AI, Health & Confidentiality
summary
Panelists describe Intel’s hardware-backed confidential computing and an open-source federated learning stack used to train models without moving raw records, citing a UPenn brain-tumor study spanning 71 sites on six continents that reports a 33% uplift in true-positive accuracy. A European research effort funded via Gaia-X prototypes an Internet of health data: personal health clouds with ledgered access logs (blockchain-like) that let individuals grant revocable, privacy-preserving research access and aim to give newborns lifetime portable medical records.
AI-written summary · about
chapters (5)
- 0:00AI in healthcare intro
- 3:41Gaia X and health data clouds
- 8:56Confidential computing at Intel
- 14:51Federated learning with hardware security
- 20:23Personalized healthcare timeline
related talks
14mFaster AI model training with confidential computing
13mProvably Secure AI for Health Data – Powered by Confidential Computing
01h 16mEuropean AI powered Solutions to Combat Dementia. How to Implement AI in Health...
30mPanel Discussion: State of the Union: Confidential AI From the Field
17mTowards the medicine of the future with confidential computing
01h 20mUnlocking societal benefits of AI with privacy protective technologies.
39mOpenFL: Building Better AI Models with Private Data
01h 16mHealthy AI for access, sharing and protection of sensitive data.