@aifbw Really nice paper. Would be good to see net re-classification/differential diagnosis in an external dataset, and the inclusion of more relevant phenotype information in the models to determine if the increased prediction due to genetics is independent of clinical info.