Study on Links between Cerebral Infarction and Climate Change Based on Hidden Markov Models

Abstract

Several studies have shown that variability in weather has been linked to stroke occurrence. However, the association has not been clear. The purpose of this study was to determine the effect of the change of weather pattern on stroke onset using Hidden Markov Models (HMMs). The daily data of emergency transport from 2002 to 2005 in Nagoya City in Japan were used. As "weather patterns", we used 11-classes classified by Japan Weather Association (JWA), including "East high pressure and West Low pressure", "Two centers of Low pressure" and so on. We proposed a new type of application of HMMs. Normally, HMMs were used to predict a hidden state from observed data. In this paper, HMMs were applied to show the existence of influence of weather variability on stroke occurrence. Our HMM was compared with the stochastic process which ignores the influence of weather states. A statistical test leaded to conclude that weather influenced the occurrence of stroke. These findings suggest that further research is possible to give more detailed study of the relation between weather states and stroke occurrence, if we use more specified data of weather elements such as daily temperature, sea level pressure and so on.