Capstone 2016-2017

Clinical trials are research studies that explore whether a medical treatment is safe and effective for humans. For example, a clinical trial to determine the maximally tolerated doses of dinaciclib and bortezomib for the treatment of relapsed multiple myeloma. The identification of patients who satisfy a set of predefined criteria for the trial is instrumental. However, the process of distinguishing these patients on the basis of their clinical records is a time consuming and challenging task. The specific criteria about a patient is often contained within the unstructured text portions of a clinical record such as radiology reports, progress notes, and discharge summaries. A large number of studies have been published describing ways to automate this process but require the development of complex manually validated term searches. One difficulty with this is data normalization; there are many ways to describe a single concept. For example, Heart Attack and Myocardial Infarction both refer to the death of the heart muscle. The goal of this project is to develop a framework to process clinical records for cohort discovery.

Cohort Assessment & Retrieval Environment (CARE)