A secure, private, and explainable IoHT framework to support sustainable health monitoring in a smart city

Internet of Health Things (IoHT) have allowed connected health paradigm ubiquitous. 5 G supported healthcare vertical allows IoHT to offer connected health monitoring with quality of service and ultra-low latency. Deep learning has shown potential in processing massive amount of IoHT data that are generated daily, automate connected healthcare workflows, and help in decision making processes. However, three important challenges need to be addressed to attain long term healthcare-related sustainability - data security, data privacy, and social acceptance of deep learning process. In this paper, we propose a framework that will allow healthcare sustainability through the following contributions 1) ensure privacy of training dataset, 2) support the aggregation of the global model gradients through a private Blockchain-brokered entity, 3) support trustworthiness and provenance of the federated clients by blockchain and off-chain, 4) share the dataset, train the model and share trained model among the federated clients in an encrypted fashion, and 5) add explainability and reasoning of deep learning process to make the model acceptable by the society. We will present the detailed design of our proposed sustainable system, the implementation details and test results. The test results show promising prospect of achieving sustainability of IoHT-enabled connected health applications.

Automated summary

Internet of Health Things (IoHT) faces challenges in data security, privacy, and social acceptance, but deep learning helps in processing large amounts of IoHT data and providing connected health monitoring. This paper proposes a framework to achieve healthcare sustainability through ensuring privacy, supporting global model gradient aggregation, adding explainability, etc.