Event Details

2^nd October at 13:00-14:00
Hybrid event
in Säätiö lecture hall, Medisiina D / Zoom

Professor Andreas Mayr, Philipps University Marburg, Germany
Boosting polygenic risk scores and gene-environment interactions in large-scale cohort studies

Coffee/tea and pulla available for the first 40 attendees! Or join to listen to the lecture via Zoom: https://utu.zoom.us/j/61366447409

Andreas Mayr is Head of the Institute for Medical Biometry and Statistics (IMBS) at Philipps University Marburg, Germany. His methodological research focuses on statistical modelling and prediction, with particular emphasis on distributional regression and modern regression techniques. He is the author of both methodological and clinical publications, as well as a book on distributional regression. Before joining Marburg, he held positions at the University of Bonn (Epidemiology) and LMU Munich (Applied Stochastics). He completed his PhD and habilitation at FAU Erlangen-Nürnberg Medical School and studied statistics at LMU Munich and the University of Buenos Aires (UBA), Argentina. He is an editor of the Statistical Modelling Journal and a member of the Clinical Ethics Committee in Marburg.

Large-scale cohort studies are a key resource for understanding how genetic predisposition interacts with environmental and lifestyle factors to influence disease risk over time. At the same time, incorporating high-dimensional genomic data into such studies remains challenging, particularly when aiming to move beyond standard polygenic risk score (PRS) approaches based on GWAS summary statistics. In his talk, Prof. Mayr will present boosting-based methods (Mayr et al. 2014) for deriving and extending PRS using individual-level data from large-scale cohorts. Building on the snpboost framework (Klinkhammer et al., 2023, 2024), he illustrates how statistical boosting enables scalable estimation of polygenic models while accounting for the complex structure of genetic data.  A particular focus will be on recent work on boosting distributional regression for detecting gene–environment interactions (Wu et al., 2026). By modeling not only the mean but also the variability of clinical outcomes, this approach allows the identification of genetic variants that may modify the effect of environmental exposures, thereby supporting more targeted prevention strategies. The talk will highlight how these methods can be applied in prospective cohort settings to improve risk prediction, better understand heterogeneity in disease development, and ultimately contribute to better prevention strategies.

Selected publications

Mayr, A., Binder, H., Gefeller, O., & Schmid, M. (2014). The evolution of boosting algorithms – from machine learning to statistical modelling. Methods of Information in Medicine, 53(06), 419-427.

Klinkhammer, H., Staerk, C., Maj, C., Krawitz, P. M., & Mayr, A. (2024). Genetic prediction modeling in large cohort studies via boosting targeted loss functions. Statistics in Medicine, 43(28), 5412-5430.

Klinkhammer, H., Staerk, C., Maj, C., Krawitz, P. M., & Mayr, A. (2023). A statistical boosting framework for polygenic risk scores based on large-scale genotype data. Frontiers in Genetics, 13, 1076440.

2529164123. Wu, H. Klinkhammer, K. Kunwar, C. Staerk, C. Maj, & A. Mayr (2026). Detecting gene–environment interactions to guide personalized intervention: Boosting distributional regression for polygenic scores. Proceedings of the National Academy of Sciences, 123(14), e2529164123.