Extracting Medication Information from Discharge Summaries

Extracting medication information from clinical records has many potential applications and was the focus of the i2b2 challenge in 2009. We present a hybrid system, comprised of machine learning and rule-based modules, for medication information extraction. With only a handful of template-filling rules, the system’s core is a cascade of statistical classifiers for field detection. It achieved good performance that was comparable to the top systems in the i2b2 challenge, demonstrating that a heavily statistical approach can perform as well or better than systems with many sophisticated rules. The system can easily incorporate additional resources such as medication name lists to further improve performance.