Modeling of premature mortality rates from chronic diseases in Europe, investigation of correlations, clustering and granger causality

Abstract

In this study mortality rates due to chronic diseases in 31 European countries are examined, through certain time series modeling, based on the electronic database of Eurostat. The time series are grouped into 6 clusters using Fuzzy K-means method combined with Jensen-Shannon divergence enabling a more accurate investigation of the homogeneity within clusters. The intuitive interpretation of these results relies on studies concerning the differentiation of Europeans’ habits due to geopolitical position, nutritional, sociopolitical and environmental factors. Spearman coefficient, distance correlation and cross-correlation are used to reveal the underlying correlations while causality is checked through a parametric test, aiming to disclose the existence of Granger causality by examining all the pairs of the respective time series of premature mortality rates. Correlation indices and causality tests, highlight many statistically significant relationships with respect to their dynamics, comprehending the relationships governing the living standards of European countries. Finally, Europe’s average mortality is modelled while special emphasis is given to the emergence of the ideal representative of Europe’s rates. A detailed investigation of the relationships and patterns that characterize premature mortality from chronic diseases across Europe is provided, examining each country either individually or in connection to the other countries, while each statistically generated outcome is supported by results concerning the economic status, environment, nutrition and especially status of public health.