Seminar

Disease prediction using nation-wide health data and genetics

Andrea Ganna

The presentation will outline recent advances in disease prediction through the integration of national health registries, genomic data, and artificial intelligence. It will introduce a new foundation model trained on nationwide Finnish data (over 7 million individuals and 6.5 billion records), while highlighting challenges related to the amplification of disparities across regions and socioeconomic groups, underscoring issues of fairness and generalizability. The presentation will also demonstrate how polygenic scores capture lifelong disease risk and complement electronic health record–derived risk scores, with each providing strengths across different disease domains. Integrating genomic and EHR data enables improved trial emulation, strengthens causal inference, and supports the design of more representative clinical studies. Overall, the talk will emphasize that equitable and ethically grounded integration of AI and genetics is essential to achieving precision prevention at a population scale.