AI in Education & Trauma Data Science

Deep learning models can teach as well as assist. We have shown that AI-based feedback improves medical students’ interpretation of hip fractures (Insights into Imaging, 2020) and studied AI-assisted training for rib fracture interpretation in a prospective study of undergraduate medical students (World Journal of Emergency Surgery, 2026).

In parallel, the lab conducts registry-based trauma outcomes research built on the Linkou CGMH trauma registry (2008–present, more than 33,000 patients). Ongoing projects include the impact of massive transfusion protocol implementation and prediction models for blood product utilization, dynamic hemorrhage prediction for trauma patients, and applications of large language models for trauma registry coding such as Injury Severity Score assignment.