Swarm Learning-based Secure and Fair Model Sharing for Metaverse Healthcare‏

Abstract

Metaverse is a digital space that aims to build a fully immersive, hyper spatio-temporal virtual world for human interaction. It is very promising to apply metaverse into personal healthcare to improve the quality and efficiency of personal healthcare services. Several challenges, such as patient data security and privacy issues, are hindering the application of metaverse healthcare. To protect data security and privacy of healthcare, swarm learning is proposed by integrating distributed machine learning and blockchain. However, the existing swarm learning framework often faces additional issues of security and fairness in metaverse healthcare, such as security concerns caused by multiple anonymous avatars and uneven distribution of data quality. In this article, we propose a swarm learning-based model sharing framework to enhance the security and fairness of healthcare-AI model sharing in the metaverse. The proposed metaverse swarm learning can support the privacy-protection global model and partial model parameters merging. Moreover, the decentralized autonomous organization blockchain network is proposed to guarantee the fairness of model sharing gains among the imbalance of healthcare resource data. Simulation results on two practical healthcare datasets show that our proposed model-sharing can achieve better accuracy than local training and approximate accuracy compared to central training.

Keywords: Swarm learning, Metaverse, Healthcare, Model sharing, Security, Fairness.