Genetic factors increase the identification efficiency of predictive models for dyslipidaemia: a prospective cohort study

Abstract Background Few studies have developed risk models for dyslipidaemia, especially rural populations. Furthermore, the performance of genetic factors in predicting dyslipidaemia has not been explored. The purpose this study is to develop and evaluate prediction with without Methods A total 3596 individuals from Henan Rural Cohort Study were included study. According ratio 7:3, all divided into a training set testing set. conventional conventional+GRS (genetic score) Cox regression, artificial neural network (ANN), random forest (RF), gradient boosting machine (GBM) classifiers area under receiver operating characteristic curve (AUC), net reclassification index (NRI), integrated discrimination (IDI) used assess ability models, calibration was show Results Compared lowest quartile GRS, hazard ( HR ) (95% confidence interval CI )) highest GRS 1.23(1.07, 1.41) population. Age, family history diabetes, physical activity, body mass (BMI), triglycerides (TGs), high-density lipoprotein cholesterol (HDL-C), low-density (LDL-C) AUCs Cox, ANN, RF, GBM 0.702(0.673, 0.729), 0.736(0.708, 0.762), 0.787 (0.762, 0.811), 0.816(0.792, 0.839), respectively. After adding increased by 0.005, 0.018, 0.023, 0.015 classifiers, corresponding NRI IDI 25.6, 7.8, 14.1, 18.1% 2.3, 1.0, 2.5, 1.8%, Conclusion Genetic could improve predictive model, suggesting that information be provided as potential predictor screen clinical dyslipidaemia. Trial registration registered at Chinese Clinical Register. (Trial registration: ChiCTR-OOC-15006699 . Registered 6 July 2015 - Retrospectively registered).