Background: Cardiovascular disease (CVD) is the leading cause of mortality worldwide. Accurately identifying subjects at high-risk of CVD may improve CVD outcomes. We sought to systematically examine the feasibility and performance of 7 widely used machine learning (ML) algorithms in predicting CVD risks. Methods: The final analysis included 1508 Kazakh subjects in China without CVD at baseline who completed follow-up. All subjects were randomly divided into the training set (80%) and the test set (20%). L1-penalized logistic regression (LR), support vector machine with radial basis function (SVM), decision tree (DT), random forest (RF), k-nearest neighbors (KNN), Gaussian naive Bayes (NB), and extreme gradient boosting (XGB) were employed for prediction CVD outcomes. Ten-fold cross-validation was used during model developing and hyperparameters tuning in the training set. Model performance was evaluated in the test set in light of discrimination, calibration, and clinical usefulness. RF was applied to obtain the variable importance of included variables. Twenty-two variables, including sociodemographic characteristics, medical history, cytokines, and synthetic indices, were used for model development. Results: Among 1508 subjects, 203 were diagnosed with CVD over a median follow-up of 5.17 years. All 7 models had moderate to excellent discrimination (AUC ranged from 0.770 to 0.872) and were well calibrated. LR and SVM performed identically with an AUC of 0.872 (95% CI: 0.829– 0.907) and 0.868 (95% CI: 0.825– 0.904), respectively. LR had the lowest Brier score (0.078) and the highest sensitivity (97.1%). Decision curve analysis indicated that SVM was slightly better than LR. The inflammatory cytokines, such as hs-CRP and IL-6, were identified as strong predictors of CVD. Conclusion: SVM and LR can be applied to guide clinical decision-making in the Kazakh Chinese population, and further study is required to ensure their accuracies.

Background: Cardiovascular diseases (CVD) are currently the leading cause of premature death worldwide. Model-based early detection of high-risk populations for CVD is the key to CVD prevention. Thus, this research aimed to use machine learning (ML) algorithms to establish a CVD prediction model based on routine physical examination indicators suitable for the Xinjiang rural population.Method: The research cohort data collection was divided into two stages. The first stage involved a baseline survey from 2010 to 2012, with follow-up ending in December 2017. The second-phase baseline survey was conducted from September to December 2016, and follow-up ended in August 2021. A total of 12,692 participants (10,407 Uyghur and 2,285 Kazak) were included in the study. Screening predictors and establishing variable subsets were based on least absolute shrinkage and selection operator (Lasso) regression, logistic regression forward partial likelihood estimation (FLR), random forest (RF) feature importance, and RF variable importance. The selected subset of variables was compared with L1 regularized logistic regression (L1-LR), RF, support vector machine (SVM), and AdaBoost algorithm to establish a CVD prediction model suitable for this population. The incidence of CVD in this population was then analyzed.Result: After 4.94 years of follow-up, a total of 1,176 people were diagnosed with CVD (cumulative incidence: 9.27%). In the comparison of discrimination and calibration, the prediction performance of the subset of variables selected based on FLR was better than that of other models. Combining the results of discrimination, calibration, and clinical validity, the prediction model based on L1-LR had the best prediction performance. Age, systolic blood pressure, low-density lipoprotein-L/high-density lipoproteins-C, triglyceride blood glucose index, body mass index, and body adiposity index were all important predictors of the onset of CVD in the Xinjiang rural population.Conclusion: In the Xinjiang rural population, the prediction model based on L1-LR had the best prediction performance.

Background: To externally validate the Prediction for ASCVD Risk in China (PAR) risk equation for predicting the 5-year atherosclerotic cardiovascular disease (ASCVD) risk in the Uyghur and Kazakh populations from rural areas in northwestern China and compare its performance with those of the pooled cohort equations (PCE) and Framingham risk score (FRS). Methods: The final analysis included 3347 subjects aged 40–74 years without CVD at baseline. The 5-year ASCVD risk was calculated using the PAR, PCE, and FRS. Discrimination, calibration, and clinical usefulness of the three equations in predicting the 5-year ASCVD risk were assessed before and after recalibration. Results: Of 3347 included subjects, 1839 were female. We observed 286 ASCVD events in within 5-year follow-up. All three risk equations had moderate discrimination in both men and women. C-indices of PAR, PCE, and FRS were 0.727 (95% CI, 0.725–0.729), 0.727 (95% CI, 0.725–0.729), and 0.740 (95% CI, 0.738–0.742), respectively, in men; the corresponding C-indices were 0.738 (95% CI, 0.737–0.739), 0.731 (95% CI, 0.730–0.732), and 0.761 (95% CI, 0.760–0.762), respectively, in women. PCE, PAR and FRS substantially underestimated the 5-year ASCVD risk in women by 70, 23 and 51%, respectively. However, PAR and FRS fairly predicted the risk in men and PAR was well calibrated. The calibrations of the three risk equations could be changed by recalibration. The decision curve analyses demonstrated that at the threshold risk of 5%, PCE was the most clinically useful in both men and women after recalibration. Conclusions: All three risk equations underestimated the 5-year ASCVD risk in women, while PAR and FRS fairly predicted that in men. However, the results of predictive performances for three risk equations are inconsistent, more accurate risk equations are required in the primary prevention of ASCVD aiming to this Uyghur and Kazakh populations.

Objective: To compare the screening ability of various anthropometric and atherogenic indices for Metabolic syndrome (MetS) using three common criteria and to evaluate the validity of suitable parameters in combination for the screening of MetS among a Kazakh population in Xinjiang. Methods: A total of 3752 individuals were selected using the stratified cluster random sampling method from nomadic Kazakhs (≥18 years old) in Xinyuan county, Xinjiang, China, which is approximately 4407 km away from the capital Beijing. MetS was defined by the International Diabetes Federation (IDF), National Cholesterol Education Program Adult Treatment Panel III (ATP III) and Joint Interim Statement (JIS) criteria. The receiver operating characteristic curve (ROC) was used to compare the area under the ROC curve (AUC) of each index. The sensitivity, specificity, Youden’s index and cut-offs of each index for the screening of MetS were calculated. Results: According to the IDF, ATP III and JIS criteria, 18.61%, 10.51%, and 24.83% of males and 23.25%, 14.88%, and 25.33% of females had MetS. According to the IDF criteria, the waist-to-height ratio (WHtR) was the index that most accurately identified individuals with and without MetS both in males (AUC = 0.872) and females (AUC = 0.804), with the optimal cut-offs of 0.53 and 0.52, respectively. According to both the ATP III and JIS criteria, the lipid accumulation product (LAP) was the best index to discriminate between individuals with and without MetS in males (AUC = 0.856 and 0.816, respectively) and females (AUC = 0.832 and 0.788, respectively), with optimal cut-offs of 41.21 and 34.76 in males and 28.16 and 26.49 in females, respectively. On the basis of the IDF standard, Youden’s indices of WHtR and LAP serial tests for the screening of MetS were 0.590 and 0.455 in males and females, respectively, and those of WHtR and LAP parallel tests were 0.608 and 0.479, accordingly. Conclusion: According to the IDF, ATP III and JIS criteria, both the WHtR and LAP were better indices for the screening of MetS. The WHtR and LAP parallel test was the most accurate.

Background: This study aimed to assess the association of baseline insulin resistance (IR) surrogates and their longitudinal trajectories with cardiovascular diseases (CVD) to provide a useful reference for preventing CVD.Methods: This study was a prospective cohort study conducted in the 51st Regiment of the Third Division of Xinjiang Corps. A total of 6362 participants were recruited in 2016 to conduct the baseline survey, and the follow-up surveys in 2019, 2020, 2021, and 2022. The Kaplan–Meier method was used to estimate the cumulative incidence of CVD according to the baseline IR surrogates of metabolic insulin resistance score (METS-IR) and triglyceride-glucose (TyG) index. Cox regression models were used to assess the association between the baseline IR surrogates and CVD. The impact of the longitudinal trajectories of the IR surrogates on CVD was analyzed after excluding those with IR surrogate data measured ≤2 times. Based on the group-based trajectory model (GBTM), the trajectory patterns of IR surrogates were determined. The Kaplan-Meier method was used to estimate the cumulative incidence of CVD in each trajectory group of METS-IR and TyG index. Cox regression models were used to analyze the association between different trajectory groups of each index and CVD. In addition, the Framingham model was utilized to evaluate whether the addition of the baseline IR surrogates increased the predictive potential of the model.Results: Baseline data analysis included 4712 participants. During a median follow-up of 5.66 years, 572 CVD events were recorded (mean age, 39.42 ± 13.67 years; males, 42.9%). The cumulative CVD incidence increased with the ascending baseline METS-IR and TyG index quartiles (Q1–Q4). The hazard ratio and 95% confidence interval for CVD risk in Q4 of the METS-IR and TyG index were 1.79 (1.25, 2.58) and 1.66 (1.28, 2.17), respectively, when compared with Q1. 4343 participants were included in the trajectory analysis, based on the longitudinal change patterns of the METS-IR and TyG index, the following three trajectory groups were identified: low-increasing, moderate-stable, and elevated-increasing groups. Multivariate Cox regression revealed that the hazard ratio (95% confidence interval) for CVD risk in the elevated-increasing trajectory group of the METS-IR and TyG index was 2.13 (1.48, 3.06) and 2.63 (1.68, 4.13), respectively, when compared with the low-rising group. The C-index, integrated discrimination improvement value, and net reclassification improvement value were enhanced after adding the baseline METS-IR and TyG index values to the Framingham model (P<0.05).Conclusions: Elevated baseline IR surrogates and their higher long-term trajectories were strongly associated with a high risk of CVD incidence in Xinjiang’s rural areas. Regular METS-IR and TyG index monitoring can aid in the early detection of CVD-risk groups.