A standardised, high-throughput approach to diagnostic group testing method validation

Abstract

1AbstractBackgroundGroup testing, combining the samples of multiple patients into a single pool to be tested for infection, is an approach to increase throughput in clinical diagnostic and population testing by reducing the number of tests required. In order to further increase the throughput and accuracy of these approaches, mathematicians regularly devise novel combinatorial methods. However, although these novel methods are easily validated in silico, they are often never implemented in diagnostic laboratories because of the lack of clear and standardised pathways to clinical validation.MethodsWe develop a standardised analytical workflow that makes use of high-throughput automation and virus-like particle standards to validate theoretical group testing approaches. We then utilise specially developed virus-like particles for SARS-CoV-2, Influenza A, Influenza B, and Respiratory Syncytial Virus (RSV) to develop and validate a novel multiplex group testing approach based on simulated annealing and Bayesian optimization. Our approach improves the inference of positive samples in group testing, leveraging the quantitative nature of RT-qPCR test results.ResultsOur results show a higher positive predictive value of our novel approach for the inference of positive samples compared to the standard approach using binary test outcomes. In large-scale surveillance testing our method can greatly reduce the number of false positive identifications. Our in vitro findings show the viability of group testing for multiplexed testing of respiratory infections and demonstrate the potential of a novel inference method. Both innovations increase the number of people that can be tested with the available resources, which is particularly important in low-resource settings.