Bayesian time‐varying autoregressive models of COVID‐19 epidemics

Abstract

The COVID‐19 pandemic has highlighted the importance of reliable statistical models which, based on the available data, can provide accurate forecasts and impact analysis of alternative policy measures. Here we propose Bayesian time‐dependent Poisson autoregressive models that include time‐varying coefficients to estimate the effect of policy covariates on disease counts. The model is applied to the observed series of new positive cases in Italy and in the United States. The results suggest that our proposed models are capable of capturing nonlinear growth of disease counts. We also find that policy measures and, in particular, closure policies and the distribution of vaccines, lead to a significant reduction in disease counts in both countries.

A Systematic Review of INGARCH Models for Integer-Valued Time Series

“…Hence, it is this temporal variability in the data that drew Roy [ 12 ] to explore its trends by the time-varying Bayesian INGARCH model. Similarly, Giudici et al [ 3 ] was also concerned about the time-varying features of COVID-19 and proposed Bayesian time-dependent Poisson autoregressive models. Additionally, Gning et al [ 76 ] focused on COVID-19 in Senegal and China.…”

“…For example, during an influenza outbreak, the number of new confirmed cases is reported daily, even down to each community. The analysis of such data is one of the fundamental tasks in epidemic forecasting and policy implementation (see Agosto and Giudici [ 1 ], Agosto et al [ 2 ] and Giudici et al [ 3 ] among others). Secondly, -valued count time series taking values in the range are the appropriate tool to employ when attention is turned to, for example, the changes in athletic performance by the difference between the number of goals scored in each game and that in the previous one.…”

Projections of wastewater as an indicator of COVID-19 cases in corrections facilities: a modelling study

“…We present the development of a hierarchical Bayesian spatial-temporal model within corrections facilities. In contrast to extant count time series models, [14][15][16][17][18] the present model not only accounts for potential instances of non-reports but also explicitly incorporates the dynamic temporal lags between wastewater-based virus concentration and the occurrence of positive clinical cases. This model aims to provide insights and answers to the wastewater concentration associated with the detection of at least one positive COVID-19 case of an incarcerated person.…”