An adversarial-resilient intrusion detection framework for internet of medical things (IoMT) using digital twin-enabled behavioral threat modeling and federated hybrid ensemble learning

The Internet of Medical Things (IoMT) has transformed healthcare by enabling continuous patient monitoring and remote diagnostics. However, this growth introduces considerable security challenges. This paper examines vulnerabilities in IoMT devices to advanced cyberattacks that jeopardize patient safety and data integrity. We review the limitations of traditional security methods and motivate the need for adaptive defenses. We evaluate machine learning and deep learning models for real-time threat detection and identification of anomalous behavior within IoMT networks. We further propose a security framework that integrates digital twin technology with edge-cloud computing to improve the reliability of IoMT applications. Results show that hybrid and deep-learning models maintain detection performance under the resource constraints typical of medical devices. The proposed XGBoost component achieved a precision of 0.97, a recall of 0.98, and an ROC-AUC of 0.999 on the SmartWard dataset, while the hybrid ensemble showed measurable adversarial robustness under FGSM perturbations. The decision-critical inference path runs in under 0.05 seconds, supporting deployment on resource-constrained medical devices.