BMI 6115 - Clinical Information Extraction

Web-based delivery. This course will focus on common methods for automated information extraction from clinical narrative text documents, and on pattern-matching, ontology-driven, and machine learning approaches. A corpus of de-identified clinical documents from earlier i2b2 NLP challenges will be used to expose students to the specificities and difficulties of clinical text, and will allow experimenting with different information extraction methods and comparing. The hands-on experiments will consist of two tasks: disease and syndromes extraction, and contextual analysis of the latter.