Öppet seminarium med Xu Zong om befolkningens hälsa, åldrande och maskininlärning

Seminariets titel: Machine learning for social determinants of health: From identifying predictors to estimating heterogeneous and causal effects

Xu Zong är postdoktor vid Helsinki Institute of Demography and Population Health, Helsingfors universitet, där han även disputerade. Hans forskning fokuserar på befolkningens hälsa, åldrande och maskininlärning. Hans forskning har publicerats i tidskrifter som Social Science & Medicine och uppmärksammats i internationella medier som The Telegraph.

Bakgrund till Zongs föredrag (på engelska):

Drawing on a series of recent studies, this talk illustrates three complementary ways in which machine learning can advance population health research. First, supervised and explainable methods, such as tree-based ensemble models combined with SHAP-based interpretation, can identify the most important determinants of health outcomes. Second, methods such as causal forests can move beyond average effects to uncover heterogeneity, revealing how these associations vary across countries and population subgroups. Third, approaches such as double (debiased) machine learning can help strengthen causal inference by flexibly adjusting for high-dimensional confounders while preserving valid statistical estimation. At the seminar, our guest speaker will illustrate when and how each of these tools can strengthen public health and social science research, and also discuss their limitations.

Seminariet arrangeras av Institutionen för folkhälsovetenskap vid Stockholms universitet.