Estimation of the health status of people in the vicinity of pharmacies using pharmacy big data

Abstract

The purpose of this study was to propose a method for visualizing the patterns of the geographical propagation of influenza infection, and to elaborate parameters for the characterization of these patterns. First, a motion picture was prepared for the quotidian propagation of influenza infection in the Greater Tokyo Metropolitan area, which is considered a typical epidemic area for the 2012/2013 flu season. Second, hebdomadal recordings of patients with influenza infection in the 47 prefectures of Japan were grouped into 3 categories (1-peak, 2-peak, or multi-peak). The prefectures were arranged according to the weeks with the maximum number of patients, to examine variations in the temporal infection order of the districts among the flu seasons. These characteristics were analyzed using Cramer's coefficient of association and Spearman's rank correlation coefficient. Finally, the propagation of influenza infection was compared between urban and remote areas: the Greater Tokyo Metropolitan area and Tochigi prefecture. Regarding influenza virus infection, differences in population density, public transportation systems, and lifestyles between the urban and rural areas were found to lead to distinct endemic patterns of infection. Emphasis was placed on the so-called big data hubris.