Results indicated higher accuracy for the AA groups over the EA group models. The research of biomarkers, prediction models, and machine learning algorithms rely on group homogeneity and relevant features. Therefore, the higher AA accuracy could be due to better fit of the features to the target and that EA genetic-based ancestry definition has variation leading to the reduction in EA group homogeneity. Studies have shown that EA forms a structured population due to historical immigration of diverse source populations67. Future studies might consider dividing EA groups to subgroups according to genetic variation or finding new approaches to define ancestry68,69.
