Privacy-Preserving Methods for Feature Engineering Using Blockchain: Review, Evaluation, and Proof of Concept

Area of Science:

• Computer Science
• Biomedical Informatics
• Data Privacy Engineering

Background:

• Protecting private data is crucial in research involving identifiable participant information.
• Limiting data collection scope and preventing secondary use are key risk management strategies.
• Ideal data collection frameworks integrate secure feature engineering without third parties.

Purpose of the Study:

• Compare current data privacy approaches for practicality and effectiveness.
• Evaluate traditional, cryptographic, secure hardware, and blockchain-based techniques.
• Assess privacy preservation in feature engineering for biomedical research.

Main Methods:

• Defined properties for evaluating data privacy approaches.
• Conducted a qualitative comparison of various privacy techniques.
• Utilized a geolocation data sharing use case for biomedical research evaluation.

Main Results:

• Trusted third-party approaches lack sufficient privacy guarantees in modern ecosystems.
• Cryptographic techniques offer strong privacy but face limitations in use cases and computational complexity.
• Blockchain alone is insufficient; hybrid approaches (blockchain with cryptography or TEEs) show promise, with a software implementation provided.

Conclusions:

• Blockchain and smart contracts facilitate privacy-preserving feature engineering, reducing reliance on trusted parties.
• Combining blockchain with cryptography or secure hardware presents promising avenues for data privacy.
• Hybrid blockchain and TEE frameworks offer practical tools for privacy-preserving applications.