QUANTITATIVE GENETIC MODELS FOR PREDICTING POLYGENIC DISEASE RISK ACROSS POPULATIONS

Abstract

Background: Polygenic diseases, such as diabetes, cardiovascular disorders and some cancers are diseases caused by the combined effect of multiple genetic variants and environmental factors. Quantitative genetic models are increasingly important for estimating disease susceptibility with polygenic risk scores in diverse populations.

Objective: The aim of this study is to assess the predictive accuracy and transferability of quantitative genetic models in predicting the risk of polygenic diseases in multi-ethnic populations.

Methodology: Analysis of genome-wide association study datasets from ~50,000 individuals from European, Asian and African ancestry groups. Polygenic risk prediction models were constructed and validated using statistical approaches such as Bayesian regression, linear mixed models and machine learning algorithms. We measured predictive performance using area under the curve (AUC), odds ratios, and cross-population calibration metrics.

Findings: The proposed models showed better predictive accuracy with an average AUC of 0.81 within-population and 0.74 in cross-population validation. Bayesian-based models were almost 12% more accurate than conventional regression approaches in classifying risk. The addition of ancestry-specific variants made prediction more reliable in underrepresented populations significantly.

Conclusion: Quantitative genetic models offer a powerful approach for predicting polygenic disease risk at the population level. Incorporating population diversity and ancestry-informed variants can significantly improve predictive accuracy and foster equitable delivery of personalized genomic medicine.