Differential Privacy-Enhanced Federated Learning in Medical Data Environments
DOI:
https://doi.org/10.64252/fxajf118Keywords:
Differential Privacy, Noise-based Privacy Protection, Privacy-Preserving Data Perturbation, FedAvg, Distributed Model Aggregation, Federated Learning, Decentralized Machine Learning, Collaborative Learning without Data Sharing, Edge-Based Model Training, Health Data Privacy, Confidentiality of Medical Records, Protection of Patient Information, Secure Handling of Health Data. Privacy Protection, Data Confidentiality Measures, Privacy Preservation Techniques, Information Security Controls, Sensitive Data Protection.Abstract
In today’s data-driven world, safeguarding the privacy and security of sensitive medical information particularly disease-related data has become a pressing concern. This research investigates the application of federated learning, differential privacy, and federated averaging to enable secure and private analysis of healthcare data. A novel framework is proposed that integrates these advanced privacy-preserving techniques to ensure individual data remains confidential while allowing collaborative analytics among various healthcare institutions. Through simulations and experimental evaluations, the framework’s ability to protect patient privacy without compromising data utility is assessed. The results highlight the potential of this approach to support secure data sharing and analysis in modern healthcare environments, contributing to the advancement of privacy-centric health data solutions.
