Minimizing the usage of SARS-CoV-2 lab test resources through test pooling enhanced by classification techniques

Abstract

Testing is an effective practice to limit the spread of the SARS-CoV-2. PCR is an accurate method to detect SARS-CoV-2 infected individuals, but PCR lab test kits are scarce and expensive resources. Therefore, their usage should be optimized. Testing in batch (pooling) is a procedure that merges individuals’ swabs, allowing group diagnosis without affecting the accuracy of the results. Savings on test kits depend on the prevalence of the disease, pool composition, and size. We propose a novel approach for optimizing pooling to minimize the usage of lab test kits. We show that estimating the probability of an individual being infected by means of a binary classifier leads to improvements in the efficiency of pooling strategies. We use simulation to select the components of a new pooling strategy based on a classifier and evaluate our approach using a real dataset.