P-14 A Deep Learning Approach to Identifying Outcomes

Reading through medical abstracts to find key data points can be a time consuming task. The ACRES system, built by Dr. Rodney Summerscales, automatically extracts key information from many abstracts. This allows a large set of results to be compared, the most successful of which can be studied further. This research is focused on the identification and extraction of text containing outcomes measured for each treatment group in a clinical trial using an artificial neural network. The goal is to score a higher degree of accuracy than the previous approach which used trained classifiers built with supervised learning techniques.