A Parallelized Approach Toward Solving the Weighted Consensus Model for Classifying COVID-19 Infection

CT scans are proved to be one of the best ways to identify the presence of COVID-19 and diagnose the patients. To reduce the workload on doctors, many researchers have come up with automatic classification techniques. However, all the research was done to improve the proposed model’s accuracy. To improve the reliability and robustness of predictions, recent work [1] focused on allotting weightage to predictions. This paper aims at improving the performance of the existing one in both reliability and computational terms. We compare different metrics’ ability to compute the weights of the base models and show the computational time improvement by parallelism. The proposed approach achieved about 4.2 speedup and an efficiency of around 60%. The proposed parallelizable approach works best when a bulk of test samples are to be tested to reduce the total testing time.