Abstract:In the Internet of medical things, data from a single source can be easily analyzed. Besides, it is paramount to collect data from multiple sources to provide consistent, accurate, and vital information as compared to that collected from a single source. Data fusion enables the combination of data from multiple sensors. However, privacy concerns need to be addressed during data aggregation. Furthermore, since medical data is becoming increasingly sophisticated via multimodal means, there is a high likelihood that a complex relationship among biological processes exists. This challenge motivates the study. This study is proposing a secure federated learning system based on data fusion using multi-party computation and additive secret-sharing method. The gradient parameters of the federated training model are protected rather than the actual data. The purpose of protecting the gradient parameters is to address the tradeoff between prediction accuracy and data privacy. In the proposed model, a convolutional neural network, which has an efficient weight-sharing approach, is employed for the prediction. The weight of the model is encrypted using a multi-computation and additive secret sharing. Extensive simulations are carried out to ascertain the efficiency of the proposed model. Besides, the proposed system model is compared with existing works in the literature in terms of accuracy, precision, recall and f1-score. Simulation results show that the proposed model has 97% accuracy as compared to 93% for random forest, 91% for Naive Bayes, 92% for logistic regression and 95% for decision tree. Furthermore, F1-score of the proposed model is 89% as compared to 85% for random forest, 80% for Naive Bayes, 82% for logistic regression and 79% for decision tree. Security analysis of the proposed system model is performed, which shows that the system is robust against weighted majority vote attack, honest-but-curious party and other security related attacks.

Explainable federated learning scheme for secure healthcare data sharing

A Federated Learning Framework Via Decentralized Data Valuation for Chronic Disease Healthcare

A federated learning system with data fusion for healthcare using multi-party computation and additive secret sharing

A Comprehensive Privacy-Preserving Federated Learning Scheme with Secure Authentication and Aggregation for Internet of Medical Things

Federated learning for preserving data privacy in collaborative healthcare research

A privacy preserving federated learning scheme using homomorphic encryption and secret sharing

Homomorphic Encryption-based Privacy-preserving Federated Learning in IoT-enabled Healthcare System

Medical Federated Model with Mixture of Personalized and Sharing Components

Federated Medical Learning Framework Based on Blockchain and Homomorphic Encryption

Designing ECG Monitoring Healthcare System with Federated Transfer Learning and Explainable AI

Medical Federated Model with Mixture of Personalized and Shared Components

Towards Fair and Privacy Preserving Federated Learning for the Healthcare Domain

Research on medical data security sharing scheme based on homomorphic encryption

A Security-Enhanced Federated Learning Scheme Based on Homomorphic Encryption and Secret Sharing

Efficient and privacy-preserving online diagnosis scheme based on federated learning in e-healthcare system

Privacy-Preserving Federated Learning for Internet of Medical Things Under Edge Computing

FEEL: A Federated Edge Learning System for Efficient and Privacy-Preserving Mobile Healthcare

Security and Privacy Issues and Solutions in Federated Learning for Digital Healthcare

A Scalable Federated Learning Approach for Collaborative Smart Healthcare Systems with Intermittent Clients using Medical Imaging

Adaptation of Federated Explainable Artificial Intelligence for Efficient and Secure E-Healthcare Systems

Review on security of federated learning and its application in healthcare