For many incoming students, the high enrollment gateway courses such as algebra-based physics courses pose a serious challenge. Techniques are needed for nurturing students in such courses, especially students who are at high risk of performing poorly. Our objective is to identify patterns in the daily lives of undergraduates that predict success or failure in introductory physics classes. We intend to use recently developed long-term ambulatory sensors to monitor students’ sleep patterns and to use data mining methods to identify patterns that predict poor performance. Objective metrics for identifying at-risk students can potentially enable instructors and counselors to intervene before major problems arise. Furthermore, showing students’ direct links between their sleep patterns and their academic performance can empower them to gradually adjust their behaviors to counteract negative trends. Pilot data from this project can be used to develop proposals for both the National Science Foundation and the Institute of Educational Sciences.