Categorization of Warning Statements for Over-the-Counter Medications

Abstract

The use of dietary supplements (including vitamin, minerals, fiber acids, etc.), a kind of over-the-counter (OTC) drugs, by U.S. adults is significantly increasing nowadays. However, the supplements can cause side effects and interactions with conventional prescriptions. The DailyMed provides OTC drug labels containing black box warnings. The information embedded in narrative is not directly searchable, thus preventing its wide use by clinical decision support system and consumers. The objective of this research is to develop an automatic program to categorize (e.g., interactions, side effects) warning statements from drug labels. After developing the patterns to extract safety information, we can categorize warning statements with precision of 0.948 and recall of 0.918.