Controlled Randomization in the Simulation of Infectious Diseases

In this study, we explore a novel approach of using controlled randomization to improve the performance of COVID-19 disease models. The modeling of COVID-19 illnesses is a difficult task that requires striking a compromise between realism and simplification to capture the stochastic character of infectious disease dynamics. The COVID-19 presentation is variable, ranging from moderate illness that gets better in a few days to severe disease that may linger for weeks or more and result in mortality. The therapeutic relevance of therapy would be unmistakably shown by an improvement in mortality, but sample sizes for studies examining mortality are high and may be problematic, especially considering the huge number of possible medicines to be evaluated. Despite the varying temporal course of COVID-19, the proper summary metric for severity ratings has been debated. Fixed time points, such as comparing severity ratings at day 14, may overlook clinical benefits. An endpoint like time to improvement eliminates timing issues. Data will be analyzed using regression, T-test (TT), and other types of statistical analysis. Our analysis of recent COVID-19 therapeutic results and models for simulation trials allows us to assess the power of statistics for prospective trial results. Fixed time-point techniques' power relies on evaluation time. Time-to-event techniques outperform fixed time-point methods in statistical power, even at the optimum time.