Abstract

Dictated and transcribed medical records are converted to XML. Several transformations are used to impose structure on the documents. The XML documents are then used to extract medical facts. Finally this information is presented to end users via a web browser. I discuss the various DTDs used to represent the information, describe the annotation format used to mark up the facts extracted from the documents, and discuss the issues we encountered with this system along the way. Finally, I describe how this system benefited from using XML as the storage format for the documents and annotations.