Predicting Whom to Test is More Important Than More Tests - Modeling the Impact of Testing on the Spread of COVID-19 Virus By True Positive Rate Estimation

Paul F Rodriguez

Using these two examples one could produce a plausible ROC curve of sensitivity (TP rate) vs specificity (FP rate), Figure 1 . In this case, the values on the axis are percentages from 0 to 1. Importantly, the graph is not easily parameterized by the absolute number of tests but should be thought of as parameterized by relative values as follows. If we think of the cruise ship example, we imagine that all the passengers & crew are given a score that predicts their likelihood of having COVID-19, and then tested in sequence of rank 1 to rank P, the population size, according to the score. As we test from 1/P to P/P fraction then the TP rate and FP rate . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.