Binary disease prediction using tail quantiles of the distribution of continuous biomarkers

In the analysis of binary disease classification, numerous techniques exist, but they merely work well for mean differences in biomarkers between cases and controls. Biological processes are, however, much more heterogeneous, and differences could also occur in other distributional characteristics (e.g. variances, skewness). Many machine learning techniques are better capable of utilizing these higher-order distributional differences, sometimes at cost of explainability. In this study, we propose quantile based prediction (QBP), a binary classification method based on the selection of multiple continuous biomarkers and using the tail differences between biomarker distributions of cases and controls. The performance of QBP is compared to supervised learning methods using extensive simulation studies, and two case studies: major depression disorder (MDD) and trisomy. QBP outperformed alternative methods when biomarkers predominantly show variance differences between cases and controls, especially in the MDD case study. More research is needed to further optimise QBP.