An application of a mixture of exponential distributions for assessing hazard rates from COVID-19

Abstract

In the present study, we are interested in modeling repose time periods (the length of the time intervals between successive deaths) caused by a new, widespread disease called covid-19. This is useful for predict-ing probabilities of new deaths that occur within pre-determined time intervals. In practical applications, the choice of the statistical model is crucial for obtaining accurate estimates of death hazard rates. Based on an earlier research, we propose to use a mixture of exponential distributions; this model is simple to implement when hazard rates obtained from the components of the mixture are easily calculated, and it is adequate for dealing with nonstationary time series as those appearing in the case of this disease. The model is then applied to the example of Italy, and it appears to be also useful for comparing hazard rates along time.