Predicting health literacy in Brazil: a machine learning approach.

Abstract

Health literacy is essential for promoting well-being and the ability to make informed decisions. We investigated the level of health literacy in Brazil and identified the predictive factors that influence it. Our data contribute to the international context, with a focus on countries in the Global South and, in particular, Latin America. By analyzing health literacy in Brazil, this study sheds light on the challenges faced by populations with similar socioeconomic backgrounds in low- and middle-income countries, where disparities in access to education and health services are widespread. In addition to descriptive analysis, we used the Random Forest machine learning algorithm, which uses bootstrap aggregation (bagging). To make the results interpretable, we implemented Shapley's Additive exPlanation values. The results show a predominance of problematic levels of health literacy among the population. The analysis reveals that factors such as medication use, dependence on the Unified Health System (Sistema Único de Saúde), and educational level are significant predictors of health literacy. The findings highlight the need for public policies aimed at reducing socioeconomic disparities and improving the public health system in order to promote better access to and understanding of health information.