SecuMed-SIoT: A Hybrid CNN-Transformer IDS for Enhanced Security in Healthcare SIoT Networks

Abstract

Healthcare IoT (Internet of Things) systems have transformed patient monitoring and data management, but their extensive interconnectivity and sensitive data make them highly susceptible to cyber-attacks. Traditional intrusion detection systems (IDS) often fail to meet the stringent security demands of healthcare IoT environments due to high false-positive rates, computational inefficiencies, and limited adaptability to emerging threats. This paper introduces SecuMed-SIoT, a novel, security-focused hybrid IDS specifically designed for healthcare IoT, leveraging Social IoT (SIoT) principles and a CNN-Transformer architecture (CTLGNet) to enhance threat detection capabilities. SecuMed- SIoT incorporates security-driven interaction modeling to evaluate device behaviours and collaborates with a network of trusted devices to detect anomalies in real time, achieving high detection accuracy with minimal false alarms. Extensive experiments demonstrate that SecuMed-SIoT attains a detection accuracy of 94.2% and a false-positive rate of 3.1%, significantly outperforming conventional IDS models in both performance and efficiency. These findings underscore SecuMed-SIoT effectiveness in protecting sensitive healthcare data and ensuring device security within SIoT networks.