A Review of Privacy Enhancement Methods for Federated Learning in Healthcare Systems.

Area of Science:

• Artificial Intelligence
• Machine Learning
• Healthcare Informatics

Background:

• Federated learning (FL) enables collaborative model training without sharing raw patient data.
• FL is crucial for Medical Internet of Things (MIoT) and Electronic Health Records (EHRs) to maintain data privacy.
• Despite privacy benefits, FL faces challenges as model updates can risk data re-identification.

Purpose of the Study:

• To review and analyze existing privacy and security enhancement methods for FL in healthcare.
• To identify and discuss seven core techniques used to protect data in healthcare FL systems.
• To explore the strengths, limitations, and future potential of these privacy-enhancing techniques.

Main Methods:

• Literature review of privacy and security enhancement methods in healthcare FL.
• Analysis of seven identified techniques: Differential Privacy, Homomorphic Encryption, Blockchain, Hierarchical Approaches, Peer to Peer Sharing, Intelligence on the Edge Device, and Mixed/Hybrid Approaches.
• Discussion of the trade-offs and future directions for each technique.

Main Results:

• Seven primary categories of privacy-enhancing techniques for healthcare FL were identified.
• Each technique's strengths, limitations, and applicability in healthcare FL were evaluated.
• The study highlights the ongoing research focus on securing FL in sensitive medical environments.

Conclusions:

• Various techniques exist to mitigate privacy risks in healthcare FL, each with unique trade-offs.
• Further research is needed to optimize these methods for robust privacy and security in medical applications.
• The future of FL in healthcare relies on effectively implementing and advancing these privacy-preserving strategies.