Researchers have developed TIBECG, a novel temporal information bottleneck architecture designed to protect demographic privacy in electrocardiogram (ECG) waveforms. This method aims to allow the release of ECG data for diagnostic purposes without exposing sensitive information like age or sex. TIBECG was evaluated on individual leads from hospital recordings, demonstrating its ability to retain diagnostic utility while significantly reducing demographic inference compared to conventional methods. The study also highlights that direct downsampling can suppress demographic inference but often at a greater cost to diagnostic accuracy than TIBECG. AI
IMPACT This research could enable more secure sharing of sensitive medical data for AI-driven diagnostics and research.
RANK_REASON The cluster contains an academic paper detailing a new method for privacy preservation in medical data. [lever_c_demoted from research: ic=1 ai=1.0]
- alphaXiv
- arXiv
- DagsHub
- electrocardiography
- Hugging Face
- IArxiv
- Kullback--Leibler divergence
- TIBECG
- Tsai-Min Chen
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →