Mainzelliste SecureEpiLinker (MainSEL): privacy-preserving record linkage using secure multi-party computation

Area of Science:

• Computer Science
• Medical Informatics
• Cryptography

Background:

• Record linkage is crucial for data analysis but faces privacy challenges with personally identifiable information (PII).
• Existing methods often require a trusted third party, which is prohibited by privacy laws for sensitive data like medical records.
• Privacy-preserving record linkage (PPRL) aims to link records without leaking PII.

Purpose of the Study:

• To develop a fault-tolerant framework for privacy-preserving record linkage (PPRL).
• To enable secure linkage of medical records without compromising patient privacy.
• To implement PPRL without relying on a trusted third party.

Main Methods:

• Utilized secure multi-party computation (sMPC) for privacy-preserving record linkage.
• Integrated the sMPC framework with the Mainzelliste medical record keeping software.
• Ensured no leakage of personally identifiable information (PII) under standard cryptographic assumptions.

Main Results:

• Demonstrated a fault-tolerant PPRL framework using sMPC.
• Achieved efficient patient record linkage in simulated realistic network conditions.
• Linked a patient record against 10,000 records in 3.9-48 seconds depending on network latency.

Conclusions:

• The developed PPRL framework is feasible for real-world applications.
• Secure multi-party computation offers a viable solution for privacy-preserving data linkage in healthcare.
• The approach guarantees the non-leakage of PII, adhering to privacy regulations.