How hard will COVID-19 hit South Africa? A story in three parts.
Prof Wilfred Ndifon (AIMS) & Neil Turok (AIMS and Perimeter)
Monday, 4 May 2020, 16:00 (GMT+2)
Suppressing COVID-19 will likely require us to rapidly identify and isolate infected individuals, on an ongoing basis. However, RT-PCR tests are costly, making regular testing of every individual extremely expensive. In this talk we present a way to reduce the cost by collecting samples from everyone, then subdividing and combining them in various combinations, prior to testing. We propose an algorithm based on the geometry of a hypercube, which improves on traditional “group testing” (Dorfman 1943) approaches. It enables extremely efficient identification of infected individuals when the prevalence of the virus is low. For example, finding an infected individual in a population of 10,000 could take fewer than 20 tests. We shall describe field trials of this approach, now under way in Rwanda.