Alliance for Pandemic Preparedness

July 16, 2020

Smart Pooling AI-Powered COVID-19 Testing



  • [Pre-print, not peer reviewed] Escobar et al. developed a pooling method for SARS-CoV-2 testing protocols that uses machine learning to predict the probability of a negative result from sociodemographic data. The technique was most efficient at lower disease prevalence, and showed efficiency gains of 42% with respect to individual testing at disease prevalence up to 25%. 

Escobar et al. (July 15, 2020). Smart Pooling AI-Powered COVID-19 Testing. Pre-print downloaded July 16 from