Methodology for predicting hospital admissions and evaluating recovery rates for coronavirus disease in Japan

In this study, we aimed to propose a method to predict the number of patients needing hospitalization using a combination of available technologies. We developed a method to predict the number of hospital admissions by combining a simple susceptible-infected-recovered (SIR) model with the relationship between the number of new positive cases and the number of hospital admissions, increasing the reliability of each prediction. The accuracy of the concordance between the actual number of patients and the predicted number of hospitalized patients was 99%. Owing to the high accuracy, we were also able to establish a method to evaluate recovery rates. This facilitated determination of the effectiveness of measures implemented throughout Japan to reduce the number of treatment days. The model developed in this study facilitates immediate estimation of the maximum number and timing of hospitalizations based on the peak of new positive cases. Moreover, it provides a statistically true value of the recovery rate required by the mathematical model for investigating countermeasures.