RETRACTED ARTICLE: Securing health care data through blockchain enabled collaborative machine learning.

Area of Science:

• Machine Learning
• Blockchain Technology
• Data Privacy

Background:

• Traditional data transfer in machine learning raises privacy concerns, especially in healthcare.
• Centralized data transfer methods are limited and pose security risks.
• Decentralized approaches are needed for secure and efficient model exchange.

Purpose of the Study:

• To investigate privacy-preserving model transfer between users and organizations using federated learning.
• To explore the use of blockchain technology for rewarding client contributions.
• To establish a secure and efficient decentralized framework for machine learning model sharing.

Main Methods:

• Federated learning techniques for privacy-preserving model training and transfer.
• Blockchain technology for incentivizing and rewarding participating clients with tokens.
• Utilizing the COVID-19 dataset to evaluate the federated learning process.

Main Results:

• The federated learning approach enabled successful model transfer between users and volunteer organizations.
• Client contributors were effectively rewarded with tokens via blockchain.
• Individual model accuracies reached 88% (contributor a), 85% (contributor b), and 74% (contributor c).
• The FedAvg algorithm achieved an overall accuracy of 82%.

Conclusions:

• Federated learning provides a viable solution for secure, decentralized model transfer.
• Blockchain integration offers a robust mechanism for rewarding contributions in federated learning systems.
• The proposed framework enhances collaboration and data privacy in machine learning applications.