern of abnormalities within the preset period are statistically analyzed according to a fixed time period. The preset abnormal distribution characteristics are used to verify the distribution abnormality of M physical examination reports to determine whether there are any phenomena that significantly deviate from the normal abnormal trend in the M physical examination reports, and the distribution abnormality verification results are obtained.

[0034] S6: After performing joint anomaly verification using the distribution anomaly verification results and the commonality verification results, perform abnormal physical examination item location and send it to the target physical examination center for reminder.

[0035] Specifically, joint anomaly verification is performed using the results of distribution anomaly verification and commonality verification. This involves integrating and judging the data, tracing back to identify the specific indicators that caused the anomalies, such as identifying key medical examination indicators that repeatedly appear in all anomaly reports. This information is then combined with the correlation structure in the anomaly commonality association map to further pinpoint potential systemic anomalies. Warning messages are generated and sent to the target medical examination center to alert them, enabling the center to promptly understand potential anomaly trends in medical examination items and thus perform equipment calibration or data review.

[0036] Furthermore, this application also includes: extracting a first optional physical examination indicator from the plurality of optional physical examination indicators; determining a first indicator anomaly space for the first optional physical examination indicator based on the historical anomaly data; identifying other indicators that have anomaly common associations with the first optional physical examination indicator within the first indicator anomaly space based on the historical anomaly data, and calculating the common confidence level to generate a first anomaly common association map; and adding the first anomaly common association map to the anomaly common association map.

[0037] Specifically, the first optional physical examination indicator is randomly selected from several optional physical examination indicators, such as blood pressure, blood sugar, and white blood cell count.

[0038] Next, historical abnormal data refers to the values ​​of medical examination indicators that deviate from the normal range in the target medical examination center's past examination records. The abnormal space refers to the abnormal value range of the medical examination indicator corresponding to the abnormal data. For example, a normal blood glucose level is 4 to 6 mmol / L; therefore, the portion below 4 or above 6 is considered the abnormal space. Based on historical abnormal data, the first abnormal space of the first optional medical examination indicator is determined, that is, the abnormal space is further refined into the first abnormal space corresponding to the first optional medical examination indicator, such as above 7 mmol / L and below 3 mmol / L.

[0039] Subsequently, within the anomaly space of the first indicator, other indicators with abnormal commonalities that are associated with the first optional physical examination indicator are identified based on historical anomaly data. Anomaly commonalities refer to the pattern where an abnormality in one indicator is accompanied by abnormalities in one or more other indicators; for example, when blood sugar is high, triglycerides and body mass index are also abnormal. Then, the confidence level of commonalities is calculated to generate the first anomaly commonality association map. The confidence level of commonalities is a numerical value that measures the stability and reliability of commonalities.

[0040] Finally, the first abnormal common association graph is added to the entire abnormal common association graph. That is, the corresponding abnormal association network established around the random physical examination indicators is included in the abnormal common association graph. The abnormal common association graph is obtained by using the same method as the first abnormal common association graph obtained according to the first optional physical examination indicator, and then obtaining the second abnormal common association graph in sequence, until several optional physical examination indicators are traversed, and then all abnormal common association graphs are combined to obtain the graph.

[0041] Furthermore, this application also includes: the first indicator anomaly space includes an abnormally high range and an abnormally low range.

[0042] Specifically, the first abnormality space refers to all numerical ranges of a specific physical examination indicator that exceed its normal reference range. This first abnormality space further includes elevated and lower abnormality ranges. The elevated range is the numerical range where the indicator value exceeds its upper normal limit. For example, if the upper normal limit for blood glucose is 6 mmol / L, then all values ​​above 6, such as 6.1, 7, and 10 mmol / L, fall within the elevated range. The lower abnormality range is the numerical range where the physical examination indicator falls below its lower normal limit. For example, if the lower normal limit for blood glucose is 4 mmol / L, then values ​​below 4, such as 3.8, 3.5, or lower, fall within the lower abnormality range. The threshold values ​​for the elevated and lower abnormality ranges can be determined not only based on normal physiological reference values ​​but also by the limitations of the instrument's detection range. For instance, if the equipment can only detect values ​​from 0 to 20 mmol / L, then 20 is the maximum boundary of the elevated range; values ​​exceeding this value will no longer yield accurate results.

[0043] Furthermore, this application also includes: segmenting the first indicator anomaly space according to a preset step size to generate multiple first indicator anomaly sub-regions; mapping the historical anomaly data to the multiple first indicator anomaly sub-regions according to the indicator values ​​of the first optional physical examination indicator to generate multiple sub-region mapping data; identifying other indicators that have anomaly common associations with the first optional physical examination indicator based on the multiple sub-region mapping data to generate multiple sub-region anomaly common association indicators; calculating the common anomaly ratio of the multiple sub-region anomaly common association indicators to generate a common confidence level; extracting sub-region anomaly common association indicators with a common confidence level greater than a preset coefficient to generate the first anomaly common association map.

[0044] Specifically, the preset step size is a numerical interval set by those skilled in the art based on actual conditions. For example, the abnormal blood glucose range from 6 mmol / L to 12 mmol / L is divided into six sub-regions, such as 6 to 7, 7 to 8, and 8 to 9, with a step size of 1 mmol / L. The abnormal space of the first indicator is segmented according to the preset step size, generating multiple abnormal sub-regions of the first indicator, thus refining the broad abnormal range into several smaller continuous intervals. Next, in each historical abnormal data point, the specific abnormal value of the corresponding physical examination indicator is extracted, and the abnormal sub-region of the first indicator to which the physical examination indicator belongs among the first selectable physical examination indicators is determined.

[0045] Then, based on the mapping data of multiple sub-regions, other indicators that have abnormal common associations with the first optional physical examination indicator are identified. Other physical examination indicators that frequently show abnormalities simultaneously in the sub-region mapping data are found, and their names and associations are recorded, generating multiple sub-region abnormal common association indicators. For example, in the sub-region mapping data of blood glucose 8 to 9, cholesterol and blood pressure may also be found to be high simultaneously, and these are recorded as abnormal common association indicators of the first optional physical examination indicator.

[0046] Next, the common anomaly ratio is calculated for the common correlation indicators of multiple sub-regions, that is, the proportion of common anomalies in each first-index anomaly sub-region out of the total data, generating a common confidence level. The preset coefficient is a threshold set in advance by those skilled in the art based on the actual situation. The common correlation indicators of sub-regions with a common confidence level greater than the preset coefficient are extracted and considered to be correlated, generating a first anomaly common correlation map, which is used to screen out reliable anomaly relationships.

[0047] Furthermore, this application also includes: performing similarity analysis on the common anomaly correlation indicators and corresponding common confidence levels of the multiple sub-regions; if the similarity of the common anomaly correlation indicators and corresponding common confidence levels of any two or more sub-regions is greater than a preset similarity, performing sub-region merging and dimensionality reduction optimization on the first common anomaly correlation map.

[0048] Specifically, the preset similarity is a similarity value customized by those skilled in the art based on actual circumstances. Similarity analysis is performed on common anomaly correlation indicators and their corresponding common confidence levels across multiple sub-regions to determine the structural and numerical similarity of these indicators. If the similarity of any two or more common anomaly correlation indicators and their corresponding common confidence levels exceeds the preset similarity, sub-region merging is performed on the sub-regions corresponding to these indicators. The common anomaly correlation indicators of highly similar sub-regions are then fused. The merged sub-regions represent a broader but consistent range of anomalies, avoiding data redundancy and computational complexity. The first anomaly common correlation graph is then dimensionality-reduced and simplified by reducing the number of nodes or edges and merging redundant structures, thereby improving processing efficiency and subsequent analysis performance.

[0049] Furthermore, this application also includes: extracting the first abnormal marker, the second abnormal marker, and up to the Nth abnormal marker from the first medical examination report in the M medical examination reports; inputting the first abnormal marker, the second abnormal marker, and up to the Nth abnormal marker into the abnormal common association map; performing matching verification in the abnormal common association map according to the feature values ​​of the markers; determining whether the abnormal common association between the first abnormal marker, the second abnormal marker, and up to the Nth abnormal marker matches the abnormal common association map; and generating the common verification result.

[0050] Specifically, indicators that are too high or too low in the physical examination report will be clearly marked and can be read directly. The first physical examination report from M random physical examination reports is extracted, and the first, second, up to the Nth abnormality markers in the first report are extracted. All physical examination indicators marked as abnormal in the first report are identified and extracted.

[0051] The first, second, and up to the Nth anomaly marker are input into the anomaly commonality association map. Matching and verification are performed in the anomaly commonality association map according to the feature values ​​of the anomaly markers. That is, based on the specific characteristics of the anomaly indicators, such as numerical range and anomaly type (high or low), the corresponding association path is searched in the anomaly commonality association map to verify whether the anomaly markers appear according to the common relationships in historical data. It is determined whether the anomaly commonality associations between the first, second, and up to the Nth anomaly markers match the anomaly commonality association map. The association patterns of the anomaly indicators in the current report are compared with the association patterns in the historical maps to determine whether they are consistent or highly similar, thereby evaluating the anomaly commonality verification results. The commonality verification results reflect whether the anomalies in the medical examination report conform to previously discovered anomaly combination patterns. If the matching degree is high, it indicates that the anomalies may have a potential association mechanism.

[0052] Furthermore, this application also includes: analyzing the abnormal detection fluctuation range of the several optional physical examination indicators corresponding to different degrees of abnormality under the preset period; calculating the real-time abnormal detection rate of the several optional physical examination indicators to different degrees of abnormality under the preset period based on the M physical examination reports; determining whether the deviation between the real-time abnormal detection rate and the abnormal detection fluctuation range is greater than a preset deviation threshold, and if so, the distribution anomaly verification result is that an anomaly exists.

[0053] Specifically, the preset period is a period that can be customized by those skilled in the art based on actual conditions. The preset period can be one day. If the fluctuation range of the abnormal detection rate within one day is too high, further attention to the abnormalities is required. The abnormal detection fluctuation range of several optional physical examination indicators corresponding to different degrees of abnormality is analyzed under the preset period. The range and fluctuation of abnormal values ​​of optional physical examination indicators are statistically analyzed. Different degrees of abnormality refer to the possibility that the indicator may have mild, moderate or severe abnormalities. For example, blood sugar exceeding the normal value by one time is considered mild abnormality, and exceeding it by two times is considered moderate abnormality. The abnormal detection fluctuation range is thus obtained.

[0054] Based on M physical examination reports, the real-time abnormality detection rate of several selectable physical examination indicators under a preset period is calculated. That is, by statistically analyzing all physical examination report data within the current time period, the actual detection ratio of each physical examination indicator at different abnormality levels is calculated.

[0055] The preset deviation threshold can be customized by those skilled in the art based on actual conditions. The system determines whether the deviation between the real-time anomaly detection rate and the anomaly detection fluctuation range exceeds the preset deviation threshold. It compares the current actual anomaly detection rate with the historical fluctuation range. If the difference exceeds the warning value, it indicates that the anomaly frequency significantly exceeds the normal fluctuation range, suggesting the possible existence of an abnormal event or trend. This leads to the distribution anomaly verification result indicating the presence of an anomaly. Table 1 shows a partial record of the most recent physical examination indicator distribution anomaly verification analysis.

[0056] Table 1: Partial Records of Verification Analysis of Abnormal Distribution of Indicators in the Most Recent Physical Examination

[0057] Furthermore, this application also includes: when both the distribution anomaly verification and the commonality verification results show anomalies, the commonality-related anomaly indicators are located based on the commonality verification results as anomaly indicators; the detection instruments and detection items corresponding to the anomaly indicators are determined, and the abnormal physical examination items are generated.

[0058] Specifically, when both the distribution anomaly verification and commonality verification results show anomalies, the indicators that are associated with common anomalies based on the commonality verification results are identified as anomalous indicators. The portions that match the abnormal items in the physical examination reports are extracted, and the indicators that share commonalities among the abnormal items are considered the most critical and likely to be anomalous. For example, if high blood sugar, high insulin, and high triglycerides are frequently and simultaneously abnormal in multiple physical examination reports, and the commonality confidence level is high, then high blood sugar can be prioritized as an anomalous indicator.

[0059] The process involves identifying the corresponding testing instruments and items for abnormal indicators, generating abnormal health check items, and further tracing down the key abnormal indicators to the instruments used to measure them in actual testing, thus clarifying the specific testing items and forming the final abnormal health check items. For example, if the abnormal indicator is blood glucose, the corresponding testing instrument might be a biochemical analyzer, and the corresponding testing item might be fasting blood glucose, thereby marking the conclusion of "abnormal fasting blood glucose".

[0060] Furthermore, this application also includes: when the target medical examination center issues an abnormal detection response for an abnormal medical examination item, the M medical examination users who read the M medical examination reports perform medical examination abnormality traceability management.

[0061] Specifically, when the target medical examination center issues an abnormal test result, it will be reviewed manually or by machine to confirm that there is indeed an abnormality and respond to the abnormal test result, which may include a recommendation for re-examination or marking it as requiring special attention.

[0062] The system manages the abnormal health checkup results of M individuals who have M health checkup reports. It retrieves and analyzes each of the previously generated M health checkup reports, identifies the M individuals to whom each report belongs, and traces the abnormal indicators of these individuals throughout the entire process. This involves finding the individuals' historical health checkup records, trends in relevant indicators, and potential influencing indicators to determine whether the abnormality is sudden, gradually evolving, or has been latent for a long time. This facilitates the timely sending of alerts for abnormal health checkup equipment and the arrangement of secondary health checkups.

[0063] In summary, the knowledge graph-based abnormal correlation analysis method for physical examination data provided in this application has the following technical effects: by realizing the technical goal of constructing an abnormal indicator correlation network based on knowledge graph, automatically identifying common abnormal patterns in physical examination data and performing accurate verification and tracing, it achieves the technical effects of improving the accuracy of abnormal detection in physical examinations, strengthening the linkage identification capability of complex health problems, and supporting physical examination centers to efficiently carry out personalized health interventions.

[0064] Example 2: Based on the same inventive concept as the knowledge graph-based abnormal correlation analysis method for physical examination data in the foregoing examples, this application also provides a knowledge graph-based abnormal correlation analysis system for physical examination data. Please refer to the appendix. Figure 2 The system includes: an optional physical examination indicator generation module 11, used to determine all physical examination items of the target physical examination center and generate several optional physical examination indicators; an abnormal common correlation map establishment module 12, used to perform abnormal common correlation and common confidence level analysis between the indicators based on historical abnormal data for the several optional physical examination indicators, and establish an abnormal common correlation map; an M physical examination report generation module 13, used to receive M physical examination reports generated by the target physical examination center in the first time zone; a common verification result generation module 14, used to read the abnormal marker items in the M physical examination reports, perform abnormal common correlation verification based on the abnormal common correlation map, and generate common verification results; a distribution anomaly verification module 15, used to perform abnormal distribution feature analysis under a preset period for the several optional physical examination indicators, and perform distribution anomaly verification on the M physical examination reports with preset abnormal distribution features; and an abnormal physical examination item location module 16, used to perform joint anomaly verification with the distribution anomaly verification results and the common verification results, and then perform abnormal physical examination item location and send it to the target physical examination center for reminder.

[0065] Furthermore, the knowledge graph-based physical examination data anomaly correlation analysis system is also used for: extracting a first optional physical examination indicator from the plurality of optional physical examination indicators; determining a first indicator anomaly space for the first optional physical examination indicator based on the historical anomaly data; identifying other indicators that have anomaly common correlations with the first optional physical examination indicator within the first indicator anomaly space based on the historical anomaly data, and calculating the common confidence level to generate a first anomaly common correlation graph; and adding the first anomaly common correlation graph to the anomaly common correlation graph.

[0066] Furthermore, the knowledge graph-based physical examination data anomaly correlation analysis system is also used to: the first indicator anomaly space includes an abnormal range of excessively high values ​​and an abnormal range of excessively low values.

[0067] Furthermore, the knowledge graph-based physical examination data anomaly correlation analysis system is also used for: segmenting the first indicator anomaly space according to a preset step size to generate multiple first indicator anomaly sub-regions; mapping the historical anomaly data to the multiple first indicator anomaly sub-regions according to the indicator values ​​of the first optional physical examination indicator to generate multiple sub-region mapping data; identifying other indicators that have anomaly common correlations with the first optional physical examination indicator based on the multiple sub-region mapping data to generate multiple sub-region anomaly common correlation indicators; calculating the common anomaly ratio of the multiple sub-region anomaly common correlation indicators, generating a common confidence level, extracting sub-region anomaly common correlation indicators with a common confidence level greater than a preset coefficient, and generating the first anomaly common correlation graph.

[0068] Furthermore, the knowledge graph-based physical examination data anomaly association analysis system is also used to: perform similarity analysis on the common anomaly association indicators and corresponding common confidence levels of the multiple sub-regions; if the similarity of the common anomaly association indicators and corresponding common confidence levels of any two or more sub-regions is greater than a preset similarity, perform sub-region merging and perform dimensionality reduction optimization on the first anomaly common association graph.

[0069] Furthermore, the knowledge graph-based physical examination data anomaly association analysis system is also used to: extract the first abnormal marker, the second abnormal marker, and up to the Nth abnormal marker from the first physical examination report in the M physical examination reports; input the first abnormal marker, the second abnormal marker, and up to the Nth abnormal marker into the anomaly common association graph; perform matching verification in the anomaly common association graph according to the feature values ​​of the markers; determine whether the anomaly common association between the first abnormal marker, the second abnormal marker, and up to the Nth abnormal marker matches the anomaly common association graph; and generate the common verification result.

[0070] Furthermore, the knowledge graph-based physical examination data anomaly correlation analysis system is also used to: analyze the abnormal detection fluctuation range of the several optional physical examination indicators corresponding to different degrees of abnormal values ​​under the preset period; calculate the real-time abnormal detection rate of the several optional physical examination indicators for different degrees of abnormal values ​​under the preset period based on the M physical examination reports; determine whether the deviation between the real-time abnormal detection rate and the abnormal detection fluctuation range is greater than a preset deviation threshold, and if so, the distribution anomaly verification result is that an anomaly exists.

[0071] Furthermore, the knowledge graph-based physical examination data anomaly correlation analysis system is also used for: when both the distribution anomaly verification and the commonality verification results show anomalies, locating the commonality correlation anomaly indicators as anomaly indicators based on the commonality verification results; determining the detection instruments and detection items corresponding to the anomaly indicators, and generating the abnormal physical examination items.

[0072] Furthermore, the knowledge graph-based physical examination data anomaly correlation analysis system is also used to: when the target physical examination center issues an anomaly detection response for an abnormal physical examination item, to perform physical examination anomaly traceability management on the M physical examination users who read the M physical examination reports.

[0073] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The knowledge graph-based abnormal correlation analysis method and specific examples in the aforementioned embodiment one are also applicable to the knowledge graph-based abnormal correlation analysis system for physical examination data in this embodiment. Through the foregoing detailed description of the knowledge graph-based abnormal correlation analysis method for physical examination data, those skilled in the art can clearly understand the knowledge graph-based abnormal correlation analysis system for physical examination data in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here.

[0074] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0075] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for abnormal association analysis of physical examination data based on knowledge graphs, characterized in that, include: Identify all medical examination items at the target medical examination center and generate several optional medical examination indicators; For the aforementioned optional physical examination indicators, an abnormal common correlation and common confidence level analysis are performed on the indicators based on historical abnormal data to establish an abnormal common correlation map; Receive M medical examination reports generated by the target medical examination center in the first time zone; Read the abnormal markers from the M physical examination reports, perform abnormal common association verification based on the abnormal common association map, and generate common verification results; An abnormal distribution feature analysis is performed on the several optional physical examination indicators under a preset period, and the M physical examination reports are verified for distribution anomalies using the preset abnormal distribution features. After performing joint anomaly verification using the distribution anomaly verification results and the commonality verification results, the abnormal physical examination item is located and sent to the target physical examination center for reminder.

2. The method for abnormal correlation analysis of physical examination data based on knowledge graph as described in claim 1, characterized in that, For the aforementioned optional physical examination indicators, anomaly commonalities and common confidence levels among the indicators are analyzed based on historical abnormal data to establish an anomaly commonality correlation map, including: Extract the first optional physical examination indicator from the plurality of optional physical examination indicators; Based on the historical abnormal data, determine the first indicator abnormality space of the first optional physical examination indicator. Within the first indicator anomaly space, other indicators that have anomaly common associations with the first optional physical examination indicators are identified based on the historical anomaly data, and the common confidence level is calculated to generate the first anomaly common association map. Add the first common anomaly association map to the common anomaly association map.

3. The method for abnormal correlation analysis of physical examination data based on knowledge graph as described in claim 2, characterized in that, The first indicator's abnormal space includes both excessively high and excessively low abnormal ranges.

4. The method for abnormal correlation analysis of physical examination data based on knowledge graph as described in claim 3, characterized in that, Within the first indicator anomaly space, based on the historical anomaly data, other indicators that share anomaly commonalities with the first optional physical examination indicator are identified, and the confidence level of commonality is calculated to generate a first anomaly commonality association map, including: The first indicator anomaly space is divided according to a preset step size to generate multiple first indicator anomaly sub-regions; Based on the index value of the first optional physical examination index, the historical abnormal data is mapped to the multiple first index abnormal sub-regions to generate multiple sub-region mapping data; Based on the mapping data of the multiple sub-regions, other indicators that have abnormal common associations with the first optional physical examination indicators are identified respectively, and multiple sub-region abnormal common association indicators are generated; The common anomaly ratio is calculated for the common anomaly correlation indicators of the multiple sub-regions, a common confidence level is generated, and the common anomaly correlation indicators of the sub-regions with a common confidence level greater than a preset coefficient are extracted to generate the first anomaly common correlation map.

5. The method for abnormal correlation analysis of physical examination data based on knowledge graph as described in claim 4, characterized in that, A similarity analysis is performed on the common anomaly correlation indicators and corresponding common confidence levels of the multiple sub-regions. If the similarity of the common anomaly correlation indicators and corresponding common confidence levels of any two or more sub-regions is greater than a preset similarity, sub-region merging is performed, and the first common anomaly correlation map is optimized for dimensionality reduction.

6. The method for abnormal correlation analysis of physical examination data based on knowledge graph as described in claim 1, characterized in that, Read the abnormal markers from the M medical examination reports, perform abnormal common association verification based on the abnormal common association map, and generate common verification results, including: Extract the first abnormality marker, the second abnormality marker, up to the Nth abnormality marker from the first of the M physical examination reports; The first anomaly marker, the second anomaly marker, and so on up to the Nth anomaly marker are input into the anomaly commonality association graph. Matching and verification are performed in the anomaly commonality association graph according to the feature values ​​of the markers. It is determined whether the anomaly commonality association between the first anomaly marker, the second anomaly marker, and so on up to the Nth anomaly marker matches the anomaly commonality association graph, and the commonality verification result is generated.

7. The method for abnormal correlation analysis of physical examination data based on knowledge graph as described in claim 1, characterized in that, An abnormal distribution feature analysis is performed on the aforementioned selectable physical examination indicators under a preset period. The M physical examination reports are then verified for distribution anomalies using the preset abnormal distribution features, including: The analysis focuses on the abnormal detection fluctuation range of the several selectable physical examination indicators corresponding to different degrees of abnormality under the preset period; Based on the M physical examination reports, calculate the real-time abnormality detection rate of the several optional physical examination indicators with different degrees of abnormality under the preset period; Determine whether the deviation between the real-time anomaly detection rate and the anomaly detection fluctuation range is greater than a preset deviation threshold. If so, the distribution anomaly verification result indicates that an anomaly exists.

8. The method for abnormal correlation analysis of physical examination data based on knowledge graph as described in claim 1, characterized in that, After performing joint anomaly verification based on the distribution anomaly verification and the commonality verification results, the abnormal physical examination item is located and sent to the target physical examination center for notification, including: When both the distribution anomaly verification and the commonality verification results show anomalies, the commonality-related anomaly indicators are located based on the commonality verification results as anomaly indicators. The detection instruments and detection items corresponding to the abnormal indicators are determined, and the abnormal physical examination items are generated.

9. The method for abnormal correlation analysis of physical examination data based on knowledge graph as described in claim 8, characterized in that, After sending a reminder to the target medical examination center, the process also includes: When the target medical examination center issues an abnormal detection response for an abnormal medical examination item, the M medical examination users who read the M medical examination reports will be subject to medical examination abnormality traceability management.

10. A knowledge graph-based system for analyzing abnormal correlations in physical examination data, characterized in that: The steps for implementing the knowledge graph-based abnormal correlation analysis method for physical examination data according to any one of claims 1 to 9 include: The optional physical examination indicator generation module is used to determine all physical examination items of the target physical examination center and generate several optional physical examination indicators. The abnormal common correlation map establishment module is used to perform abnormal common correlation and common confidence level analysis between the indicators based on historical abnormal data for the several optional physical examination indicators, and establish an abnormal common correlation map. The module for generating M medical examination reports is used to receive M medical examination reports generated by the target medical examination center in the first time zone. The commonality verification result generation module is used to read the abnormality markers in the M physical examination reports, perform abnormality commonality association verification based on the abnormality commonality association map, and generate commonality verification results; The distribution anomaly verification module is used to perform anomaly distribution feature analysis on the several selectable physical examination indicators under a preset period, and to verify the distribution anomalies of the M physical examination reports using the preset anomaly distribution features. The abnormal physical examination item location module is used to perform joint abnormal verification based on the distributed abnormality verification results and the commonality verification results, and then locate the abnormal physical examination item and send it to the target physical examination center for reminder.

Images