Abstract
The aim of this research is to develop a bio-weather preventive medicine methodology individualized to people's health and living environments based on analyzing relationships between meteorological elements and meteorotropism. In this paper, using a large database of emergency medical transfers, we analyze quantitatively the effect of weather on disease by decision tree-making as a data mining method. In the analysis for cerebral infarction, a data set composed of meteorological elements and transfer counts for each season and period of average is constructed, decision trees are extracted from the data, branches in the decision tree at which the onset rate increases are investigated, and these results are compared from the viewpoints of the factors responsible for the increase. The result that the decision tree induced from the average meteorological conditions within every 6-hour period makes clear the bifurcation of the onset rate is obtained. Seasonal analysis suggests that the onset rate increases remarkably when the relative humidity decreases in winter and when the temperature increases in summer. Consequently, this analyzing method is efficient at inducing the relationships between meteorological elements and increases in the disease.
