Deep Learning on Electronic Health Records for Research in Pharmacoepidemiology: Examples From the Field of Oncology

Deep learning has lately attracted a lot of attention and seen many great achievements in the fields of imaging, genomics and digital pathology. However, the experience of this methodology on potential use cases in pharmacoepidemiology and the performance in comparison with established approaches is rather limited. This talk highlighted two recent studies which utilized deep learning in an oncological electronic health record database for the development of inferential and prognostic models.