Disease prediction using nation-wide health data and genetics


See more information at https://metrics-and-models.github.io/!

Utilising data from Finland, where genetic details of approximately 500,000 individuals are linked to a national health registry, this presentation will explore the use of polygenic scores and electronic health records (EHRs) to predict disease susceptibility. It will highlight the challenges and successes of employing polygenic scores for disease prediction. Additionally, the talk will discuss the application of machine learning algorithms to predict healthcare outcomes from EHRs, assessing their fairness and generalizability across different populations. The presentation will conclude with a direct comparison of genetic and EHR-based predictions to assess which approach is most predictive and generalizable across healthcare systems.