For every epidemic, the public and decision-makers worry that the number of infected people will be far more than the
healthcare system can afford, thus it is important to predict the cases which need medical services. This is similar to cases of natural
disasters such as floods when a devastating outcome occurs with incoming and outgoing water volume being higher than a river’s
storage capacity. The similarities between floods and epidemics inspire a modified rainfall-runoff model, I&α. This novel model
focuses on the prediction of active cases, i.e., y0, which measures the number of people who need medical services. This model is
based on the past data for the prediction of future epidemic trends. This model can effectively predict the maximum y0 and its peak
date when applied to model COVID-19, with an average error of 3.8% and 2.7 days, respectively. The average error for y0 on May
12, 2020 is 22.7%.