○ Topic: Electronic Health Records: Question Answering & Multi-Modality

○ Lecturer: Professor Yoon-Jae Choi (KAIST AI Graduate School)

○ Date: October 1, 2021 4:30PM – 5:30PM

○ ZOOM Meeting ID: 832 6600 9463
https://snu-ac-kr.zoom.us/j/83266609463

○ Inquiries: Jongjin Lee (ga0408@snu.ac.kr)

○ Abstract: Electronic health records (EHR) contain valuable information regarding how patients are treated individually, as well as statistical insights on medical practice conducted by large hospitals. However, due to its complex structure and heterogeneous data modalities, non-database experts often have difficulties in utilizing such information. In this talk, we introduce question answering based on EHR, which enables users to interact with machines using natural language to retrieve information from the raw EHR. Furthermore, focusing on the multi-modality of EHR, we discuss how to handle such heterogeneity using multi-modal learning with Transformer architectures.