Statistical analysis of influenza propagation pattern using prescription data from Tochigi Prefecture

Abstract

In this study, we estimate the propagation patterns of influenza using prescription data obtained from pharmacies in multiple areas. In our model, we assume that a peak in the volume of sales of medicine corresponds to a peak in an influenza epidemic, and we use a cross-correlation function to estimate propagation patterns by estimating the interregional gap in the time-series of the volume of sales of anti-influenza medicine. We also examine the causal relation between different time-series using the Granger causal test. Based on the propagation patterns estimated from these causal relations, we determine that the influenza virus consistently spreads outward from city centers. Finally, we assess the reproducibility of results obtained from the two estimation methods used in this study using a stochastic SIR model.