Systematic review of privacy-preserving Federated Learning in decentralized healthcare systems

Federated Learning (FL) has become a promising method for training machine learning models while protecting patient privacy. This systematic review examines the use of privacy-preserving techniques in FL within decentralized healthcare systems. It compares existing methods such as Differential Privacy (DP), Trusted Execution Environment (TEE), Zero Knowledge Proofs (ZKP), Homomorphic Encryption (HE), Watermarking, Blockchain, and Secure Multi-Party Computation (SMPC) based on regulatory compliance, scalability, computational cost, complexity, and mathematical foundations. The principle challenges in decentralized healthcare like heterogeneous data, privacy risks, security threats, and compliance issues have been discussed. The review also highlights the importance of adhering to global regulations like HIPAA, GDPR, and country-specific data protection laws. Furthermore, it discusses open challenges and suggests future research directions to overcome current limitations, including computational efficiency, adversarial attacks, and the creation of policy frameworks for standardization. Overall, this review provides a unique perspective on ethical, secure, and scalable privacy-preserving FL models for the next generation of healthcare applications. • Analyzes essential techniques: Differential Privacy, SMPC, HE, TEE, ZKP, and Blockchain. • Reviews key privacy techniques: DP, SMPC, HE, TEE, ZKP, and Blockchain. • Compares methods based on cost, scalability, and resilience in FL. • Identifies issues such as non-IID data, high communication, and compliance. • Suggests hybrid and hardware-aided frameworks for secure FL. • presents future needs in terms of explainability, interoperability, and quantum security.