Machine learning for determining lateral flow device results for testing of SARS-CoV-2 infection in asymptomatic populations

• DOI: 10.1016/j.xcrm.2022.100784
• Publication Date: 2022-09-27T09:41:27Z
• Authors: Andrew D. Beggs; Camila C.S. Caiado; Mark Branigan; Paul Lewis-Borman; Nishali Patel; Tom Fowler; Anna Dijkstra; Piotr Chudzik; Paria Yousefi; Avelino Javer; Bram Van Meurs; Lionel Tarassenko; Benjamin Irving; Celina Whalley; Neeraj Lal; Helen Robbins; Elaine Leung; Lennard Lee; Robert Banathy
• Keywords: Machine Learning; 03 medical and health sciences; COVID-19 Testing; 0302 clinical medicine; SARS-CoV-2; Humans; COVID-19; Sensitivity and Specificity; Article; 3. Good health

Rapid antigen tests in the form of lateral flow devices (LFDs) allow testing a large population for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). To reduce variability device interpretation, we show design and an artifical intelligence (AI) algorithm based on machine learning. The learning (ML) is trained combination artificially hybridized LFDs LFD data linked to quantitative real-time PCR results. Participants are recruited from assisted test sites (ATSs) health care workers undertaking self-testing, images analyzed using ML algorithm. A panel clinicians used resolve discrepancies. In total, 115,316 returned. ATS substudy, sensitivity increased 92.08% 97.6% specificity 99.85% 99.99%. self-read 16.00% 100% 99.15% 99.40%. An ML-based classifier results outperforms human reads self-reading.