Active learning for concept extraction from clinical free text

This thesis is a step towards automating information extraction from clinical free-text. It establishes a Cost-efficient Enhanced Active Learning framework to significantly reduce annotation cost, while ensuring high-quality extracted information. The practical significance of this research is three-fold: (1) benefitting the overall patient healthcare by facilitating downstream eHealth workflows such as supporting clinical information processing and efficient decision making, (2) benefitting the research in medical informatics by facilitating the development of rich annotated corpora from clinical free text resources, and (3) benefitting the research in machine learning by developing domain-independent and effective active learning approaches