Causal Modelling of Spatio-Temporal Epidemiological and Clinical Data

The assumption that data are independent and identically distributed samples from a single underlying population is pervasive in statistical and machine learning modelling. However, most real-world data do not satisfy this assumption. Regression models have been extended to deal with structured data collected over time, space, and different populations. But what about causal network models, which often use regression for their local distributions? In this talk, I will discuss how to learn well-specified models with causal discovery from environmental, epidemiological and clinical data. As a motivating example, I will focus on scalable and interpretable techniques, modelling the interplay between weather patterns, pollution, mental conditions and dermatologic problems.