PROJECTS

This project investigates privacy threats in publicly shared electrocardiogram (ECG) data, where biometric features make individuals vulnerable to re-identification. Unlike prior work assuming idealized adversaries, we evaluate risks under realistic conditions with partial attacker knowledge. Using data from 109 participants, our approach achieves 85% re-identification accuracy, showing that even limited knowledge enables effective linkage. These findings reveal the inadequacy of simple anonymization and underscore the need for stronger safeguards — including differential privacy, access control, and encrypted computation — to protect sensitive biosignals while preserving their utility for healthcare research.