Disease prediction with multi-omics and biomarkers empowers case–control genetic discoveries in the UK Biobank

Aiming to enhance early disease detection, here we present the MILTON (MachIne Learning with phenoType associatiONs), the machine-learning framework that utilized commonly measured clinical biomarkers, plasma protein levels and other quantitative traits in the UK Biobank and further validated in the FinnGen Biobank. MILTON, which was first trained on 67 biomarkers, outperformed traditional polygenic risk scores, achieving an area under the curve (AUC) of over 0.7 for 1,091 diseases and over 0.9 for 121 diseases. This approach holds potential for improving precision medicine by offering more accurate predictions and genetic insights into disease development.
All extracted gene–disease associations and incident disease predictive biomarkers are publicly available (http://milton.public.cgr.astrazeneca.com).