Pooling of coronavirus tests under unknown prevalence
Alexander Pikovski, Kajetan Bentele
Abstract
Diagnostic testing for the novel coronavirus is an important tool to fight the coronavirus disease (Covid-19) pandemic. However, testing capacities are limited. A modified testing protocol, whereby a number of probes are 'pooled' (i.e. grouped), is known to increase the capacity for testing. Here, we model pooled testing with a double-average model, which we think to be close to reality for Covid-19 testing. The optimal pool size and the effect of test errors are considered. The results show that the best pool size is three to five, under reasonable assumptions. Pool testing even reduces the number of false positives in the absence of dilution effects.
Topics & Concepts
PoolingCoronavirusCoronavirus disease 2019 (COVID-19)False positive paradoxPandemicSevere acute respiratory syndrome coronavirus 2 (SARS-CoV-2)Statistics2019-20 coronavirus outbreakProtocol (science)Diagnostic testStatistical hypothesis testingVirologyMedicineComputer scienceMathematicsDiseaseArtificial intelligenceInfectious disease (medical specialty)OutbreakPathologyVeterinary medicineAlternative medicineSARS-CoV-2 detection and testingSARS-CoV-2 and COVID-19 ResearchMachine Learning and Algorithms