Natália M. Arruda , Federal Institute of São Paulo
Tiago Carvalho, Federal Institute of São Paulo
Luciana C. Alves, University of Campinas (UNICAMP)
The aim of the present study was to investigate the relationships among socioeconomic, structural, contextual and health factors and the probability of adult mortality in the Brazilian small areas in 2010. The analyses were based on data from the 2010 Demographic Census and Mortality Information System of DATASUS. The machine learning method was used to establish the determinants of probability of adult mortality. Machine learning methods allow a better understanding of the interactions between the different factors. The results showed that mortality rates due to external causes, unemployment rate, proportion of blacks, vaccination coverage and proportion of whites were the ones that obtained the greatest predictive power in the risk of adult mortality probability using the algorithms Random Forest, Extreme Boosted Trees, Support Vector Machine and Naive Bayes. The algorithms obtained good performance and were effective in analyzing the variables, although some correlated, with the outcome of adult mortality probability. Identifying the determinants of adult mortality and the main disparities between social groups and in small areas is extremely important in helping to build public policies that respond adequately to the specific needs of each region and social group, thus contributing to reduce the socioeconomic inequalities and mortality.f probability of adult mortality.
Presented in Session P2. Poster Session Ageing, Health and Mortality