Target Audience

Healthcare institutions, research organizations, and technology partners seeking to develop AI models without compromising patient data privacy.

Challenge

Healthcare organizations face significant challenges in leveraging large datasets for AI development due to strict privacy regulations (e.g., HIPAA) and cybersecurity risks. Centralized data-sharing models pose re-identification risks, while siloed data limits collaborative innovation.

Solution Approach

Mayo Clinic and Google implemented a 'data under glass' model, where de-identified patient data remains in a controlled enclave (Mayo Clinic Cloud) while third-party algorithms can be tested and trained without data leaving the institution. This federated learning approach ensures data privacy while enabling collaborative AI development.

Value Add

Enables secure, scalable AI development by reducing data procurement and storage costs, fostering partnerships without compromising privacy, and accelerating medical research through shared insights without exposing raw data.

References

Mayo Clinic (2019–2021) in partnership with Google, with governance oversight from Mayo’s DaTA Board and One Table task force.

Read more here: https://www.ncbi.nlm.nih.gov/books/NBK594445/