LLNL’s Data Science Challenge (DSC) took place in July with students from UC Merced and UC Riverside. During the intensive two-week internship at the Lab’s UC Livermore Collaboration Center, undergraduate and graduate students worked together on a challenge problem in cardiology, exploring a data-driven approach to reconstructing electroanatomical maps of the heart at clinically relevant resolutions and combining input from the standard 12-lead electrocardiogram (ECG) with advanced machine learning techniques. The ECG provides a noninvasive and cost-effective tool for the diagnosis of heart conditions. However, the standard 12-lead ECG is inadequate for mapping out the electrical activity of the heart in sufficient detail for many clinical applications (e.g., identifying the origins of an arrhythmia). In order to construct a more detailed map of the heart, current techniques require not only ECG readings from dozens of locations on a patient’s body, but also patient-specific anatomical models built from expensive medical imaging procedures. For this challenge, the students used a cardiac electrophysiology repository developed by senior staff scientist Mikel Landajuela.