2020/07/20 by Shayan Hassantabar, Stefano Novati, Hassantabar, Shayan +15
Computer Science · Medicine · #Anomaly Detection Techniques and Applications #COVID-19 diagnosis using AI #FOS: Computer and information sciences #Human-Computer Interaction (cs.HC) #Neural and Evolutionary Computing (cs.NE) #SARS-CoV-2 detection and testing
paper · pdf · doi:10.48550/arxiv.2007.10497
openalex publication_date 2020/07/20 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
The novel coronavirus (SARS-CoV-2) has led to a pandemic. The current testing\nregime based on Reverse Transcription-Polymerase Chain Reaction for SARS-CoV-2\nhas been unable to keep up with testing demands, and also suffers from a\nrelatively low positive detection rate in the early stages of the resultant\nCOVID-19 disease. Hence, there is a need for an alternative approach for\nrepeated large-scale testing of SARS-CoV-2/COVID-19. We propose a framework\ncalled CovidDeep that combines efficient DNNs with commercially available WMSs\nfor pervasive testing of the virus. We collected data from 87 individuals,\nspanning three cohorts including healthy, asymptomatic, and symptomatic\npatients. We trained DNNs on various subsets of the features automatically\nextracted from six WMS and questionnaire categories to perform ablation studies\nto determine which subsets are most efficacious in terms of test accuracy for a\nthree-way classification. The highest test accuracy obtained was 98.1%. We also\naugmented the real training dataset with a synthetic training dataset drawn\nfrom the same probability distribution to impose a prior on DNN weights and\nleveraged a grow-and-prune synthesis paradigm to learn both DNN architecture\nand weights. This boosted the accuracy of the various DNNs further and\nsimultaneously reduced their size and floating-point operations.\n