Processing Electronic Medical Records : Ontology-Driven Information Extraction and Structuring in the Clinical Domain

Natural Language Processing and Information Extraction systems can annotate free text with meta-data, thus making it possible to automate processes within clinical care as well as helping medical research gain insight into the tremendous amount of data which lies hidden in electronic medical records. In this thesis we will present an overview of recent developments of methods and resources used in intelligent and robust systems that can extract accurate meta-data automatically from unstructured sources. We will focus on the use of ontologies as a mean to extract information. Further we will evaluate a modern IE system by using its structured output to automatically diagnosing patients based on their discharge summary.