Five-second facial video AI screens hypertension and diabetes
A study to be presented at ESC Congress 2026 shows an AI algorithm can detect undiagnosed hypertension (about 90% accuracy) and diabetes (about 81% accuracy) from a 5-second facial video, using pulse-wave dynamics and skin blood-flow signals. The method could enable contactless, scalable screening outside clinics, with prior 30-second results indicating higher accuracy. Validation in larger, more diverse cohorts is needed, and cuff-free blood pressure estimation was also explored.