Causal data science in psychiatric epidemiology
"What if" an individual with psychosis takes medication for sleeping problems? Will their mental health improve? If running a randomised controlled trial is not an option, but we have an understanding of the mechanisms regulating the interplay between symptoms we may use the tools of causal inference to predict the effect of an intervention aiming to modulate certain symptoms. More often we have an idea of the important variables, but no full picture of how they come together to determine the overall mechanism of action. Finding causal diagrams compatible with the observed data is a way of gaining insights into the connections between symptoms and ultimately predicting plausible intervention effects, which may help prioritise experimental studies. Drawing upon real-world case studies in psychiatric epidemiology, we will explore state of the art strategies for causal discovery from observational data and describe an end-to-end operational pipeline to learn diagrams from data and predict intervention effects.
"What if" an individual with psychosis takes medication for sleeping problems? Will their mental health improve? If running a randomised controlled trial is not an option, but we have an understanding of the mechanisms regulating the interplay between symptoms we may use the tools of causal inference to predict the effect of an intervention aiming to modulate certain symptoms. More often we have an idea of the important variables, but no full picture of how they come together to determine the overall mechanism of action. Finding causal diagrams compatible with the observed data is a way of gaining insights into the connections between symptoms and ultimately predicting plausible intervention effects, which may help prioritise experimental studies. Drawing upon real-world case studies in psychiatric epidemiology, we will explore state of the art strategies for causal discovery from observational data and describe an end-to-end operational pipeline to learn diagrams from data and predict intervention effects.
Speakers
Giusi Moffa is an Assistant Professor of Statistics at the Department of Mathematics and Computer Science of the University of Basel. At the heart of her research are causal graphical models and their translational value in health care, clinical research and epidemiology. Real-world problems motivate her work on statistical methods for causal discovery and techniques for estimating intervention effects from observational data. Giusi holds a PhD in computational statistics from the University of Bristol, UK. After conducting post-doctoral research in statistical bioinformatics at the University of Regensburg, Germany, Giusi gained experience as a statistician in clinical drug development in the pharma industry at Novartis, Basel, before returning to academia.
Join us on April 11th & 12th 2024
.png)
