Accelerating probabilistic privacy-preserving medical record linkage: A three-party MPC approach

Area of Science:

• Computer Science
• Data Security
• Bioinformatics

Background:

• Record linkage is crucial for integrating data across healthcare and research.
• Probabilistic Privacy-Preserving Record Linkage (PPRL) is vital for sensitive data integration.
• Balancing privacy, accuracy, and efficiency in large-scale PPRL is a significant challenge.

Purpose of the Study:

• To introduce a novel and efficient Probabilistic Privacy-Preserving Record Linkage (PPRL) method.
• To enhance the speed and scalability of PPRL using a secure 3-party computation (MPC) framework.
• To address the growing demand for secure and efficient data integration methods.

Main Methods:

• Developed a novel PPRL method utilizing a secure 3-party computation (MPC) framework.
• Enabled multiple parties to compute linkage results without revealing private inputs.
• Focused on improving the speed and efficiency of the linkage process.

Main Results:

• The proposed PPRL method achieves up to 14 times faster performance than state-of-the-art (SOTA) methods.
• Maintains linkage quality comparable to existing SOTA MPC-based PPRL techniques.
• Demonstrated significant scalability and efficiency, e.g., linking 10,000 records in 8.74s (700 Mbps) vs. 92.32s (SOTA).

Conclusions:

• The novel PPRL method offers an efficient and scalable solution for large-scale record linkage.
• The approach ensures robust privacy protection through secure 3-party computation.
• Presents a promising tool for secure data integration in privacy-sensitive applications.