A neural network approach to analyzing health care information

The article describes an investigation into the use of artificial neural networks for the analysis of health information. The Kohonen self-organizing map technique was used to group 39 European countries according to data extracted from the World Health Organization's Health for All database. The groups were seen to exhibit significantly different characteristics with respect to life expectancy, probability of dying before 5 years of age, infant mortality rate, standardized death rate (SDR) for diseases of the circulatory system, and SDR for external causes of injury and poisoning. Results obtained using the technique were subsequently confirmed by the use of traditional statistical tests.