Evaluation of Group Testing for SARS-CoV-2 RNA

Nasa Sinnott-Armstrong; Daniel Klein; Brendan Hickey

We provide an overview of the simulation methods employed. We evaluated models using a sampling approach, drawing samples from either a Bernoulli distribution with a probability equal to the assumed prevalence, or a beta-Bernoulli distribution with α parameter 5 and β parameter 5/prevalence to account for inter-individual variation in risk. All draws were independent. These were averaged over 1000 replicates to smooth over any sampling noise; sample sizes are adjustable in our web interface. Next, the draws are used to assign true case status for each simulation run. The number of random values thus corresponds directly to the count of cases tested by a given method in a single batch. Of note, the draws from the beta are used as the case probability directly for the row major ordered case prevalence tests ( Figure S1 ), which are only informative under the beta-Bernoulli simulations.