This article presents a comparative study on the predictive performance and generalizability of two types of risk scores for common diseases: Polygenic Scores (PGS), which use genetic data, and Electronic Health Record-based Phenotype Risk Scores (PheRS), which use individuals' health history. The research analyzed data from over 845,000 individuals across three European biobanks (FinnGen, UK Biobank, and Estonian Biobank) to predict the onset of 13 diseases. The findings indicate that PheRS are generally well-transferable across different healthcare systems, and importantly, PheRS and PGS are largely independent predictors of disease risk, demonstrating an additive benefit when combined for more accurate disease prediction. The authors conclude that integrating routinely collected EHR data with genetic information offers a cost-effective and powerful approach for estimating disease risk.

References:

• Detrois K E, Hartonen T, Teder-Laving M, et al. Cross-biobank generalizability and accuracy of electronic health record-based predictors compared to polygenic scores[J]. Nature Genetics, 2025: 1-10.