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

Conversely, we can also imagine the curve is parameterized as a function of the score threshold. As we get more and more tests our threshold for who gets tested will decrease, and TP rate will increase. Of course, the false positive rate (i.e. individuals who are predicted to be positive but are negative) will increase as well but that cost is much less than misses (false negatives). A third way to consider the parameterization is a function of number of infected individuals (I), which is perhaps most relevant for current situation. If our testing is on the order of magnitude as I, and if we think we need to have 15* I tests, then the units could be considered multiples of I. In either case, the shape of the graph is the same but the increase in the efficacy of testing will depend on whether we can increase the number of tests, or decrease the pool of potentially infected persons, or just test for more days. If we could also improve the score thresholding, then that could increase the whole shape of the graph and change the inflection point.