Privacy-preserving data sharing infrastructures for medical research: systematization and comparison

Area of Science:

• Medical research
• Data privacy
• Health informatics

Background:

• Data sharing is vital in medical research but faces privacy challenges.
• Existing methods for anonymous data sharing often restrict analysis capabilities.
• A clear systematization of privacy-protection trade-offs in data sharing is lacking.

Purpose of the Study:

• To develop a systematization for evaluating privacy protection and utility in data sharing approaches.
• To categorize existing data sharing methods based on privacy and usefulness.
• To identify research gaps in privacy-preserving data sharing.

Main Methods:

• Developed a systematization framework with axes for privacy (safe data, settings, outputs) and usefulness (de-duplication, flexibility, scalability).
• Adopted privacy axes from the Five Safes Framework.
• Applied the framework to categorize popular data sharing approaches.

Main Results:

• Categorized approaches into distributed data analyses, secure multi-party computation, and data enclaves.
• Identified research gaps, such as the lack of de-duplication for distributed data and limited flexibility.
• Highlighted fundamental differences and functional gaps among data sharing methods.

Conclusions:

• The developed systematization enhances transparency of privacy-preserving data sharing infrastructures.
• It aids decision-makers and regulatory authorities in understanding the trade-offs involved.
• Further research is needed to address identified functionality gaps in data sharing methods.