S1 Text - Assessing Biases in the Evaluation of Classification Assays for HIV Infection Recency

The supplementary material contains additional information regarding: 1) the impact of biomarker variability on assay performance; 2) the assay’s performance at different TSI categories for the simulated distributions; 3) the effect of the TSI distribution on the diversity biomarker cutoff; 4) the comparison of the binary classification approach (as we do in the main text) with a probabilistic approach to classifying individuals into recent or chronic; 5) the uncertainty around the assay performance at different TSI categories for the empirical distributions; and 6) the use of weighted resampling as a way to overcome, in some cases, the evaluation biases shown in this work. (PDF)