IDENTIFICATION OF HYPERLIPIDEMIA RISK FACTORS AND QUANTIFICATION OF TCM CONSTITUTIONAL CONTRIBUTIONS: A MULTIDIMENSIONAL METRIC FRAMEWORK

Abstract

Against the backdrop of rising hyperlipidemia prevalence and limited management resources, the precise identification of key risk factors in high-risk populations is of great significance for improving public health. Based on clinical data from 1,000 middle-aged and elderly patients, this study constructed a composite measurement framework combining linear contraction and nonlinear evaluation, aiming to systematically identify core risk indicators for hyperlipidemia and assess the marginal contribution of Traditional Chinese Medicine (TCM) constitutions. The study first employed a LASSO regression model with L1 regularization to perform embedded screening of the initial high-dimensional features, effectively overcoming the interference of multicollinearity among lipid indicators. Subsequently, supplementary validation was conducted by integrating importance rankings from random forests, ultimately identifying eight core early warning indicators, including total cholesterol (TC), triglycerides (TG), BMI, and the total score of the physical activity questionnaire. These indicators demonstrated strong clinical predictive utility. Regarding the quantification of constitutional risk contributions, to address statistical biases caused by mixed constitutions, the study introduced a regularized logistic regression model with signed constraints for corrective estimation. Empirical results indicate that the Phlegm-Dampness constitution is the susceptibility factor with the highest risk of disease onset and the greatest statistical significance, with an odds ratio of 1.15, while the Harmonious constitution exhibits a clear health-protective effect. This study provides empirical evidence for the development of an early warning system for chronic diseases that deeply integrates traditional Chinese medicine (TCM) and Western medicine.