A glycolipid metabolism index prediction model for predicting cardiovascular events and a construction method thereof

Abstract

The application discloses a glycolipid metabolism index prediction model for predicting cardiovascular events and a construction method thereof. The model aims to solve the problems that existing cardiovascular risk assessment tools are not accurate in describing glycolipid metabolism state and are single in prediction dimension. The construction method comprises the following steps: obtaining demographic information and clinical data of a subject; measuring the content of high-density lipoprotein subcomponent HDL2b; taking body mass index, lipoprotein a, triglyceride-glucose index and HDL2b as core evaluation parameters, taking whether a cardiovascular event occurs within five years as a target variable, and adopting a multivariate logistic regression to construct a prediction model. The model can realize rapid and accurate stratification of the risk of cardiovascular event occurrence of an individual by quantifying index scores and adding up. The application integrates multi-dimensional key indicators of glycolipid metabolism, has the advantages of high prediction performance, clear risk stratification and strong repeatability, and is suitable for early risk screening and evaluation of cardiovascular diseases.

Description

Technical Field

[0001] This invention relates to a glucose and lipid metabolism index prediction model for predicting cardiovascular events, and also to a method for constructing the glucose and lipid metabolism index prediction model, belonging to the field of healthcare informatics technology. Background Technology

[0002] Cardiovascular disease, as the leading cause of death and disability among chronic non-communicable diseases worldwide, poses a significant challenge to public health systems due to its high incidence and severity. Abnormal glucose and lipid metabolism has been proven to be a core risk factor for the progression of atherosclerosis and cardiovascular events (such as cardiac death, acute myocardial infarction, ischemic stroke, and coronary artery recanalization or reconstruction). Therefore, developing precise cardiovascular event prediction tools based on glucose and lipid metabolism-related indicators has crucial clinical value for early identification of high-risk populations and the development of targeted intervention strategies.

[0003] Current clinically used cardiovascular risk assessment tools (such as ASCVD risk assessment models and Framingham risk scores) have many technical limitations: First, they rely primarily on traditional risk factors such as age, gender, blood pressure, and smoking history, and their assessment of glucose and lipid metabolism status is limited to basic indicators such as total cholesterol and low-density lipoprotein cholesterol. They fail to delve into more specific sub-indicators in the process of glucose and lipid metabolism, resulting in insufficient sensitivity in identifying potential risks caused by early metabolic abnormalities and difficulty in capturing risk signals when traditional indicators have not yet shown obvious abnormalities. Second, although some models attempt to incorporate single blood glucose or blood lipid indicators, they lack a systematic integration of key parameters across multiple dimensions of glucose and lipid metabolism. Specifically, existing technologies mostly focus only on the surface correlation of basic lipid metabolism indicators, failing to fully explore the potential synergistic predictive value between lipid metabolism subcomponents, glucose metabolism alternative indicators, and traditional indicators. They have neither explored combined application modes of glucose and lipid metabolism indicators of different dimensions nor established standardized methods for indicator screening and integration, resulting in incomplete coverage of the mechanisms of cardiovascular events and limited predictive efficacy. Thirdly, existing models have a single predictive dimension and fail to fully consider the complexity of the glucose and lipid metabolism system. They process indicators only through simple superposition or weighting, failing to accurately capture the interaction relationships between indicators. This results in insufficient accuracy in estimating the probability of cardiovascular events, making it difficult to accurately distinguish between low- and medium-risk and high-risk populations in clinical practice, thus affecting the targeting and effectiveness of intervention measures.

[0004] In addition, there are shortcomings in the detection and application of glucose and lipid metabolism indicators in existing technologies: on the one hand, there is no standardized protocol for the detection of lipid metabolism-related sub-indicators, the clinical application value of the relevant indicators has not been fully explored, and they have not been transformed into core assessment parameters for risk prediction models; on the other hand, the construction of existing models often lacks rigorous indicator selection logic, does not clarify the technical basis for the range of indicator values, and lacks stability analysis, resulting in poor clinical applicability and reproducibility of the models, making it difficult to promote and apply them in different medical institutions.

[0005] Therefore, there is an urgent need for a technical solution that can break through the limitations of existing indicator selection and integration, explore key parameters with greater predictive value in the glucose and lipid metabolism system, and achieve the synergistic application of multidimensional indicators through standardized detection methods and rigorous statistical modeling. This solution would address the technical problems of existing models, such as imprecise characterization of glucose and lipid metabolism, single predictive dimension, limited efficacy, and poor clinical applicability, and provide more reliable tool support for the early prevention and control of cardiovascular diseases. Summary of the Invention

[0006] In view of the shortcomings of the existing technology, the primary technical problem to be solved by the present invention is to provide a method for constructing a predictive model of glucose and lipid metabolism indicators for predicting cardiovascular events.

[0007] Another technical problem to be solved by the present invention is to provide a glucose and lipid metabolism index prediction model for predicting cardiovascular events obtained by the above construction method.

[0008] The third technical problem to be solved by the present invention is to provide a data analysis method for predicting cardiovascular events based on the above model.

[0009] To achieve the above objectives, the present invention adopts the following technical solution: According to a first aspect of the present invention, a method for constructing a predictive model of glucose and lipid metabolism indicators for predicting cardiovascular events is provided, comprising the following steps: S1: Obtain the subject's demographic information and clinical data. The demographic information includes any one or more of age, sex, height, weight, and body mass index. The clinical data includes any one or more of triglycerides, total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, lipoprotein a, fasting blood glucose, glycated hemoglobin, and triglyceride-glucose index. S2: Obtain the content of HDL2b subfraction of high-density lipoprotein in the subject; S3: Using body mass index, lipoprotein a, triglyceride-glucose index, and high-density lipoprotein subfraction HDL2b as assessment parameters, and taking whether the subject has experienced cardiovascular events within five years as the target variable, multivariate logistic regression was used for statistical analysis to construct a glucose and lipid metabolism index prediction model for predicting cardiovascular events. The cardiovascular events mentioned include cardiac death, acute myocardial infarction, coronary artery recanalization or reconstruction, and ischemic stroke.

[0010] Preferably, the Body Mass Index (BMI) is calculated using the following formula: BMI = weight / height 2 .

[0011] Preferably, the triglyceride-glucose index is calculated according to the following formula: Triglyceride-glucose index = ln[triglycerides (mg / dL) × glucose (mg / dL) / 2].

[0012] Preferably, the glucose and lipid metabolism index prediction model is a logistic regression binary classifier, and the construction method further includes a model evaluation step S4: assessing the occurrence of cardiovascular events based on the body mass index, high-density lipoprotein subfraction HDL2b, lipoprotein a, and triglyceride-glucose index. Specifically, the score of each variable is calculated using a nomogram based on the specific values ​​of the four variables, the scores of the four variables are added together to obtain the total score, and the probability of cardiovascular events is determined based on the total score.

[0013] Preferably, in step S4, the total score is increased based on the increase in the values ​​of the body mass index, the lipoprotein a, or the triglyceride-glucose index, and the decrease in the value of the high-density lipoprotein subfraction HDL2b.

[0014] Preferably, in step S4, the relationship between the total score and the probability of cardiovascular events is as follows: when the total score is below 128, the probability of occurrence is less than 10%; when the total score is between 128 and 140, the probability of occurrence is 10% to 30%; when the total score is between 140 and 152, the probability of occurrence is 30% to 50%; when the total score is between 152 and 164, the probability of occurrence is 50% to 70%; when the total score is between 164 and 176, the probability of occurrence is 70% to 90%; and when the total score exceeds 176, the probability of occurrence exceeds 90%.

[0015] According to a second aspect of the present invention, a glucose and lipid metabolism index prediction model for predicting cardiovascular events is provided, which is constructed using the above-described method.

[0016] According to a third aspect of the present invention, a data analysis method for predicting cardiovascular events is provided, which uses the above-mentioned glucose and lipid metabolism index prediction model to assess the risk of cardiovascular events in individuals to be assessed, including the following steps: A1: Obtain demographic information and clinical data of the person to be evaluated. The demographic information includes age, sex, height, weight, and body mass index. The clinical data includes triglycerides, total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, lipoprotein a, fasting blood glucose, glycated hemoglobin, and triglyceride-glucose index. A2: Measure the content of HDL2b, a subfraction of high-density lipoprotein, in the individuals to be evaluated; A3: Input the body mass index, high-density lipoprotein subfraction HDL2b, lipoprotein a, and triglyceride-glucose index into the glucose and lipid metabolism index prediction model to obtain the probability of cardiovascular events in the person to be evaluated.

[0017] According to a fourth aspect of the present invention, an assessment system for predicting cardiovascular events is provided, for implementing the above-described data analysis method, the assessment system comprising: The data acquisition module is used to acquire demographic information and glucose and lipid metabolism-related indicators of the individual to be evaluated. These indicators include age, height, weight, body mass index, triglycerides, fasting blood glucose, triglyceride-glucose index, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, and lipoprotein a. The subgroup detection module is used to measure the HDL2b content of the high-density lipoprotein subgroup in the individual being evaluated. The risk assessment module is used to receive the data of body mass index, high-density lipoprotein subfraction HDL2b, lipoprotein a and triglyceride-glucose index, call the above-mentioned glucose and lipid metabolism index prediction model, and output the cardiovascular event probability assessment result.

[0018] Preferably, the data acquisition module interfaces with the hospital information system (HIS), laboratory information system (LIS), and wearable devices to achieve automated data entry.

[0019] Compared with existing technologies, this invention systematically integrates multidimensional key indicators of glucose and lipid metabolism, employs standardized detection techniques and rigorous statistical modeling to construct a predictive model, and combines experimental data verification, demonstrating significant technical advantages and practical value over existing technologies, as detailed below: (1) Predictive efficacy is accurate and efficient, significantly improving risk identification ability: This invention screens out four core assessment parameters through multi-factor logistic regression: body mass index (BMI), high-density lipoprotein subfraction HDL2b, lipoprotein a, and triglyceride-glucose index (TyG index). Validated on a training set of 80 cases, the area under the receiver operating characteristic curve (AUC) of the combined prediction of the four indicators reached 0.919 (P < 0.001), which is significantly higher than the predictive efficacy of a single indicator (AUC of HDL2b = 0.830, AUC of lipoprotein a = 0.836, AUC of BMI = 0.754, AUC of TyG index = 0.791). Further validation on a validation set of 20 cases showed that the AUC of the combined indicators increased to 0.940 (P < 0.001), and the Hosmer-Lemeshow test showed that the model fit was good (P = 0.155), with the calibration curve highly overlapping with the ideal curve. Experimental data show that the accuracy of this model in predicting cardiovascular events is significantly better than existing models that rely on traditional indicators. It can accurately capture risk signals caused by early metabolic abnormalities and effectively solve the problem of insufficient sensitivity in risk identification of existing technologies.

[0020] (2) Scientific and rigorous indicator selection, specifically depicting the fine-grained state of glucose and lipid metabolism: This invention, through a case-control study, confirmed that there were highly significant statistical differences (P < 0.001) between the cardiovascular event group and the control group in four indicators, namely BMI, HDL2b, lipoprotein a, and TyG index, while there were no significant differences (P > 0.05) in traditional glucose and lipid metabolism indicators (such as total cholesterol, high-density lipoprotein cholesterol, and fasting blood glucose). Furthermore, HDL2b, as a functionally specific subcomponent of high-density lipoprotein, had a significantly lower relative content (20.72 ± 7.04%) in the cardiovascular event group than in the control group (29.09 ± 7.15%), making it a specific indicator associated with cardiovascular events. This invention specifically incorporates the above four core indicators, achieving a precise depiction of the fine-grained regulatory state of glucose and lipid metabolism, and solving the technical deficiency of existing models that rely solely on basic indicators and cannot reflect the complexity of the metabolic system.

[0021] (3) Clear and explicit risk stratification, with strong guidance for clinical application: This invention establishes standardized scoring calculation and risk correspondence rules. By quantifying the contribution of indicators (adding 0.84 points for every 1 unit increase in BMI, 0.58 points for every 1 unit increase in lipoprotein a, 27.8 points for every 1 unit increase in TyG index, and subtracting 1.1 points for every 1 unit increase in HDL2b), the total score is accurately correlated with the probability of cardiovascular events. Experimental verification shows that when the total score is below 128 points, the probability of occurrence is <10%; 128-140 points correspond to a probability of 10%-30%; 140-152 points correspond to a probability of 30%-50%; 152-164 points correspond to a probability of 50%-70%; 164-176 points correspond to a probability of 70%-90%; and when the score exceeds 176 points, the probability is >90%. In the 20-case validation set, the model predicted a probability of over 60% for those who experienced cardiovascular events, while the predicted probability for those who did not experience events was mostly below 30%, demonstrating high accuracy in risk stratification. This design provides clinicians with an intuitive and operational risk assessment standard, facilitating the rapid differentiation of high, medium, and low-risk populations and providing a clear basis for developing targeted intervention strategies.

[0022] (4) Standardized detection and modeling schemes with strong repeatability and scalability: This invention clarifies the standardized detection method for HDL2b (microfluidic chip electrophoresis), and specifies in detail the sample collection, processing, electrophoresis conditions, and result calculation logic to ensure consistency of detection results across different laboratories. It also clarifies the statistical methods for model construction (multifactor logistic regression), the basis for sample size (80 training cases + 20 validation cases), and the range of index values ​​(BMI 15–30, lipoprotein a 80–240, TyG index 7.0–10.5, HDL2b 10–50). Repeated experiments have verified that the model prediction results are stable (the 95% confidence interval of AUC after 1000 Bootstrap samplings is [0.886, 0.952]). This standardized design solves the problems of inconsistent detection methods and poor model repeatability in existing technologies, facilitating its application in medical institutions at all levels.

[0023] (5) Wide range of applicable scenarios and significant social and clinical value: The assessment system provided in this embodiment of the invention can be connected to hospital information systems (HIS), laboratory information systems (LIS), and wearable devices to achieve automated data entry. It is applicable to various scenarios such as screening of the general population without a clear history of cardiovascular disease and follow-up of high-risk groups. By identifying high-risk individuals early (such as those with a total score of more than 176 points and a predicted probability of >90%), it can guide early clinical intervention, reduce the incidence and mortality of cardiovascular events, and alleviate the burden on public health. It has important clinical practice significance and social value. Attached Figure Description

[0024] Figure 1A flowchart illustrating the method for constructing a glucose and lipid metabolism index prediction model according to an embodiment of the present invention; Figure 2 This is a comparison chart of body mass index between the cardiovascular event group and the control group in an embodiment of the present invention; Figure 3 This is a comparison chart of triglyceride-glucose index between the cardiovascular event group and the control group in an embodiment of the present invention; Figure 4 This is a comparison chart of high-density lipoprotein subfraction HDL2b between the cardiovascular event group and the control group in an embodiment of the present invention; Figure 5 This is a comparison chart of lipoprotein a between the cardiovascular event group and the control group in an embodiment of the present invention; Figure 6A shows the ROC curves of individual core indicators (HDL2b, lipoprotein a, BMI, TyG index); Figure 6B shows the ROC curves predicted by the combined analysis of four indicators: body mass index, HDL2b, lipoprotein a, and TyG index. Figure 7 This is a nomogram model of a glucose and lipid metabolism index prediction model for predicting cardiovascular events, as described in this embodiment of the invention. Figure 8 This is the calibration curve of the nomogram model of the glucose and lipid metabolism index prediction model for predicting cardiovascular events in this embodiment of the invention. Figure 9 In this embodiment of the invention, the receiver operating characteristic curve of a glucose and lipid metabolism index prediction model for predicting cardiovascular events using a validation set is shown. Detailed Implementation

[0025] The embodiments of the technical solution of the present invention will now be described in detail with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and are therefore merely examples and should not be used to limit the scope of protection of the present invention. It should be noted that, unless otherwise stated, the technical or scientific terms used in this invention should have the ordinary meaning understood by those skilled in the art.

[0026] Example 1: Method for constructing a predictive model for glucose and lipid metabolism indicators To establish a predictive model for glucose and lipid metabolism indicators in predicting cardiovascular events, it is necessary to first identify relevant glucose and lipid metabolism indicators. In this embodiment, subjects aged 80 and above who visited the Second Medical Center of the PLA General Hospital between October and December 2019 were consecutively included. Subjects with malignant tumors, severe hematological diseases, severe liver and kidney dysfunction, or those who withdrew from follow-up midway were excluded. A total of 100 subjects were included and randomly divided into a training set (80 subjects) and a validation set (20 subjects) at an 8:2 ratio.

[0027] The flowchart of the construction of the glucose and lipid metabolism index prediction model provided in this embodiment of the invention is as follows: Figure 1 As shown in the figure, this diagram visually illustrates the complete construction process of the glucose and lipid metabolism indicator prediction model, which includes four core steps: "obtaining demographic information and clinical data of subjects," "obtaining the content of HDL2b subfraction of high-density lipoprotein in subjects," "multivariate logistic regression statistical analysis," and "model construction and evaluation." The diagram clarifies the connection between each technical step, providing a clear process guide for those skilled in the art to repeat this invention. The specific steps are as follows: S1: Obtain the subject's demographic and clinical information.

[0028] The demographic information includes any one or more of the following: age, sex, height, weight, and body mass index; the clinical data includes any one or more of the following: triglycerides, total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, lipoprotein a, fasting blood glucose, glycated hemoglobin, and triglyceride-glucose index. Cardiovascular events will be followed up for five years from the date of information collection. These cardiovascular events include any one or more of the following: cardiac death, acute myocardial infarction, coronary artery recanalization or revascularization (PTCA, PCI, CABG), and ischemic stroke.

[0029] The Body Mass Index is calculated using the following formula: The triglyceride-glucose index is calculated using the following formula: S2: Obtain the content of high-density lipoprotein subfractions in the subjects.

[0030] In this embodiment, the selected high-density lipoprotein subfractions specifically include: HDL2b: High-density lipoprotein 2b HDL3: High-density lipoprotein 3 In determining high-density lipoprotein (HDL) subfraction markers, blood samples from subjects were analyzed for HDL subfractions to screen for differences in HDL subfraction markers between those who experienced cardiovascular events and the control group. A high-density lipoprotein (HDL) subfraction typing kit (microfluidic chip method) (Guangdong Guosheng Medical Technology Co., Ltd.) was used.

[0031] The detection instrument uses a fully automatic capillary electrophoresis instrument (model: GS-Micep CE; registration number: Yue Sui Medical Equipment Preparation 20240290; manufacturer: Guangdong Guosheng Medical Technology Co., Ltd.). The fully automatic capillary electrophoresis instrument separates components based on the phenomenon that charged particles migrate directionally at different speeds under the action of a high-voltage electric field in an electrolyte solution. After the separation of each component, it is detected by a detector, and qualitative and quantitative analysis is carried out based on the migration time and response value of each component.

[0032] The kit contains a lipophilic fluorescent dye that specifically binds to lipoprotein (HDL) in the sample; a specific polymer gel medium is poured into the microfluidic chip. Under the action of an external electric field, the electroosmotic flow transports the sample into the separation channel. HDL in the test sample can be separated into different subcomponents under the action of molecular sieve and charge. By detecting the fluorescent dye bound to the lipoprotein, an electrophoretic trace of the fluorescence signal intensity ratio to migration time of each sample is formed. The data analysis software analyzes the electrophoresis pattern and calculates the percentage content of the peak areas of HDL-2b and HDL-3 subcomponents in the total HDL peak area.

[0033] The test procedure is as follows: Instrument preparation: When the instrument is running, the indoor temperature is required to be in the range of 18 - 26 °C and the humidity is less than 70%. If it is not within the required range, it will affect the accuracy of the test results. Turn on the instrument power supply. Open the instrument operation software and click "Pipeline Pump Pressure Flushing" on the main interface to drain the possible condensed water or impurities in the internal pipeline of the equipment. After "Pipeline Pump Pressure Flushing" is completed, the experiment can be carried out.

[0034] 2) Preparation of electrophoresis gel: The chip and related reagents must be pre-equilibrated to room temperature (placed at room temperature for about 20 - 30 min) before use. Before use, briefly centrifuge the fully rewarmed electrophoresis gel tube for 10 s. Transfer 500 μL of electrophoresis gel to a centrifugal filter column and centrifuge at 3700 rpm for 15 min at room temperature. The electrophoresis gel is filtered to the bottom of the tube. Discard the filter column and let the electrophoresis gel stand at room temperature for standby.

[0035] 3) Preparation of quality control product: Vortex the quality control product for 10 s and briefly centrifuge for 5 s before use. Take 10 μL for each test.

[0036] 4) Sample Pretreatment: Add 5 μL of sample dye and 10 μL of Maker dye to 1 mL of sample buffer to prepare a sample dilution (mixing ratio 1:2:200). Vortex for 10 s to mix, then centrifuge briefly for 5 s. Mix 98 μL of sample dilution with 2 μL of serum / plasma sample. Mix 90 μL of sample dilution with 10 μL of quality control A and 10 μL of quality control B. Vortex all samples for 10 s to mix, then centrifuge briefly for 5 s and incubate in the dark for 10 min. After 10 min, transfer the mixed samples to the corresponding wells on the sample plate.

[0037] 5) Ladder and Buffer Tube Preparation: Add 95 μL of the prepared quality control A to the ladder tube provided in the kit. Add 1.2 mL of sample buffer to the buffer tube provided in the kit. Click "Unload Sample Plate" on the main software interface to pop up the sample plate holder. Place the ladder tube, buffer tube, and the sample plate with added samples into the instrument's sample plate holder. After placing them, click "Load Sample Plate" on the main software interface, and the sample plate will retract into the instrument compartment.

[0038] 6) Microfluidic Chip Preparation: Let the chip storage container stand on the table for 30 minutes to allow it to reach room temperature. Connect the pipette tip to the vacuum pump and use the tip to remove all liquid from the chip wells. Rinse wells (1, 3, 4, 7, 8, and 10) twice with ultrapure water and remove the water. Add the corresponding amount of electrophoresis gel to wells 3, 7, 8, and 10 using reverse pipetting. Add the corresponding amount of ultrapure water to well 4. Before placing the chip into the instrument, ensure that well 1 (waste liquid well) is empty. Click "Unload Chip" on the main software interface to open the instrument compartment door. Remove the chip from the chip storage container and check if the circular detection window is clean. If the detection window needs cleaning, wipe it with lint-free paper dampened with 70% isopropanol. Place the chip into the chip slot of the instrument. Click the "Load Chip" button on the main software interface and close the instrument compartment door.

[0039] 7) Run the experiment: On the main interface, click the "Run" button. Select the experiment type as "HT HDL / LDL Subfraction Assay", and select the experimental plate and sample wells.

[0040] 8) Results Analysis: The percentage area of ​​the HDL-2b and HDL-3 curves was used as the estimate of the HDL-2b and HDL-3 percentages using the peak fitting algorithm. The specific formula is as follows.

[0041] This step allows us to obtain the relative levels of the high-density lipoprotein subfractions HDL2b and HDL3 in the subjects.

[0042] In specific implementations, the execution order of steps S1 to S2 is not important.

[0043] S3: Using the above-mentioned demographic information, clinical data, and high-density lipoprotein subfractions of the subjects in the training set as evaluation parameters, and taking the occurrence of cardiovascular events as the target variable, multivariate logistic regression was used for statistical analysis to construct a glucose and lipid metabolism index prediction model for predicting cardiovascular events.

[0044] In this specific implementation, SPSS 27.0 and R (version 4.3.3) software were used for statistical analysis. In this analysis, a cardiovascular event group and a control group were defined. Cardiovascular events were defined as cardiac death, acute myocardial infarction, coronary artery recanalization or revascularization (PTCA, PCI, CABG), and ischemic stroke. Individuals experiencing any of these cardiovascular events during the five-year follow-up period were included in the cardiovascular event group; otherwise, they were in the control group.

[0045] For multivariate logistic regression, covariates are used to adjust for confounding factors. The Shapiro-Wilk test is used to determine the normality of continuous variables. Normally distributed variables are expressed as mean ± standard deviation, and independent samples t-tests are used for comparisons between groups. Non-normally distributed data are expressed as median (interquartile range), and the Mann-Whitney U test is used for comparisons between groups of non-normally distributed data. Chi-square tests are used to compare categorical variables between groups, and P < 0.05 is considered statistically significant. Indicators with P < 0.05 in univariate analysis are included in multivariate logistic regression, and indicators with P < 0.05 in multivariate regression are included in the glucose and lipid metabolism indicator prediction model. Receiver operating characteristic curves are used to determine the diagnostic power of positive indicators for endpoint events. A glucose and lipid metabolism indicator prediction model for cardiovascular events is constructed using nomograms.

[0046] Statistical analysis, as shown in Table 1, revealed no statistical differences in demographic information (age, gender) and some clinical laboratory indicators (triglycerides, total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, fasting blood glucose, and glycated hemoglobin) between the cardiovascular event group and the control group.

[0047] High-density lipoprotein cholesterol (HDL-C) participates in reverse cholesterol transport, transferring cholesterol to the liver for metabolism and lowering blood cholesterol levels, thus playing a role in preventing atherosclerosis (AS). It also inhibits oxidative stress and inflammation, protects vascular endothelial cells, and is closely related to cardiovascular events. Therefore, we selected HDL subfractions to construct a glucose and lipid metabolism predictive model for cardiovascular events. We analyzed the differences in HDL subfractions in the serum of two groups of subjects using a HDL subfraction typing kit (microfluidic chip method) (Guangdong Guosheng Medical Technology Co., Ltd.). We observed that in individuals who experienced cardiovascular events, the relative content of HDL2b was significantly lower than that of the control group (P < 0.001), while the relative content of HDL3 did not change significantly. This suggests that HDL2b may be a specific biomarker for cardiovascular events.

[0048] The indicators that did not show significant differences were not included in the glucose and lipid metabolism predictive model for cardiovascular events. Body mass index, triglyceride-glucose index, high-density lipoprotein subfraction HDL2b, and lipoprotein a showed statistically significant differences (P < 0.001). Figures 2-5 As shown, the glucose and lipid metabolism indicators were incorporated into the predictive model for cardiovascular events. Figure 2 shows a comparison of body mass index (BMI) between the cardiovascular event group and the control group. Data shows that the mean BMI of the cardiovascular event group was 24.64 ± 2.77 kg / m². 2 The control group had a concentration of 21.78 ± 3.16 kg / m³. 2 The difference between the two groups was statistically significant (P < 0.001), which directly demonstrates that BMI is a key risk indicator associated with the occurrence of cardiovascular events, providing experimental evidence for its inclusion in the prediction model.

[0049] Figure 3 shows a comparison of the triglyceride-glucose index (TyG index) between the cardiovascular event group and the control group. The results showed that the mean TyG index in the cardiovascular event group was 9.07±0.56, while that in the control group was 8.48±0.43, with a highly significant difference between the two groups (P<0.001). This confirms that the TyG index, as a surrogate indicator of insulin resistance, can effectively distinguish between high-risk and general populations and is one of the core evaluation parameters of the model.

[0050] Figure 4 compares the HDL2b subfraction of high-density lipoprotein between the cardiovascular event group and the control group. Data shows that the relative HDL2b level in the cardiovascular event group was 20.72±7.04%, significantly lower than the 29.09±7.15% in the control group (P<0.001). This directly confirms the association between decreased HDL2b levels and the occurrence of cardiovascular events, demonstrating its technical value as a specific biomarker and providing direct experimental support for model screening of this indicator.

[0051] Figure 5 The figure shows a comparison of lipoprotein a levels between the cardiovascular event group and the control group. The results showed that the mean lipoprotein a level in the cardiovascular event group was 194.97 ± 22.74 mg / dL, while it was 159.98 ± 26.64 mg / dL in the control group. The difference between the two groups was highly significant (P < 0.001), clearly demonstrating the role of lipoprotein a as an independent cardiovascular risk factor and validating the scientific validity of its inclusion in the model. See also Table 1.

[0052] Table 1 In one embodiment of the present invention, the glucose and lipid metabolism index prediction model for predicting cardiovascular events can be a logistic regression binary classifier, wherein the evaluation parameters include body mass index, high-density lipoprotein subfraction 2b, lipoprotein a, and triglyceride-glucose index, and the evaluation result is the risk of cardiovascular events. Figure 6A and Figure 6B Figure 6A shows the receiver operating characteristic (ROC) curves of the prediction model. The horizontal axis represents "1-specificity," and the vertical axis represents "sensitivity." The area under the curve (AUC) for HDL2b is 0.830, for lipoprotein a is 0.836, for BMI is 0.754, and for TyG index is 0.791 (all P < 0.05), demonstrating that each individual indicator has some predictive value for cardiovascular event risk, but its efficacy is limited. Figure 6B The ROC curve for the joint prediction of the four indicators shows an AUC of 0.919 (P < 0.001), significantly higher than any single indicator. This clearly demonstrates the technical advantage of the "multi-dimensional indicator synergistic integration" of this invention, proving that the prediction accuracy of this glucose and lipid metabolism indicator prediction model is significantly better than that of a single indicator evaluation. According to the ROC curve analysis, the optimal cut-off value is 0.20. At this value, the sensitivity is 0.92, indicating a high ability to identify true positives and a low false negative rate; the specificity is 0.80, indicating a good ability to identify true negatives and a well-controlled false positive rate; and the Youden index is 0.72, indicating that the model's overall discriminative power is superior at this cut-off value.

[0053] Using R software, a nomogram model was constructed based on body mass index (BMI), lipoprotein a (PLA-A), triglyceride-glucose index (TG-GLA), and high-density lipoprotein subfraction HDL2b (HDL2b) to assess the risk of cardiovascular events. A score was calculated for each of the four variables based on their values. These scores were then summed to obtain a total score, which corresponds to the probability of the occurrence of the clinical endpoint event. Specifically, within a specific range... Figure 7 As shown. Figure 7 This is a nomogram of the predictive model. This chart is a visual risk calculation tool, with horizontal axes representing the numerical values ​​of four indicators: BMI, lipoprotein a, TyG index, and HDL2b; a score axis in the middle; and a cardiovascular event probability axis on the right. The usage is as follows: (1) Find the actual BMI value of the person to be assessed, draw a vertical line upward on the corresponding value axis, and the intersection with the "score axis" is the individual score of BMI; (2) Obtain the individual scores of lipoprotein a, TyG index, and HDL2b in the same manner; (3) Sum the individual scores of the four indicators to obtain the total score; (4) Based on the total score, draw a vertical line downward on the "score axis". The intersection of the line with the "probability axis" on the right is the probability of cardiovascular events occurring in the person to be assessed within five years, thus achieving rapid quantitative assessment of risk.

[0054] Specifically, the contribution of each evaluation parameter to the total score is quantified using the nomogram model described above. In a preferred embodiment, the Body Mass Index (BMI) ranges from 15 to 30, with each unit increase boosting the score by 0.84 points; the Lipoprotein A (LA) ranges from 80 to 240 mg / dL, with each unit increase boosting the score by 0.58 points; the Triglyceride-Glucose Index (TG-GGI) ranges from 7.0 to 10.5, with each unit increase boosting the score by 27.8 points; and the High-Density Lipoprotein Subfraction (HDL2b) ranges from 10% to 50%, with each unit increase decreasing the score by 1.1 points. It should be understood that the specific values ​​and scoring coefficients described above are based on optimization results from a specific sample set, and those skilled in the art can adaptively adjust these coefficients according to different training data.

[0055] On the other hand, when the total score is below 128, the probability of cardiovascular events is less than 10%; between 128 and 140, the probability is 10%–30%; between 140 and 152, the probability is 30%–50%; between 152 and 164, the probability is 50%–70%; between 164 and 176, the probability is 70%–90%; and when the total score exceeds 176, the probability of cardiovascular events exceeds 90%. Through glucose and lipid metabolism indicators, we can predict the probability of cardiovascular events.

[0056] like Figure 8 As shown, the Hosmer-Lemeshow test verified that the model had a high good fit (P > 0.05). The graph shows "predicted probability" on the horizontal axis and "true probability" on the vertical axis, containing three curves: the ideal curve (diagonal, representing perfect consistency between predicted and true probabilities), the actual curve (representing the model's prediction result), and the calibration curve (representing the calibrated model's prediction result). The results show a high degree of overlap between the calibration curve and the ideal curve, and the Hosmer-Lemeshow test P = 0.155 (P > 0.05), demonstrating that the glucose and lipid metabolism index prediction model constructed in this invention has a good good fit, high consistency between the predicted probability and the actual clinical probability, and strong technical reliability.

[0057] Another 20 subjects were selected for the validation set (details as described above), and the criteria described above were used to determine whether the subjects had cardiovascular events. The glucose and lipid metabolism index prediction model constructed according to this invention requires the measurement of body mass index, lipoprotein a, triglyceride-glucose index, and high-density lipoprotein subfraction HDL2b data. For example... Figure 9 As shown in the figure, this is the ROC curve of the joint prediction of the four indicators for 20 validation set samples, with an area under the curve (AUC) of 0.980 (P < 0.001). This result is a further improvement compared to the training set AUC of 0.919, confirming that the glucose and lipid metabolism indicator prediction model constructed in this invention still has excellent predictive efficacy, strong stability and generalization ability in independent external samples, providing key data support for clinical application. Specific data were then fed into the constructed glucose and lipid metabolism indicator prediction model to predict the incidence of cardiovascular events in these subjects; the specific results are shown in Table 2.

[0058] According to the validation set results, at the optimal cutoff value (Cut-off value = 0.20), the positive predictive value of this glucose and lipid metabolism index prediction model is 88.9%, and the negative predictive value is 90.9%, indicating that the glucose and lipid metabolism index prediction model has good predictive efficacy.

[0059] Table 2 In some embodiments, the present invention also provides a method for predicting cardiovascular events based on glucose and lipid metabolism indicators, comprising at least the following steps: Step 1. Obtain demographic information and clinical data of the person to be evaluated.

[0060] For specific instructions, please refer to the above text.

[0061] Step 2. Measure the high-density lipoprotein subfraction 2b of the person to be evaluated.

[0062] For specific instructions, please refer to the above text.

[0063] In a specific implementation, the execution order of steps 1 and 2 is not important.

[0064] Step 3. Input the body mass index, lipoprotein a, triglyceride-glucose index, and high-density lipoprotein subfraction HDL2b into the glucose and lipid metabolism index prediction model for predicting cardiovascular events, and assess the cardiovascular event occurrence of the person to be assessed.

[0065] By inputting the body mass index, lipoprotein a, triglyceride-glucose index, and high-density lipoprotein subfraction HDL2b of the individuals to be assessed into a glucose and lipid metabolism index prediction model for cardiovascular events, the risk probability of cardiovascular events for the individuals to be assessed can be obtained.

[0066] It should be noted that the glucose and lipid metabolism index prediction model provided in this embodiment of the invention is only used for risk stratification, does not constitute a basis for clinical diagnosis, and cannot replace the professional judgment of doctors.

[0067] The above provides a detailed description of the glucose and lipid metabolism index prediction model and construction method for predicting cardiovascular events provided by this invention. Any obvious modifications made by those skilled in the art without departing from the essence of this invention will constitute an infringement of the patent rights of this invention and will incur corresponding legal liability.

Claims

1. A method for constructing a predictive model of glucose and lipid metabolism indicators for predicting cardiovascular events, characterized in that... Includes the following steps: S1: Obtain the subject's demographic information and clinical data. The demographic information includes any one or more of age, sex, height, weight, and body mass index. The clinical data includes any one or more of triglycerides, total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, lipoprotein a, fasting blood glucose, glycated hemoglobin, and triglyceride-glucose index. S2: Obtain the content of HDL2b subfraction of high-density lipoprotein in the subject; S3: Using body mass index, lipoprotein a, triglyceride-glucose index, and high-density lipoprotein subfraction HDL2b as assessment parameters, and taking whether the subject has experienced cardiovascular events within five years as the target variable, multivariate logistic regression was used for statistical analysis to construct a glucose and lipid metabolism index prediction model for predicting cardiovascular events. The cardiovascular events mentioned include any one or more of the following: cardiac death, acute myocardial infarction, coronary artery recanalization or reconstruction, and ischemic stroke.

2. The construction method as described in claim 1, characterized in that... The Body Mass Index (BMI) is calculated using the following formula: BMI = weight / height 2 .

3. The construction method as described in claim 1, characterized in that... The triglyceride-glucose index is calculated using the following formula: Triglyceride-glucose index = ln[triglycerides (mg / dL) × glucose (mg / dL) / 2].

4. The construction method as described in claim 1, characterized in that: The glucose and lipid metabolism index prediction model is a logistic regression binary classifier. The construction method also includes a model evaluation step S4: assessing the occurrence of cardiovascular events based on the body mass index, high-density lipoprotein subfraction HDL2b, lipoprotein a, and triglyceride-glucose index. Specifically, the score of each variable is calculated using a nomogram based on the specific values ​​of the four variables, and the scores of the four variables are added together to obtain the total score. The probability of cardiovascular events is determined based on the total score.

5. The construction method as described in claim 4, characterized in that... In step S4, the total score is increased based on the increase in the body mass index, the lipoprotein a or the triglyceride-glucose index, and the decrease in the high-density lipoprotein subfraction HDL2b.

6. The construction method as described in claim 4, characterized in that... In step S4, the relationship between the total score and the probability of cardiovascular events is as follows: when the total score is below 128, the probability of occurrence is less than 10%; when the total score is between 128 and 140, the probability of occurrence is 10% to 30%; when the total score is between 140 and 152, the probability of occurrence is 30% to 50%; when the total score is between 152 and 164, the probability of occurrence is 50% to 70%; when the total score is between 164 and 176, the probability of occurrence is 70% to 90%; and when the total score exceeds 176, the probability of occurrence exceeds 90%.

7. A predictive model for glucose and lipid metabolism indicators to predict cardiovascular events, characterized in that... It is constructed using the construction method described in any one of claims 1 to 6.

8. A data analysis method for predicting cardiovascular events, characterized in that... Using the glucose and lipid metabolism index prediction model of claim 7 to assess the risk of cardiovascular events in the individuals to be assessed includes the following steps: A1: Obtain demographic information and clinical data of the person to be evaluated. The demographic information includes any one or more of age, sex, height, weight, and body mass index. The clinical data includes any one or more of triglycerides, total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, lipoprotein a, fasting blood glucose, glycated hemoglobin, and triglyceride-glucose index. A2: Measure the content of HDL2b, a subfraction of high-density lipoprotein, in the individuals to be evaluated; A3: Input the body mass index, high-density lipoprotein subfraction HDL2b, lipoprotein a, and triglyceride-glucose index into the glucose and lipid metabolism index prediction model to obtain the probability of cardiovascular events in the person to be evaluated.

9. An assessment system for predicting cardiovascular events, used to implement the data analysis method of claim 8, characterized in that... include: The data acquisition module is used to acquire demographic information and glucose and lipid metabolism-related indicators of the individual to be evaluated. The indicators include any one or more of the following: age, height, weight, body mass index, triglycerides, fasting blood glucose, triglyceride-glucose index, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, and lipoprotein a. The subgroup detection module is used to measure the HDL2b content of the high-density lipoprotein subgroup in the individual being evaluated. The risk assessment module is used to receive data on body mass index, high-density lipoprotein subfraction HDL2b, lipoprotein a, and triglyceride-glucose index, call the glucose and lipid metabolism index prediction model described in claim 7, and output the cardiovascular event probability assessment results.

10. The evaluation system as described in claim 9, characterized in that: The data acquisition module interfaces with the hospital information system (HIS), laboratory information system (LIS), and wearable devices to achieve automated data entry.

Images