Method and system for predicting myocarditis based on protein expression

Abstract

The invention discloses a method and a system for predicting myocarditis based on protein expression. The method comprises the following steps: collecting a blood sample of a subject, quantitatively detecting expression levels of cardiac muscle specificity and inflammation related proteins such as cTnI, H-FABP and SAA at the same time by adopting an enzyme-linked immunosorbent assay, carrying out standardized correction by utilizing an internal reference protein, and inputting corrected data into a multivariable logistic regression model based on large-scale population data training. And calculating a comprehensive risk score. The system comprises a sample pretreatment module, a multiple protein quantitative detection module, an analysis module and a clinical report generation module, and supports automatic detection, dynamic weight adjustment and trend analysis. Through multi-marker combined analysis and modeling integration, the specificity and early warning capability of myocarditis diagnosis are remarkably improved, risk stratification and personalized treatment assistance are realized, and the limitation that a traditional method depends on a single marker and subjective experience is overcome.

Description

Technical Field

[0001] This application relates to techniques for predicting myocarditis, specifically to a method and system for predicting myocarditis based on protein expression. Background Technology

[0002] Currently, the clinical diagnosis of myocarditis mainly relies on a comprehensive and exclusionary set of criteria, the core of which lies in combining clinical manifestations, non-specific laboratory tests, cardiac imaging, and, when necessary, endocardial biopsy (EMB). A typical diagnostic pathway begins with the identification of suspected symptoms (such as chest pain, dyspnea, palpitations, and fatigue), combined with non-specific ST-T changes, conduction blocks, or various arrhythmias that may be shown on electrocardiogram (ECG). In terms of laboratory tests, elevated serum myocardial injury markers, particularly cardiac troponin I (cTnI or cTnT) and creatine kinase isoenzyme (CK-MB), are key indicators of myocardial cell damage. Echocardiography can assess cardiac structure and function, detecting abnormal wall motion, enlarged cardiac chambers, or pericardial effusion. Cardiac magnetic resonance imaging (CMR), with its superior ability to characterize tissues, can non-invasively detect myocardial inflammation, edema, and fibrosis using T2-weighted imaging sequences such as edema, early gadolinium enhancement (EGE), and late gadolinium enhancement (LGE), and has become an important auxiliary tool. EMB is considered the "gold standard" for histological diagnosis, providing direct evidence of inflammatory cell infiltration and cardiomyocyte necrosis.

[0003] However, this traditional diagnostic system has many inherent flaws. First, its diagnostic specificity is severely lacking. Clinical symptoms, abnormal electrocardiograms, and elevated serum cTnI are not unique to myocarditis; they can occur in acute coronary syndrome, stress-induced cardiomyopathy, other types of cardiomyopathy, and even in systemic stress states associated with severe infection, leading to a persistently high misdiagnosis rate. Second, it lacks effective means for early warning and accurate assessment of disease activity. While cTnI is a reliable marker of myocardial necrosis, its elevation often lags behind functional impairment and the onset of inflammation, and it cannot distinguish whether the damage is caused by inflammation, ischemia, or other factors. Although CMR provides rich morphological and functional information, its sensitivity to early, focal inflammation is limited, and the high cost of equipment, long examination time, and contraindications (such as implanted metal devices and severe renal insufficiency) limit its widespread and repeated application. Furthermore, differential diagnosis, especially the differentiation of dilated cardiomyopathy (DCM), which highly overlaps with clinical manifestations, is particularly difficult. Both can present with heart failure, cardiac enlargement, and elevated markers of myocardial injury, but their etiologies, pathological mechanisms, treatment strategies, and prognoses are drastically different. Currently, there is a lack of reliable biomarkers that can effectively distinguish between the two at the molecular level, often leading to delays or biases in treatment decisions. Finally, treatment lacks targeting and prognostic predictive capabilities. Current treatments are mostly empirical supportive care (such as rest, anti-heart failure medication, and anti-arrhythmic drugs) or non-specific immunomodulation (such as glucocorticoids and intravenous immunoglobulins). Because it is impossible to assess the individualized immune inflammation activation status and myocardial injury repair process in real time and quantitatively, treatment plans are often developed blindly, failing to achieve "personalized precision medicine." Some patients do not respond well to treatment, while others may suffer from unnecessary immunosuppression and the risk of side effects such as infection. Summary of the Invention

[0004] In view of this, the present invention provides a method and system for predicting myocarditis based on protein expression to solve the above-mentioned technical problems.

[0005] A method for predicting myocarditis based on protein expression includes the following steps:

[0006] Peripheral blood or serum samples were collected from the subjects, and plasma or serum components were obtained by centrifugation to obtain the test samples;

[0007] The protein expression concentration data of the myocardial specific protein in the test sample were simultaneously and quantitatively detected using enzyme-linked immunosorbent assay (ELISA).

[0008] The obtained protein expression concentration data are input into a pre-trained multivariate logistic regression model, which is constructed based on a large-scale protein expression database of healthy people and patients diagnosed with myocarditis, and outputs a comprehensive risk score.

[0009] The comprehensive risk score is compared with a preset risk threshold. If the score is higher than the threshold, the subject is determined to have a risk of myocarditis or is already in the active phase of myocarditis. The risk level is then determined based on the extent to which the score exceeds the threshold.

[0010] Furthermore, the multivariate logistic regression model predicts myocarditis risk based on the expression levels of multi-level myocardial structural proteins and inflammation-related proteins. Before prediction, the expression levels of multiple endogenous reference proteins are detected to correct the target protein expression level. The correction methods include:

[0011] The concentration values ​​of the target protein and the internal control protein were obtained separately;

[0012] Calculate the ratio of the expression level of each target protein to that of the selected internal reference protein to obtain the corrected relative expression level;

[0013] Using the average of the corresponding ratios from a healthy control group as a benchmark, the standardized fold change of the relative expression levels of each protein in the test sample was calculated to eliminate differences in total protein content between samples.

[0014] Furthermore, the construction of the multivariate logistic regression model and the calculation method of the comprehensive risk score include:

[0015] Historical cohort data were collected, including samples from healthy individuals, patients diagnosed with myocarditis, and other patients with myocardial diseases, and the expression levels of cTnI, H-FABP, SAA, and other optional candidate proteins were detected.

[0016] After standardizing and normalizing the data, logistic regression analysis was used to determine the weight coefficients of the contribution of each protein expression level to the diagnosis of myocarditis.

[0017] Establish a comprehensive risk scoring formula: Risk Score = W1 × [cTnI] + W2 × [H-FABP] + W3 × [SAA] + … + W n × [P n ], where W1, W2, W3, ... W n [P] represents the weighting coefficients for each protein. n [ ] represents the standardized expression level of the nth protein;

[0018] The optimal risk score cutoff value was determined by receiver operating characteristic (ROC) curve analysis to maximize the sensitivity value that distinguishes patients with myocarditis from healthy individuals.

[0019] The present invention also provides a system for predicting myocarditis based on protein expression, comprising:

[0020] The sample preprocessing module is used to receive blood samples from subjects, and obtain plasma or serum components through centrifugation to obtain the test sample;

[0021] A multiplex protein quantification module, connected to the sample preprocessing module, is equipped with a chemiluminescent immunoassay analyzer for parallel quantitative detection of the expression levels of a set of pre-set myocarditis-related protein markers in the sample, to obtain protein expression concentration data;

[0022] Analysis module: Connected to the multiplex protein quantification module, configured as follows:

[0023] Receive protein expression concentration data from the detection module;

[0024] Standardize and correct the data by calling the internal reference protein data;

[0025] The corrected data is input into the built-in multivariate logistic regression model to calculate the comprehensive risk score.

[0026] Based on the scoring and preset rules, structured prediction results are generated, including risk level, probability of disease state, and differential diagnosis suggestions.

[0027] Clinical report generation module: This module converts the structured prediction results into graphic reports that can be directly interpreted by doctors and outputs them in print via a network interface.

[0028] Furthermore, the multiplex protein quantification module includes:

[0029] The high-throughput protein analysis unit is equipped with prefabricated myocarditis protein detection strips, each strip having specific antibodies against cTnI, H-FABP, SAA, and internal reference proteins ALB and B2M immobilized on it.

[0030] The signal acquisition device simultaneously quantifies multiple proteins in a batch of samples, and then transmits them to a high-throughput protein analysis unit to obtain multiple chemiluminescent signals.

[0031] The signal analysis unit converts the collected chemiluminescence signals into relative optical density values ​​for each protein and automatically calculates the ratio of the target protein to the internal reference protein.

[0032] Furthermore, the analysis module includes:

[0033] Reference database, storing protein expression reference profiles from multicenter clinical studies;

[0034] The dynamic weight adjustment unit is configured to fine-tune the weight coefficients of each protein in the prediction model based on the clinical information of the subjects.

[0035] The prognostic trend analysis unit receives multiple test data from the same patient at different time points and plots curves showing the changes in key protein expression levels and comprehensive risk scores over time.

[0036] The multiplex protein quantification module includes a result reliability assessment unit, which is configured as follows:

[0037] Real-time monitoring of the chemiluminescence signal intensity fluctuation and background noise ratio of each sample;

[0038] If the signal-to-noise ratio of any target protein is lower than a preset threshold, a retest command is automatically triggered, and the detection device is controlled to retest the sample.

[0039] Simultaneously, a data quality flag is sent to the analysis module. After receiving the flag, the analysis module automatically reduces the weight of the low-quality protein data when calculating the comprehensive risk score.

[0040] Furthermore, the dynamic weight adjustment unit of the analysis module and the signal analysis unit of the multiplex protein quantification module establish a bidirectional feedback logic: after completing the detection of a batch of samples, the signal analysis unit sends the overall signal stability index of this detection to the dynamic weight adjustment unit; the dynamic weight adjustment unit judges the quality level of this batch of detection based on the index: if the index shows high stability, the original model weight is maintained; if the index shows moderate fluctuation, the weight of the internal reference protein correction factor is automatically increased; if the index shows high instability, the use of this batch of data is suspended.

[0041] Furthermore, the analysis module is equipped with self-testing logic, which includes...

[0042] a) Extract recent protein expression data and final clinical diagnosis results for all subjects from the reference database of the analysis module;

[0043] b) Use these data to iteratively calibrate the chemiluminescence signal-concentration conversion curve used in the multiplex protein quantification module;

[0044] c) Using the calibrated detection data, retrain the multivariate logistic regression model in the analysis module and update the weight coefficients and risk thresholds of each protein;

[0045] d) The updated model and transformation curve parameters are synchronously distributed to the detection and analysis modules of each terminal to complete a system-level self-optimization cycle.

[0046] Furthermore, the prognostic trend analysis unit embeds a pattern recognition subunit, which is configured as follows:

[0047] Continuous analysis was conducted on the changes and relative proportions of cTnI, H-FABP, and SAA in multiple tests of the same patient.

[0048] When a separation pattern of continuously decreasing cTnI and continuously increasing SAA is detected, an alarm is automatically triggered and this signal is pushed to the dynamic weight adjustment unit.

[0049] In response to this alert, the dynamic weighting adjustment unit temporarily increases the weighting coefficient of SAA in subsequent scoring and generates targeted differential diagnosis suggestions.

[0050] This application utilizes simultaneous quantitative detection and combined analysis of specific structural proteins reflecting cardiomyocyte injury (cTnI), cytoplasmic proteins released early in the injury phase (H-FABP), and acute-phase proteins (SAA) characterizing the intensity of systemic inflammatory response. In myocarditis, myocardial injury and immune inflammation are closely coupled core pathological processes. This invention quantifies this coupling relationship by constructing a comprehensive risk scoring model. When a characteristic pattern of simultaneous and significant increases in cTnI and SAA is detected, it strongly supports that the injury originates from an inflammatory process, rather than ischemia or other causes, thereby greatly improving diagnostic specificity. Simultaneously, the combination of early changes in H-FABP and the rapid response characteristics of SAA allows this system to issue early warning signals at an earlier stage, when clinical symptoms are atypical or traditional markers have not yet shown significant abnormalities, creating conditions for early intervention.

[0051] Through an internal control calibration process, endogenous reference proteins such as albumin (ALB) and β-2 microglobulin (B2M) are simultaneously detected to perform ratio correction on the expression levels of target proteins. This effectively eliminates differences in total protein concentration between samples and preprocessing errors, ensuring data comparability. Based on a large-scale big data benchmark and algorithm, all detection results are converted into standardized fold changes relative to the baseline of healthy individuals and input into a multivariate prediction model trained on a large historical sample cohort, outputting a continuous comprehensive risk score. This score is derived from objective data and statistical models, avoiding subjective judgment. Attached Figure Description

[0052] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only preferred embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0053] Figure 1 This is a flowchart of the method of the system of the present invention;

[0054] Figure 2This is a flowchart illustrating the method for constructing a multivariate logistic regression model and calculating a comprehensive risk score as described in this invention. Detailed Implementation

[0055] To better understand the structure of the present invention and the functional features and advantages it can achieve, the preferred embodiments of the present invention will be described in detail below with reference to the drawings.

[0056] Reference Figures 1 to 2 A method for predicting myocarditis based on protein expression includes the following steps: collecting peripheral blood or serum samples from subjects and obtaining plasma or serum components by centrifugation to obtain test samples; using enzyme-linked immunosorbent assay (ELISA) to simultaneously and quantitatively detect the expression levels of at least three myocardial-specific proteins in the obtained test samples, including: cardiac troponin I (cTnI), heart-type fatty acid-binding protein (H-FABP), and serum amyloid A (SAA), an inflammation-related protein; inputting the obtained protein expression concentration data into a pre-trained multivariate logistic regression model, which is constructed based on a large-scale protein expression database of healthy individuals and patients diagnosed with myocarditis, and outputting a comprehensive risk score; comparing the comprehensive risk score with a preset risk threshold, and if the score is higher than the threshold, determining that the subject has a risk of myocarditis or is already in the active phase of myocarditis; based on the extent to which the score exceeds the threshold, further subdividing the risk level into low risk, medium risk, and high risk, and providing corresponding clinical follow-up or intervention recommendations for each level.

[0057] In the above, the multivariate logistic regression model predicts the risk of myocarditis based on the expression levels of multilevel myocardial structural proteins and inflammation-related proteins. Before prediction, it includes detecting the expression levels of at least two endogenous reference proteins to correct the expression level of the target protein. The endogenous reference proteins are selected from at least two of albumin (ALB), transferrin (TF), or β-2 microglobulin (B2M). The correction method includes: obtaining the concentration values ​​of the target protein and the internal reference protein respectively; calculating the ratio of the expression level of each target protein to the selected internal reference protein to obtain the corrected relative expression level; and using the average of the corresponding ratios from a healthy control population as a benchmark, calculating the standardized fold change of the relative expression level of each protein in the test sample.

[0058] In the above, the construction of the multivariate logistic regression model and the calculation method of the comprehensive risk score include: collecting historical cohort data, including samples from healthy individuals, patients diagnosed with myocarditis, and other patients with myocardial diseases, and detecting the expression levels of cTnI, H-FABP, SAA, and other optional candidate proteins; after standardizing and normalizing the data, using logistic regression analysis to determine the weight coefficients of each protein expression level in the diagnosis of myocarditis; and establishing the comprehensive risk score formula: Risk Score = W1 × [cTnI] + W2 × [H-FABP] + W3 × [SAA] + … + W n ×[P n ], where W1, W2, W3, ... W n [P] represents the weighting coefficients for each protein. n [ ] represents the standardized expression level of the nth protein; the optimal risk score cutoff value was determined by receiver operating characteristic (ROC) curve analysis to maximize the sensitivity and specificity in distinguishing myocarditis patients from healthy individuals.

[0059] The meanings represented by the formulas above are shown in the table below.

[0060]

[0061] In the above, the construction and preprocessing of historical cohort data includes: the historical cohort must cover the following four groups: Group A: healthy controls (n≥200); Group B: patients with clinically diagnosed acute myocarditis (n≥150); Group C: patients with other cardiomyopathies (e.g., dilated cardiomyopathy, DCM, n≥100); Group D: patients with non-myocarditis chest pain (e.g., stable angina, n≥100). All samples underwent protein detection and correction to obtain the FC value for each protein in each sample. The FC values ​​were logarithmically transformed (log2(FC)) to better conform to a normal distribution, facilitating statistical analysis. Acute myocarditis was used as the binary dependent variable (1=yes, 0=no), and FC_cTnI, FC_H-FABP, FC_SAA, etc., were used as independent variables. Logistic regression was used for modeling. The model used stepwise regression to select the protein combination that contributed most to differentiating myocarditis and estimated its regression coefficient β. The magnitude and direction (positive or negative) of the β value directly reflect the contribution weight and direction of the protein marker in the diagnosis of myocarditis. For example, the β values ​​of cTnI and SAA may be positive and large, while the β value of the correction factor for ALB may be negative. A negative β value reflects mild hypoalbuminemia that may occur in an inflammatory state.

[0062] In the above, after the initial assessment determines the risk of myocarditis or after a diagnosis, the protein expression-based method for predicting myocarditis is repeatedly measured at different treatment stages or time points. By comparing the trend of changes in the comprehensive risk score obtained at consecutive time points, the disease progression, stability, or remission status is assessed. If the score shows a continuous downward trend, it indicates that the treatment is effective or the disease has entered the recovery period. If the score remains high or repeatedly rises, it indicates that the disease activity is ongoing or the treatment response is poor, and the treatment plan needs to be adjusted.

[0063] In the above, the multivariate model also integrates protein expression characteristic data of the patient cohort with dilated cardiomyopathy; when the subject's overall risk score is higher than the myocarditis risk threshold, and their protein expression pattern (e.g., significantly elevated SAA accompanied by moderately elevated H-FABP) is more consistent with the typical characteristics of myocarditis in the model than the characteristics of DCM (e.g., mainly with significantly elevated myocardial fibrosis-related proteins such as galactoglobulin-3), the system generates a differential diagnosis conclusion that supports the diagnosis of myocarditis.

[0064] This application also discloses a protein expression-based system for predicting myocarditis, comprising: a sample preprocessing module for receiving blood samples from subjects and automatically completing standardized preprocessing procedures such as centrifugation, plasma separation, and aliquoting, outputting plasma or serum samples that meet downstream detection requirements; a multiplex protein quantification module connected to the sample preprocessing module, integrating a protein immunoblotting or chemiluminescent immunoassay platform for parallel quantitative detection of the expression levels of a pre-set set of myocarditis-related protein markers in the sample; an analysis module connected to the detection module, configured to: receive raw protein expression data from the detection module; call internal reference protein data for standardized correction; input the corrected data into a built-in multivariate prediction model to calculate a comprehensive risk score; generate structured prediction results including risk level, disease state probability, and differential diagnosis suggestions based on the score and pre-set rules; and a clinical report generation module for converting the structured prediction results into a graphic report that can be directly interpreted by doctors and outputting it via a network interface or local printing.

[0065] In the above, the multiplex protein quantification module includes: a high-throughput protein analysis unit equipped with prefabricated myocarditis protein detection strips, each strip having immobilized specific antibodies against cTnI, H-FABP, SAA, and internal reference proteins ALB and B2M; a chemiluminescence signal acquisition device capable of simultaneously quantifying multiple proteins in a batch of samples; and a signal analysis unit capable of converting the acquired chemiluminescence signals into relative optical density values ​​for each protein and automatically calculating the ratio of the target protein to the internal reference protein.

[0066] The multiplex protein quantification module includes a result reliability assessment unit, which is configured to: monitor the chemiluminescence signal intensity fluctuation and background noise ratio of each sample in real time; if the signal-to-noise ratio of any target protein is lower than a preset threshold, automatically trigger a retest command and control the detection device to retest the sample; simultaneously, send a data quality flag to the analysis module, which, upon receiving the flag, automatically reduces the weight of the low-quality protein data when calculating the comprehensive risk score; and establish a bidirectional feedback logic between the dynamic weight adjustment unit of the analysis module and the signal analysis unit of the multiplex protein quantification module: after completing a batch of sample testing, the signal analysis unit sends the overall signal stability index of this test to the dynamic weight adjustment unit; the dynamic weight adjustment unit determines the quality level of this batch of tests based on this index: if the index shows high stability, the original model weight is maintained; if the index shows moderate fluctuation, the weight of the internal reference protein correction factor is automatically increased; if the index shows high instability, the use of this batch of data is suspended.

[0067] Furthermore, the analysis module is equipped with self-checking logic, which includes: a) extracting protein expression data and final clinical diagnosis results of all recent subjects from the reference database of the analysis module; b) using these data to iteratively calibrate the chemiluminescence signal-concentration conversion curve used by the multiplex protein quantification detection module; c) using the calibrated detection data to retrain the multivariate logistic regression model in the analysis module and update the weight coefficients and risk thresholds of each protein; d) synchronously sending the updated model and conversion curve parameters to each terminal detection module and analysis module to complete a system-level self-optimization cycle.

[0068] Furthermore, the prognostic trend analysis unit embeds a pattern recognition subunit, which is configured to: continuously analyze the change trajectory and relative proportion of cTnI, H-FABP, and SAA in multiple tests of the same patient; when a separation pattern of continuously decreasing cTnI and continuously increasing SAA is identified, an alarm is automatically triggered and this signal is pushed to the dynamic weight adjustment unit; the dynamic weight adjustment unit responds to this alarm, temporarily increases the weight coefficient of SAA in subsequent scoring, and generates targeted differential diagnosis prompts.

[0069] In the above, the analysis module includes: a reference database storing protein expression reference profiles from large-scale multicenter clinical studies, including baseline data of healthy individuals, active phase data and recovery phase data of patients with acute myocarditis, and control data of patients with other myocardial diseases such as dilated cardiomyopathy; a dynamic weight adjustment unit configured to fine-tune the weight coefficients of each protein in the prediction model based on the clinical information of the subjects (such as the time of symptom onset and whether they received treatment), for example, giving more weight to H-FABP in the very early stages of the disease and giving more weight to SAA when assessing inflammatory activity; and a prognostic trend analysis unit capable of receiving multiple test data from the same patient at different time points, plotting curves of key protein expression levels and comprehensive risk scores over time, and automatically analyzing trends.

[0070] In the aforementioned analysis module, a single calculation unit for the acute phase response index is also included. This unit is configured to: calculate an acute phase response index based on SAA and C-reactive protein (CRP) levels; perform a correlation analysis between this index and the level of myocardial injury protein (cTnI, H-FABP); when the acute phase response index is significantly elevated, accompanied by a moderate to severe elevation of myocardial injury protein, the diagnostic tendency for acute active myocarditis is strengthened; while when myocardial injury protein is significantly elevated but the acute phase response index is normal or only slightly elevated, it suggests that non-inflammatory myocardial injury should be prioritized for investigation. Furthermore, based on the patient's baseline specific protein expression pattern (e.g., extremely high SAA with moderate cTnI), pattern matching is performed with historical data of patients with good treatment responses; predicting the patient's potential response to immunomodulatory therapies such as glucocorticoids or intravenous immunoglobulins, providing auxiliary reference for the selection of initial treatment strategies.

[0071] In the above, the multiplex protein quantification module is also configured to optionally detect the level of soluble growth-stimulating gene 2 protein (sST2); correspondingly, the data integration and analysis engine is further configured to: integrate the expression level of sST2 into the prediction model or as an independent prognostic assessment parameter; combine the sST2 level (reflecting myocardial fibrosis and remodeling) with the level of a protein (cTnI, SAA) reflecting acute injury / inflammation to perform more refined risk stratification of patients and identify patient subgroups with high inflammation and high fibrosis risk who require intensive follow-up and intervention.

[0072] The clinical report generation module can automatically generate a report containing the following elements based on the results of the data analysis engine: (1) basic information of the subject and sample information; (2) quantitative results, reference ranges and abnormal markers of each detected protein; (3) the calculated comprehensive risk score and the corresponding risk level (e.g., low, medium and high); (4) differential diagnosis suggestions based on protein patterns (e.g., tending towards acute myocarditis rather than chronic ischemic cardiomyopathy); (5) trend analysis charts of patients under dynamic monitoring (e.g., for retest samples); and (6) a standardized clinical recommendation summary based on the current results.

[0073] The sample preprocessing module has an automatic sample quality assessment function. Before processing, it can automatically determine whether a sample is suitable for subsequent protein detection by detecting the hemolysis index (HI) and lipemia index (LI). For samples with severe hemolysis or lipemia, the system will issue a warning and suggest re-collection to ensure the reliability of the test results.

[0074] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any way. Any person skilled in the art can make many possible variations and modifications to the technical solutions of the present invention, or modify them into equivalent embodiments, without departing from the scope of the present invention. Therefore, any modifications, equivalent changes, and alterations made to the above embodiments based on the technology of the present invention without departing from the scope of the present invention are within the protection scope of the present invention.

Claims

1. A method for predicting myocarditis based on protein expression, characterized in that, Includes the following steps: Peripheral blood or serum samples were collected from the subjects, and plasma or serum components were obtained by centrifugation to obtain the test samples; The protein expression concentration data of the myocardial specific protein in the test sample were simultaneously and quantitatively detected using enzyme-linked immunosorbent assay (ELISA). The obtained protein expression concentration data are input into a pre-trained multivariate logistic regression model, which is constructed based on a large-scale protein expression database of healthy people and patients diagnosed with myocarditis, and outputs a comprehensive risk score. The comprehensive risk score is compared with a preset risk threshold. If the score is higher than the threshold, the subject is determined to have a risk of myocarditis or is already in the active phase of myocarditis. The risk level is then determined based on the extent to which the score exceeds the threshold.

2. The method for predicting myocarditis based on protein expression according to claim 1, characterized in that, The multivariate logistic regression model predicts myocarditis risk based on the expression levels of multilevel myocardial structural proteins and inflammation-related proteins. Before prediction, the expression levels of multiple endogenous reference proteins are detected to correct the target protein expression level. The correction methods include: The concentration values ​​of the target protein and the internal control protein were obtained separately; Calculate the ratio of the expression level of each target protein to that of the selected internal reference protein to obtain the corrected relative expression level; Using the average of the corresponding ratios from a healthy control group as a benchmark, the standardized fold change of the relative expression levels of each protein in the test sample was calculated to eliminate differences in total protein content between samples.

3. The method for predicting myocarditis based on protein expression according to claim 1 or 2, characterized in that, The construction of the multivariate logistic regression model and the calculation method of the comprehensive risk score include: Historical cohort data were collected, including samples from healthy individuals, patients diagnosed with myocarditis, and other patients with myocardial diseases, and the expression levels of cTnI, H-FABP, SAA, and other candidate proteins were detected. After standardizing and normalizing the data, logistic regression analysis was used to determine the weight coefficients of the contribution of each protein expression level to the diagnosis of myocarditis. Establish a comprehensive risk scoring formula: Risk Score = W1 × [cTnI] + W2 × [H-FABP] + W3 × [SAA] + … + W n × [P n ], where W1, W2, W3, ... W n [P] represents the weighting coefficients for each protein. n [ ] represents the standardized expression level of the nth protein; The optimal risk score cutoff value was determined by receiver operating characteristic (ROC) curve analysis to maximize the sensitivity value that distinguishes patients with myocarditis from healthy individuals.

4. A system for predicting myocarditis based on protein expression, characterized in that, include: The sample preprocessing module is used to receive blood samples from subjects, and obtain plasma or serum components through centrifugation to obtain the test sample; A multiplex protein quantification module, connected to the sample preprocessing module, is equipped with a chemiluminescent immunoassay analyzer for parallel quantitative detection of the expression levels of a set of pre-set myocarditis-related protein markers in the sample, to obtain protein expression concentration data; Analysis module: Connected to the multiplex protein quantification module, configured as follows: Receive protein expression concentration data from the detection module; Standardize and correct the data by calling the internal reference protein data; The corrected data is input into the built-in multivariate logistic regression model to calculate the comprehensive risk score. Based on the scoring and preset rules, structured prediction results are generated, including risk level, probability of disease state, and differential diagnosis suggestions. Clinical report generation module: This module converts the structured prediction results into graphic reports that can be directly interpreted by doctors and outputs them in print via a network interface.

5. The protein expression-based myocarditis prediction system according to claim 4, characterized in that, The multiplex protein quantification module includes: The high-throughput protein analysis unit is equipped with prefabricated myocarditis protein detection strips, each strip having specific antibodies against cTnI, H-FABP, SAA, and internal reference proteins ALB and B2M immobilized on it. The signal acquisition device simultaneously quantifies multiple proteins in a batch of samples, and then transmits them to a high-throughput protein analysis unit to obtain multiple chemiluminescent signals. The signal analysis unit converts the collected chemiluminescence signals into relative optical density values ​​for each protein and automatically calculates the ratio of the target protein to the internal reference protein.

6. The protein expression-based myocarditis prediction system according to claim 4, characterized in that, The analysis module includes: Reference database, storing protein expression reference profiles from multicenter clinical studies; The dynamic weight adjustment unit is configured to fine-tune the weight coefficients of each protein in the prediction model based on the clinical information of the subjects. The prognostic trend analysis unit receives multiple test data from the same patient at different time points and plots curves showing the changes in key protein expression levels and comprehensive risk scores over time.

7. The protein expression-based myocarditis prediction system according to claim 4, characterized in that, The multiplex protein quantification module includes a result reliability assessment unit, which is configured as follows: Real-time monitoring of the chemiluminescence signal intensity fluctuation and background noise ratio of each sample; If the signal-to-noise ratio of any target protein is lower than a preset threshold, a retest command is automatically triggered, and the detection device is controlled to retest the sample. Simultaneously, a data quality flag is sent to the analysis module. After receiving the flag, the analysis module automatically reduces the weight of the low-quality protein data when calculating the comprehensive risk score. Furthermore, the dynamic weight adjustment unit of the analysis module and the signal analysis unit of the multiplex protein quantification module establish a bidirectional feedback logic: after completing the detection of a batch of samples, the signal analysis unit sends the overall signal stability index of this detection to the dynamic weight adjustment unit; the dynamic weight adjustment unit judges the quality level of this batch of detection based on the index: if the index shows high stability, the original model weight is maintained. If the indicator shows moderate fluctuations, the weight of the internal reference protein correction factor will be automatically increased. If the indicators show high instability, then the use of this batch of data will be suspended.

8. The protein expression-based myocarditis prediction system according to claim 4, characterized in that, The analysis module includes self-testing logic, which includes... a) Extract recent protein expression data and final clinical diagnosis results for all subjects from the reference database of the analysis module; b) Use these data to iteratively calibrate the chemiluminescence signal-concentration conversion curve used in the multiplex protein quantification module; c) Using the calibrated detection data, retrain the multivariate logistic regression model in the analysis module and update the weight coefficients and risk thresholds of each protein; d) The updated model and transformation curve parameters are synchronously distributed to the detection and analysis modules of each terminal to complete a system-level self-optimization cycle.

9. The system for predicting myocarditis based on protein expression according to claim 6, characterized in that, The prognostic trend analysis unit has an embedded pattern recognition subunit, which is configured as follows: Continuous analysis was conducted on the changes and relative proportions of cTnI, H-FABP, and SAA in multiple tests of the same patient. When a separation pattern of continuously decreasing cTnI and continuously increasing SAA is detected, an alarm is automatically triggered and this signal is pushed to the dynamic weight adjustment unit. In response to this alert, the dynamic weighting adjustment unit temporarily increases the weighting coefficient of SAA in subsequent scoring and generates targeted differential diagnosis suggestions.

Images