Days after IBM got approval for UIMA as an OASIS standard, IBM and the Mayo clinic have announced that they’ve set up a web site dedicated to UIMA-based annotators that target Health records. The wiki pages and source repository is hosted by the National Cancer Institute.
The joint initiative is called the Health Natural Language Processing Consortium. Their goal is to help speed adoption of electronic medical records.
Unstructured Information Management Applications (UIMA) are small apps that process and can extract meaning of a portion of an unstructured document. Each tiny UIMA app has a specific purpose, like identifying and deciphering place names, people’s names, or dates, etc. The idea is that a battery of UIMA apps can be pipelined and directed towards an unstructured doc, and be able to append structure to it, creating a semi-structured document. There is an open-source implementation of UIMA that is available as an Apache Incubator project.
The system set up by IBM and the Mayo Clinic hopes to be able to extract information from doctors’ notes (good luck on that one) and also from unstructured pathology reports. The idea is to help doctors with their diagnoses and also to further medical research.
