A liver fibrosis grading determination system based on serum index data
The grading system based on serum index data solves the problems of individual differences and blurred boundaries between grades in liver fibrosis grading, and achieves continuous improvement in the accuracy and quantitative identification and diagnosis of liver fibrosis. It is suitable for non-invasive large-scale population screening and dynamic follow-up.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- JIANGXI INST OF PARASITIC DISEASE CONTROL
- Filing Date
- 2026-03-23
- Publication Date
- 2026-07-03
Smart Images
Figure CN122337583A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of liver fibrosis grading, and particularly to a liver fibrosis grading system based on serum index data. Background Technology
[0002] Liver fibrosis is a key pathological stage in the progression of various chronic liver diseases to cirrhosis and even liver cancer. Early and accurate diagnosis and grading assessment are the core prerequisites for clinical intervention.
[0003] Currently, the clinical gold standard diagnostic method mainly relies on liver biopsy. However, this technique is invasive and has limitations such as sampling error, bleeding risk, and low patient compliance, making it difficult to be widely used for early screening and dynamic monitoring.
[0004] In the field of non-invasive diagnostics, existing technologies mostly focus on the detection of serum biochemical indicators at a single time point (such as single markers like hyaluronic acid and laminin), or simply calculate using a combination of multiple indicators, which has significant technical shortcomings: First, they ignore the temporal dynamic changes in metabolic responses under pathological liver conditions. Using only static baseline data cannot effectively eliminate interfering factors such as individual physiological differences, diet, lifestyle, and short-term inflammatory fluctuations, resulting in highly volatile and insufficiently specific diagnostic results. Second, they lack standardized grading pattern libraries and precise pattern matching mechanisms, failing to develop specific and quantifiable feature clustering standards for liver fibrosis of different Metavir grades (F0-F4). Relying solely on empirical thresholds or linear regression models makes it difficult to characterize complex synergistic changes in markers, leading to problems such as blurred cross-grading boundaries and high misjudgment rates. Third, diagnostic systems are mostly one-way outputs, lacking a closed-loop linkage and dynamic optimization mechanism with gold standard data. They cannot continuously correct model parameters and classification boundaries using subsequent clinical validation data, making it difficult to iteratively improve diagnostic accuracy.
[0005] Therefore, it is necessary to propose a liver fibrosis grading system based on serum index data to solve the above problems. Summary of the Invention
[0006] The main objective of this invention is to provide a liver fibrosis grading system based on serum index data, which can effectively solve the problems in the background art.
[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A liver fibrosis grading system based on serum index data includes a standardized time-series acquisition module for liver fibrosis-associated serum physiological signals, a gold standard-paired liver fibrosis grading pattern library construction and pattern clustering module, a subject physiological signal pattern matching and grading determination module, and a diagnostic result verification and pattern library dynamic optimization module. The standardized time-series acquisition module for liver fibrosis-associated serum physiological signals is used to complete standardized time-series serum sample collection and multi-marker detection. After interference removal, baseline anchoring and normalization processing, individual difference interference is eliminated, and a liver fibrosis-associated time-series physiological signal feature set is output. The gold standard paired liver fibrosis grading pattern library construction and pattern clustering module is used to collect gold standard paired grading samples into the library to construct an initial pattern library. Through unsupervised pattern clustering, core cluster centers and pattern boundaries for each grade are generated, and grade-specific standard physiological signal pattern clusters are solidified. The subject physiological signal pattern matching and grading determination module is used to match the subject's temporal physiological signal feature set with each grading standard pattern cluster, calculate the corresponding matching degree, output the liver fibrosis grading diagnosis result after exclusion verification, and trigger standardized re-examination if the conditions are not met. The diagnostic result verification and pattern library dynamic optimization module is used to retrospectively collect valid diagnostic samples verified by the gold standard to supplement the library. When the preset sample size is reached, the hierarchical cluster centers and pattern boundaries are iteratively updated to form a self-optimizing closed loop that continuously improves diagnostic accuracy.
[0008] Preferably, the liver fibrosis-related serum physiological signal standardization time-series acquisition module includes a standardized physiological response triggering and serum sample time-series acquisition submodule and a serum physiological signal interference removal and baseline anchoring submodule, wherein the standardized physiological response triggering and serum sample time-series acquisition submodule specifically includes: It is used to trigger liver-specific metabolic responses through standardized physiological loads, thereby enabling the acquisition of diagnostic physiological signals.
[0009] Preferably, the serum sample time-series acquisition submodule specifically includes: The preprocessing of physiological signals for diagnosis is carried out as follows: the serum biomarker detection values of the subject 0 hours before the load are used as the individual physiological baseline. The relative changes of all biomarker physiological signals at the last two time points are normalized to remove signal interference caused by individual basic physiological differences of the subject. The dynamic physiological response characteristics of each biomarker and the synergistic change correlation characteristics between multiple biomarkers are extracted to finally form a subject-specific time-series physiological signal feature set associated with liver fibrosis.
[0010] Preferably, the gold standard paired liver fibrosis grading pattern library construction and pattern clustering module includes a gold standard paired physiological signal sample library entry submodule and a grading-specific physiological signal pattern clustering and boundary solidification submodule, wherein the gold standard paired physiological signal sample library entry submodule specifically includes: To construct a benchmark sample set for diagnosis, specifically: collect a standardized time-series serum physiological signal feature set of subjects diagnosed by liver biopsy gold standard, covering five stages of Metavir grade F0-F4, and add a fixed number of valid samples to the database for each grade. Each sample is simultaneously labeled with the corresponding biopsy grade gold standard result and etiology type information to form an initial pattern library with gold standard diagnostic labels.
[0011] Preferably, the pattern clustering module specifically includes: This method is used to construct a standardized diagnostic grading benchmark through pattern clustering technology. Specifically, for all physiological signal feature sets of the same grade in the initial pattern library, an unsupervised pattern clustering method is used to generate the core feature cluster center for each liver fibrosis grade. At the same time, the boundary range of physiological signal patterns for each grade is defined, outliers across grades are removed, and finally, a unique standard physiological signal pattern cluster corresponding to each liver fibrosis grade is formed.
[0012] Preferably, the subject physiological signal pattern matching and grading determination module includes a time-series physiological signal pattern matching degree calculation submodule and a grading result exclusive diagnostic output submodule, wherein the time-series physiological signal pattern matching degree calculation submodule specifically includes: Diagnostic feature matching is achieved through pattern classification technology. Specifically, the subject-specific time-series physiological signal feature set output by the preceding module is matched sequentially with five graded standard physiological signal pattern clusters in the pattern library. The matching degree between the subject feature set and the core cluster center of each grade is calculated. At the same time, it is verified whether the subject feature set falls within the pattern boundary range of the corresponding grade, and the matching degree value corresponding to each grade is output.
[0013] Preferably, the grading result exclusive diagnostic output submodule specifically includes: To achieve the final diagnostic result output, the following steps are taken: sort all the matching degree values corresponding to the grades from high to low, and output the final liver fibrosis grading diagnosis result only when the matching degree value of the grade corresponding to the highest matching degree exceeds the preset threshold of the second highest matching degree value, and the subject's feature set falls completely within the pattern boundary range of that grade; if the exclusion condition is not met, the standardized re-examination process of the subject is directly triggered.
[0014] Preferably, the diagnostic result verification and pattern library dynamic optimization module includes a diagnostic result gold standard backtesting verification submodule and a hierarchical pattern library iterative optimization submodule, wherein the diagnostic result gold standard backtesting verification submodule specifically includes: The purpose of retrospective collection of effective diagnostic samples is to collect subsequent liver biopsy verification data and clinical follow-up physiological signal re-examination data of subjects who have been given graded diagnostic results. The time-series physiological signal feature set and corresponding graded results confirmed by the gold standard are then added to the corresponding graded pattern library samples.
[0015] Preferably, the hierarchical pattern library iterative optimization submodule specifically includes: To achieve self-optimization of the system's diagnostic capabilities, specifically: when the number of new samples in each level of the pattern library reaches a preset number, the hierarchical pattern clustering and boundary fixing operations are re-executed to update the core cluster centers and pattern boundary ranges of each level.
[0016] Compared with the prior art, the present invention provides a liver fibrosis grading system based on serum index data, which has the following beneficial effects: This liver fibrosis grading system based on serum index data sets up a standardized oral amino acid loading procedure, collects serum biomarker sequences at different time points before and after the loading, and extracts dynamic physiological response characteristics using baseline normalization technology. This effectively isolates the interference of individual basal metabolic differences, diet, and environmental factors on the test results. Compared with static detection methods at a single time point, it significantly reduces the volatility of test results and ensures the stability and reliability of liver fibrosis grading.
[0017] This liver fibrosis grading system based on serum index data constructs a gold standard matching pattern library covering all Metavir grades and solidifies the exclusive feature boundaries and core cluster centers of each grade through pattern clustering algorithms. During the judgment, an exclusive pattern matching strategy is adopted, and the result is output only when both the matching degree and the threshold condition are met. This effectively solves the problems of blurred cross-grade boundaries and single model adapting to all grades in existing technologies, and achieves accurate and quantitative identification of liver fibrosis grades, significantly improving the diagnostic accuracy and specificity.
[0018] This liver fibrosis grading system based on serum index data has the functions of gold standard retrospective verification of diagnostic results and dynamic optimization of the pattern library. It can supplement the pattern library in real time with subsequent biopsy verification data and follow-up physiological signal data. The grading criteria and boundary range are optimized through periodic iterative cluster analysis. This closed-loop mechanism enables the diagnostic capability to continuously improve with the increase of sample size, breaking through the bottleneck of fixed parameters and difficulty in optimizing accuracy in traditional diagnostic models. Under long-term application, the diagnostic accuracy shows a steady upward trend.
[0019] This liver fibrosis grading system based on serum index data utilizes a completely non-invasive serum testing method, enabling accurate grading without invasive liver biopsy, thus avoiding the risks and costs associated with biopsy. Furthermore, through standardized process design and intelligent module linkage, it automates the entire process from signal acquisition, preprocessing, feature extraction to judgment output, significantly shortening the diagnostic cycle and reducing reliance on the professional skills of clinical operators. It is suitable for large-scale population screening, dynamic follow-up, and promotion in primary healthcare institutions, possessing extremely high clinical application value and socio-economic benefits. Attached Figure Description
[0020] Figure 1 This is a system block diagram of the present invention. Detailed Implementation
[0021] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.
[0022] Example 1: like Figure 1 As shown, a liver fibrosis grading system based on serum index data includes a standardized time-series acquisition module for liver fibrosis-associated serum physiological signals, a gold standard-paired liver fibrosis grading pattern library construction and pattern clustering module, a subject physiological signal pattern matching and grading determination module, and a diagnostic result verification and pattern library dynamic optimization module. The standardized time-series acquisition module for liver fibrosis-associated serum physiological signals is used to complete standardized time-series serum sample collection and multi-marker detection. After interference removal, baseline anchoring and normalization processing, individual difference interference is eliminated, and a liver fibrosis-associated time-series physiological signal feature set is output. The liver fibrosis-related serum physiological signal standardization time-series acquisition module includes a standardized physiological response triggering and serum sample time-series acquisition submodule and a serum physiological signal de-interference and baseline anchoring submodule. The standardized physiological response triggering and serum sample time-series acquisition submodule specifically includes: This method is used to trigger a liver-specific metabolic response through standardized physiological load, thereby collecting diagnostic physiological signals. Specifically, a standardized oral amino acid loading procedure is performed on the subject, with strict control over the loading dose, duration of administration, and pre-fasting conditions. Fasting venous serum samples are collected sequentially at three fixed time points: 0 hours before loading, 2 hours after loading, and 4 hours after loading. Each sample is simultaneously tested for 21 serum physiological markers associated with liver fibrosis, covering four major categories: collagen metabolism markers, liver enzyme markers, inflammatory response markers, and liver synthetic function markers. All samples are tested using the same testing equipment and the same batch of reagents, and the same batch of quality control samples are calibrated simultaneously to eliminate detection system errors, forming a time-series serum physiological signal sequence specific to the subject.
[0023] The serum sample time-series acquisition submodule specifically includes: The preprocessing of physiological signals for diagnosis is carried out as follows: the serum biomarker detection values of the subject 0 hours before the load are used as the individual physiological baseline. The relative changes of all biomarker physiological signals at the last two time points are normalized to remove signal interference caused by individual basic physiological differences of the subject. The dynamic physiological response characteristics of each biomarker and the synergistic change correlation characteristics between multiple biomarkers are extracted to finally form a subject-specific time-series physiological signal feature set associated with liver fibrosis.
[0024] The gold standard paired liver fibrosis grading pattern library construction and pattern clustering module is used to collect gold standard paired grading samples into the library to build an initial pattern library. Through unsupervised pattern clustering, core cluster centers and pattern boundaries of each grade are generated, and grade-specific standard physiological signal pattern clusters are solidified. The gold-standard paired liver fibrosis grading pattern library construction and pattern clustering module includes a gold-standard paired physiological signal sample library entry submodule and a grading-specific physiological signal pattern clustering and boundary solidification submodule. The gold-standard paired physiological signal sample library entry submodule specifically includes: To construct a benchmark sample set for diagnosis, specifically: collect a standardized time-series serum physiological signal feature set of subjects diagnosed by liver biopsy gold standard, covering five stages of Metavir grade F0-F4, and add a fixed number of valid samples to the database for each grade. Each sample is simultaneously labeled with the corresponding biopsy grade gold standard result and etiology type information to form an initial pattern library with gold standard diagnostic labels.
[0025] The pattern clustering module specifically includes: This method is used to construct a standardized diagnostic grading benchmark through pattern clustering technology. Specifically, for all physiological signal feature sets of the same grade in the initial pattern library, an unsupervised pattern clustering method is used to generate the core feature cluster center for each liver fibrosis grade. At the same time, the boundary range of physiological signal patterns for each grade is defined, outliers across grades are removed, and finally, a unique standard physiological signal pattern cluster corresponding to each liver fibrosis grade is formed.
[0026] The subject physiological signal pattern matching and grading determination module is used to match the subject's temporal physiological signal feature set with each grading standard pattern cluster, calculate the corresponding matching degree, and output the liver fibrosis grading diagnosis result after exclusion verification. If the condition is not met, a standardized re-examination is triggered. The subject physiological signal pattern matching and grading module includes a time-series physiological signal pattern matching degree calculation submodule and a grading result exclusive diagnostic output submodule. The time-series physiological signal pattern matching degree calculation submodule specifically includes: Diagnostic feature matching is achieved through pattern classification technology. Specifically, the subject-specific time-series physiological signal feature set output by the preceding module is matched sequentially with five graded standard physiological signal pattern clusters in the pattern library. The matching degree between the subject feature set and the core cluster center of each grade is calculated. At the same time, it is verified whether the subject feature set falls within the pattern boundary range of the corresponding grade, and the matching degree value corresponding to each grade is output.
[0027] The exclusive diagnostic output submodule for grading results specifically includes: To achieve the final diagnostic result output, the following steps are taken: sort all the matching degree values corresponding to the grades from high to low, and output the final liver fibrosis grading diagnosis result only when the matching degree value of the grade corresponding to the highest matching degree exceeds the preset threshold of the second highest matching degree value, and the subject's feature set falls completely within the pattern boundary range of that grade; if the exclusion condition is not met, the standardized re-examination process of the subject is directly triggered to avoid misjudgment of grading.
[0028] The diagnostic result verification and pattern library dynamic optimization module is used to retrospectively collect valid diagnostic samples verified by the gold standard to supplement the library. When the preset sample size is reached, the hierarchical cluster centers and pattern boundaries are iteratively updated to form a self-optimization closed loop that continuously improves diagnostic accuracy. The diagnostic result validation and pattern library dynamic optimization module includes a diagnostic result gold standard backtesting validation submodule and a hierarchical pattern library iterative optimization submodule. The diagnostic result gold standard backtesting validation submodule specifically includes: The purpose of retrospective collection of effective diagnostic samples is to collect subsequent liver biopsy verification data and clinical follow-up physiological signal re-examination data of subjects who have been given graded diagnostic results. The time-series physiological signal feature set and corresponding graded results confirmed by the gold standard are then added to the corresponding graded pattern library samples.
[0029] The hierarchical pattern library iterative optimization submodule specifically includes: To achieve self-optimization of the system's diagnostic capabilities, specifically: when the number of new samples in each level of the pattern library reaches a preset number, the hierarchical pattern clustering and boundary fixing operations are re-executed, updating the core cluster centers and pattern boundary ranges of each level, so that the system's hierarchical diagnostic accuracy continues to improve with the increase in the number of samples, forming a complete self-optimization closed loop for diagnostic capabilities.
[0030] Example 2: A liver fibrosis grading system based on serum index data, when used: Standardized oral amino acid loading procedures were performed on the subjects. Fasting venous serum samples were collected at three fixed time points: 0 hours before loading, 2 hours after loading, and 4 hours after loading. Multiple serum physiological markers associated with liver fibrosis were detected using standardized testing equipment and reagents to obtain time-series serum index data. The subject's time-series serum index data is input into a liver fibrosis grading system based on serum index data. The system uses the detection value at 0 hours before the load as the individual's physiological baseline, and performs normalization and interference removal processing on the indexes at subsequent time points to generate the subject's time-series physiological signal feature set. The system matches the subject's temporal physiological signal feature set with the built-in, gold-standard paired liver fibrosis grading pattern library, calculates the matching degree with each grading cluster center, and determines whether it falls within the boundary range of the corresponding grading pattern. The system compares the matching results against a threshold according to the exclusivity judgment rule. When the judgment condition is met, the liver fibrosis grading result is output. If the condition is not met, a standardized re-examination is prompted. Regularly import valid sample data verified by the gold standard into the system. The system will automatically perform pattern clustering and hierarchical boundary updates to dynamically optimize the hierarchical pattern library and improve the accuracy of subsequent judgments.
[0031] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.
Claims
1. A liver fibrosis grading system based on serum index data, comprising a standardized time-series acquisition module for liver fibrosis-related serum physiological signals, a gold standard-paired liver fibrosis grading pattern library construction and pattern clustering module, a subject physiological signal pattern matching and grading determination module, and a diagnostic result verification and pattern library dynamic optimization module, characterized in that: The standardized time-series acquisition module for liver fibrosis-related serum physiological signals is used to complete the standardized time-series collection of serum samples and the detection of multiple biomarkers. After interference removal, baseline anchoring and normalization, individual difference interference is eliminated, and a feature set of liver fibrosis-related time-series physiological signals is output. The gold standard paired liver fibrosis grading pattern library construction and pattern clustering module is used to collect gold standard paired grading samples into the library to construct an initial pattern library. Through unsupervised pattern clustering, core cluster centers and pattern boundaries for each grade are generated, and grade-specific standard physiological signal pattern clusters are solidified. The subject physiological signal pattern matching and grading determination module is used to match the subject's temporal physiological signal feature set with each grading standard pattern cluster, calculate the corresponding matching degree, output the liver fibrosis grading diagnosis result after exclusion verification, and trigger standardized re-examination if the conditions are not met. The diagnostic result verification and pattern library dynamic optimization module is used to retrospectively collect valid diagnostic samples verified by the gold standard to supplement the library. When the preset sample size is reached, the hierarchical cluster centers and pattern boundaries are iteratively updated to form a self-optimizing closed loop that continuously improves diagnostic accuracy.
2. The liver fibrosis grading system based on serum index data according to claim 1, characterized in that: The liver fibrosis-related serum physiological signal standardization time-series acquisition module includes a standardized physiological response triggering and serum sample time-series acquisition submodule and a serum physiological signal de-interference and baseline anchoring submodule. The standardized physiological response triggering and serum sample time-series acquisition submodule specifically includes: It is used to trigger liver-specific metabolic responses through standardized physiological loads, thereby enabling the acquisition of diagnostic physiological signals.
3. The liver fibrosis grading system based on serum index data according to claim 2, characterized in that: The serum sample time-series acquisition submodule specifically includes: The preprocessing of physiological signals for diagnosis is carried out as follows: the serum biomarker detection values of the subject 0 hours before the load are used as the individual physiological baseline. The relative changes of all biomarker physiological signals at the last two time points are normalized to remove signal interference caused by individual basic physiological differences of the subject. The dynamic physiological response characteristics of each biomarker and the synergistic change correlation characteristics between multiple biomarkers are extracted to finally form a subject-specific time-series physiological signal feature set associated with liver fibrosis.
4. The liver fibrosis grading system based on serum index data according to claim 1, characterized in that: The gold standard paired liver fibrosis grading pattern library construction and pattern clustering module includes a gold standard paired physiological signal sample library entry submodule and a grading-specific physiological signal pattern clustering and boundary solidification submodule. The gold standard paired physiological signal sample library entry submodule specifically includes: To construct a benchmark sample set for diagnosis, specifically: collect a standardized time-series serum physiological signal feature set of subjects diagnosed by liver biopsy gold standard, covering five stages of Metavir grade F0-F4, and add a fixed number of valid samples to the database for each grade. Each sample is simultaneously labeled with the corresponding biopsy grade gold standard result and etiology type information to form an initial pattern library with gold standard diagnostic labels.
5. The liver fibrosis grading system based on serum index data according to claim 4, characterized in that: The pattern clustering module specifically includes: This method is used to construct a standardized diagnostic grading benchmark through pattern clustering technology. Specifically, for all physiological signal feature sets of the same grade in the initial pattern library, an unsupervised pattern clustering method is used to generate the core feature cluster center for each liver fibrosis grade. At the same time, the boundary range of physiological signal patterns for each grade is defined, outliers across grades are removed, and finally, a unique standard physiological signal pattern cluster corresponding to each liver fibrosis grade is formed.
6. The liver fibrosis grading system based on serum index data according to claim 5, characterized in that: The subject physiological signal pattern matching and grading determination module includes a time-series physiological signal pattern matching degree calculation submodule and a grading result exclusive diagnostic output submodule, wherein the time-series physiological signal pattern matching degree calculation submodule specifically includes: Diagnostic feature matching is achieved through pattern classification technology. Specifically, the subject-specific time-series physiological signal feature set output by the preceding module is matched sequentially with five graded standard physiological signal pattern clusters in the pattern library. The matching degree between the subject feature set and the core cluster center of each grade is calculated. At the same time, it is verified whether the subject feature set falls within the pattern boundary range of the corresponding grade, and the matching degree value corresponding to each grade is output.
7. A liver fibrosis grading system based on serum index data according to claim 6, characterized in that: The exclusive diagnostic output submodule for the grading results specifically includes: To achieve the final diagnostic result output, the following steps are taken: sort all the matching degree values corresponding to the grades from high to low, and output the final liver fibrosis grading diagnosis result only when the matching degree value of the grade corresponding to the highest matching degree exceeds the preset threshold of the second highest matching degree value, and the subject's feature set falls completely within the pattern boundary range of that grade; if the exclusion condition is not met, the standardized re-examination process of the subject is directly triggered.
8. The liver fibrosis grading system based on serum index data according to claim 1, characterized in that: The diagnostic result verification and pattern library dynamic optimization module includes a diagnostic result gold standard backtesting verification submodule and a hierarchical pattern library iterative optimization submodule, wherein the diagnostic result gold standard backtesting verification submodule specifically includes: The purpose of retrospective collection of effective diagnostic samples is to collect subsequent liver biopsy verification data and clinical follow-up physiological signal re-examination data of subjects who have been given graded diagnostic results. The time-series physiological signal feature set and corresponding graded results confirmed by the gold standard are then added to the corresponding graded pattern library samples.
9. A liver fibrosis grading system based on serum index data according to claim 8, characterized in that: The hierarchical pattern library iterative optimization submodule specifically includes: To achieve self-optimization of the system's diagnostic capabilities, specifically: when the number of new samples in each level of the pattern library reaches a preset number, the hierarchical pattern clustering and boundary fixing operations are re-executed to update the core cluster centers and pattern boundary ranges of each level.