Bibi I, Schaffert D, Blanke P, Illian L, Lenzing F, Martin N, Leipe J, März W et al.
AutoML models were trained on data from the LURIC study (n = 3316) and validated in the UMC/M cohort (n = 423) to predict Lp(a) levels, specific cardiovascular diseases, and cardiovascular mortality. Phase 1 achieved AUC 0.6249 to 0.9101 with age, Lp(a), troponin T, BMI, and cholesterol as leading predictors; Phase 3 mortality models reached AUC 0.74 to 0.85. Data drift detected across study phases underscores the need for periodic model recalibration.
Scientific Reports, 20. Oktober 2025 | PMID 41116068