A Mixture of Regressions Model of COVID-19 Death Rates and Population Comorbidities


As the COVID-19 pandemic spread worldwide, it has become clearer that prevalence of certain comorbidities in a given population could make it more vulnerable to serious outcomes of that disease, including fatality. Indeed, it might be insightful from a health policy perspective to identify clusters of populations in terms of the associations between their prevalent comorbidities and the observed COVID-19 specific death rates. In this study, we described a mixture of polynomial time series (MoPTS) model to simultaneously identify (a) three clusters of 86 U.S. cities in terms of their dynamic death rates, and (b) the different associations of those rates with 5 key comorbidities among the populations in the clusters. We also described an EM algorithm for efficient maximum likelihood estimation of the model parameters.

MIDAS Network Members