Abstract
Influenza is an infectious disease that has been a serious threat to human beings. In this paper, we propose an influenza prediction model by introducing geographic information into a previously proposed model. Using sequence data of observed virus strains, we select a vaccine that is predicted to be most effective in the following year in Japan. We show that the vaccine strains selected by the model are closer to the circulating strains than the vaccine strains officially proposed by the governmental and international organizations. Moreover, we confirmed that the contribution of geographic information improves the prediction.
