2026 ASEE Annual Conference & Exposition

As healthcare systems increasingly adopt artificial intelligence (AI), engineering students need greater awareness of how measurement bias can influence medical care, downstream AI models, and clinical decisions. This short paper describes a hands-on pulse oximetry experiment that is part of a semester-long, high school course on data science and AI in healthcare that was implemented across 10 high schools ( approximately 400 students) in one northeastern U.S. state. Pulse oximeters estimate blood oxygen saturation using red and infrared light and are widely used despite documented accuracy limitations for individuals with darker skin pigmentation. We leverage this limitation as an anchor phenomenon that makes bias tangible through engineering practices. Students collect, visualize, and analyze measurements, and discuss the limitations of pulse oximeters and mitigation efforts. Guided by the research question, “How do high school students and teachers engage with and reason about a pulse oximetry activity as an anchor for understanding and mitigating bias in healthcare AI?, we draw on triangulated data including teacher implementation notes, teacher and student surveys and interviews, student artifacts, and platform trace indicators. Preliminary findings suggest the activity is feasible in real classrooms, supports strong engagement, and elicits emerging mechanistic and bias-oriented reasoning. To support educator adoption, the submission includes a lesson plan and practical guidance for classroom data collection and analysis.