Dissertation measures privacy-preserving AI cost on ECG data
Original title: Privacy-preserving AI can protect your heart data, the bill is the problem
On about 21,800 recordings, encryption overhead ran from roughly 30% to 1,200% while accuracy stayed near 97% to 99.9%.
A PhD dissertation by Beyazıt Bestami Yüksel at Istanbul Technical University splits ECG handling into sensor, transit and storage, and collaborative computation boundaries. It uses lightweight cryptography at the edge, TLS and AES in transit and storage, and homomorphic encryption with federated learning when hospitals train together. Tests on a public set of about 21,800 recordings, divided as simulated hospitals, kept diagnostic accuracy between about 97% and 99.9%. The work assumes an honest-but-curious aggregator and was not a live deployment.
Unconfirmed: The hospital split was simulated, not a production deployment.
- Identical measurements across different patients added about 30% compute overhead.
- Different data about the same patients raised overhead to about 450%.
- The most tangled data split reached about 1,200% overhead.
- Sites train locally and send updates that can be aggregated while encrypted.
- Organizations
- Istanbul Technical University
From the original article
3 min read at helpnetsecurity.com
A heart monitor used to be a box beside a hospital bed. Electrodes fed a cable, the cable fed a screen, and a nurse watched the screen. The threat model fit on a sticky note.
By Mirko Zorz · helpnetsecurity.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.