Prognostic disease risk prediction through integration of genetic and biomarker panels

Preventive medicine seeks to identify individuals at elevated disease risk before clinical onset. Early
risk detection enables timely, targeted lifestyle and pharmacologic interventions that can improve
health outcomes. Conventional assessments combining biomarkers and family history provide
qualitative risk estimates, but biobank-scale, multifactorial models can allow more precise risk
stratification. Although most efforts emphasize lifetime risk prediction, shorter prognostic horizons can
support more actionable decision-making. Here, we seek to extend and validate a suite of linear
prediction models trained on UK Biobank data to estimate incident disease within ten years of baseline
assessment for 28 later-in-life diseases, integrating genotype data, routine blood biomarkers, and
common biometrics. These predictors perform strongly at the population level in independent samples
and stratify high-risk individuals for targeted clinical intervention, underscoring their potential for broad
clinical implementation. We seek to further extend our models out-of-biobank and demonstrate
predictive power and clinical utility in other populations.