Healthcare AI/ML Confidential Computing Platform

Mary Beth Chalk, Business Development, Center for Digital Health Innovation, UCSF Innovation Ventures. Cyberattacks on healthcare data are a clear and present danger resulting in constricted access to high fidelity clinical data. At the same time, healthcare artificial intelligence (AI) algorithms utilized in care delivery must access this type of data to ensure the algorithms will perform consistently when utilized in the care delivery setting. The future of healthcare AI is highly dependent upon the development and adoption of zero trust data environments in which no trust is extended to any party and yet the parties are still able to collaborate. Confidential computing platforms and privacy preserving analytics enable data owners to maintain control of their data at all times while providing encrypted access to algorithm owners (whose models are also encrypted for their protection). Mary Beth discusses how confidential computing can be utilized to further protect healthcare data while also making it available for research and development of clinical workflow and decision support algorithms.