IS.10: Statistical Issues in Developing and Evaluating Polygenic Risk Models

5 August 2020 | 12:00 UTC/GMT

Time Zone Converter
Speaker Abstracts

Advances in technology, increasing sample sizes, and advances in statistical methods have enabled the development of polygenic risk models, which include hundreds to millions of inherited genetic variants. These models could inform personalized prevention, treatment recommendations, or stratified screening programs. This session, organized by the International Genetic Epidemiology Society, will introduce polygenic risk models and review statistical challenges to fitting these models (including very high-dimensional data, limited access to individual-level data, combining polygenic data with known clinical risk factors and biomarkers), to calibration (limited availability of large cohorts with complete genotyping and extensive follow-up), and to evaluating clinical utility (assumptions regarding the costs, risks and benefits of proposed interventions). Using cancer and psychiatric diseases as examples, talks will review polygenic risk scores' current empirical performance and future prospects, with a special emphasis on the importance of developing models in populations from multiple ancestries and evaluating their utility in specific medical and public health contexts.

Celia Greenwood, Lady Davis Institute for Medical Research, Jewish General Hospital

Frank Dudbridge, University of Leicester
Evaluating risk prediction of multiple outcomes

Sohee Park, Yonei University

Recent advances in individualized cancer risk prediction models in Korea

Alicia Martin, Massachusetts General Hospital
Polygenic risk scores for the world: current applications, limitations, and promise

Peter Kraft, Harvard T.H. Chan School of Public Health