Clinical documents structure analysis

The structure of clinical documents, and more specifically their sections, have a strong importance when assessing the context of clinical information extracted from clinical documents. For example, a diagnosis of asthma extracted from a Family History section has a very different meaning than the same diagnosis extracted from a Chief Complaint section! To reliably detect and classify clinical document sections, we are developing a module based on machine learning methods to classify sections into about 20 section classes, combining knowledge and resources from the Automated Problem List system, the SecTag algorithm, and standard terminologies including sections information.