NHash: Randomized N-Gram Hashing for Distributed Generation of Validatable Unique Study Identifiers in Multicenter...

Area of Science:

• Biomedical Informatics
• Data Management
• Clinical Research

Background:

• Unique study identifiers are crucial for linking research data while protecting patient privacy.
• Traditional identifiers face challenges in large, multicenter studies due to scale and potential for errors.
• Error tolerance (validatability) is a key requirement for study identifiers to ensure data integrity.

Purpose of the Study:

• Introduce Randomized N-gram Hashing (NHash), a novel method for generating unique study identifiers.
• Enable distributed and validatable identifier generation for multicenter research.
• Ensure identifiers are pseudonymous, collision-resistant, and error-tolerant.

Main Methods:

• NHash employs a two-phase approach: randomized N-gram hashing and subsequent encryption.
• Phase 1 generates an intermediate string using N-gram hashes derived from randomized inputs.
• Phase 2 encrypts the intermediate string and concatenates it with a random number for the final identifier.

Main Results:

• Experiments with synthesized data showed a negligible probability of identifier collision with NHash.
• NHash was successfully implemented for the Center for SUDEP Research (CSR) multicenter collaboration.
• The CSR involves 14 institutions focused on understanding sudden unexpected death in epilepsy (SUDEP).

Conclusions:

• The CSR Data Repository effectively utilized NHash to link deidentified multimodal clinical data.
• NHash met all objectives for generating unique, distributed, and validatable study identifiers.
• The method supports secure data linkage in complex, multicenter research environments.