Universal Health Data Schemas for Privacy-Preserving AI Community Group

The mission of this group is to define a universal, modular, and interoperable set of data schemas for health information. Our goal is to enable the aggregation and utilization of data for medical research and AI training through privacy-enhancing technologies (PETs) like Zero-Knowledge Proofs (ZKPs), while ensuring patient control and consent via Verifiable Credentials (VCs).

Scope and Problem Statement

The development of robust medical AI is hampered by siloed, non-standardized, and sensitive health data. Current data formats are incompatible across institutions, and privacy regulations prevent the sharing of raw data, creating a significant barrier to collaborative research. This group will address this by creating schemas that transform health data into standardized, verifiable, and privacy-preserving assets.

Key Deliverables

• A core set of modular, extensible Verifiable Credential schemas for common medical data types (e.g., lab results, imaging reports, prescriptions, diagnoses)
• Best practice guidelines for issuing these VCs from trusted sources (e.g., hospitals, clinics)
• Specifications for generating Zero-Knowledge Proofs from these VCs to enable privacy-preserving queries and analytics
• Use cases and implementation patterns for federated learning and AI model training using the proposed schemas and ZKP protocols

Homepage

Homepage/Blog

Shortname

uhds

Participation

Leadership

Chairs

• Amir Hameed Mir

Links

 Mailing List

public-uhds