An Advanced Hierarchical Federated Learning-Based Framework for IoMT Cybersecurity Threat Detection

The increasing reliance on the Internet of Medical Things (IoMT) in modern healthcare systems has led to a significant rise in cybersecurity vulnerabilities. With a growing number of interconnected devices such as wearable monitors, infusion pumps, and remote diagnostic tools, safeguarding sensitive medical data and ensuring device integrity has become a major challenge. Traditional Centralized Machine Learning (Centralized ML) approaches for threat detection are often unsuitable due to data privacy concerns, bandwidth limitations, and the heterogeneity of medical devices. Moreover, frequent data transmission to central servers increases latency and the risk of data breaches, posing serious concerns in real-time clinical environments. To address these issues, this work proposes a Hierarchical Federated Learning (HFL)-based cybersecurity framework tailored for IoMT networks. Unlike conventional systems, method enables local model training on edge devices and aggregates updates through a multi-tier hierarchy, preserving privacy while minimizing communication overhead. The system detects anomalies such as unauthorized access, data tampering, and abnormal device behavior using deep neural networks deployed locally. Aggregated updates are combined efficiently at regional and central servers, enabling a global model to learn from diverse, distributed data without direct data sharing. Key features of the proposed framework include privacy-preserving learning through HFL, reduced data transmission, improved threat detection accuracy, and resilience against adversarial attacks such as data poisoning. The system is evaluated on real-world healthcare datasets and demonstrates high performance in detecting complex cyber threats in heterogeneous IoMT environments. This scalable and secure solution is suitable for deployment in hospitals, remote clinics, and telemedicine infrastructure.