Abstract
Human height prediction based on genetic factors alone shows positive correlation, but predictors developed for one population perform less well when applied to population of different ancestries. In this study, we evaluated the utility of incorporating non-genetic factors in height predictors for the Han Chinese population in Taiwan. We analyzed data from 78,719 Taiwan Biobank (TWB) participants and 40,641 Taiwan Precision Medicine Initiative (TPMI) participants using genome-wide association study and multivariable linear regression least absolute shrinkage and selection operator (LASSO) methods to incorporate genetic and non-genetic factors for height prediction. Our findings establish that combining birth year (as a surrogate for nutritional status), age at measurement (to account for age-associated effects on height), and genetic profile data improves the accuracy of height prediction. This method enhances the correlation between predicted and actual height and significantly reduces the discrepancies between predicted and actual height in both males and females.