Searching of Correlation of Weather and Cardiologic Events Computer Methods for Relation Discovery and Events Prediction

摘要

This paper presents our recent research on finding correlation between weather events and medical (cardiologic) events and trends. Such correlation is intuitive, however no solid proof exists. Such correlation was investigated, but it most cases it drives to conclusion that differences are visible in large periods of time (like year seasons, winter for example). We are trying to show, that such correlation is visible in much shorter time periods (as couple of days) with unusual weather behaviour). We examine standard statistical methods, advanced events and trend detection methods and neural networks (self organizing maps) usage. We propose basic scenarios for medical events frequency increase prediction, according to weather forecasts.