Abstract

The social gradient of health literature has put considerable effort into understanding the underlying mechanisms of inequalities in health among individuals. In particular, the complexity of social reality and its dynamics hamper the researchers' design of clear causal pathways that intertwine individuals' social positions and health outcomes. To grasp better the intricacy of this social phenomenon, current epidemiologic and public health research is increasingly relying on biostatistical tools to analyze and make inferences from health data. In this presentation, I outline three methodological approaches that could be an asset to the Deparment of Public Health Sciences at Stockholm University. These methodological approaches are based on my previous work at the University of Luxembourg.

The presentation's first topic consists of applying the Bayesian regression technique in public health research. In particular, this first part will focus on the applications and potential advantages of using the Bayesian distributional model. The second part of this presentation consists of an outline of the Joint Modelling of Longitudinal and Survival data techniques, recently developed in biostatistics. The main advantage of this statistical class regards the ability to control for endogenous, longitudinal phenomena in the context of the proportional hazard model and missing data patterns. The last part of the presentation outlines the potential advantages of applying unsupervised machine learning techniques (such as clustering and classification analysis) to study the neighborhood effects on individuals' health outcomes.

Welcome

Illustration: Ruslan Nesterenko/Mostphotos