Objectives To explore convenient and pragmatic predictive method of coronary heart disease (CHD) in urban residents.
Methods Risk factors of CHD were sifted out based on the case-control study and logistic regression analysis, Bayes step by step discriminatory analysis was used to establish the multi variables quantitative predictive model, cross-validation method was adopted to estimate the model’s effectiveness.
Results CHD Statistically correlated with age, abdomen circumference, lipoprotein, apolipoprotein B, high blood pressure history, being tired, some eating habits, et al (p≤0.05). Respectively used training sample and checking sample to authenticate effectiveness of the predictive model, results showed that sensitivity, specificity, coincidence, positive predictive value and negative predictive value all be more than 90%, positive likelihood ratio and negative likelihood ratio were ideal, furthermore resultant stability was good.
Conclusions Effectiveness of this predictive model of CHD is good.