Federated & Centralized Deep Learning for Cyberattack Detection in IoMT

Abstract

As technology continues to evolve, one of the most promising recent advancements is the Internet of Things (IoT) and its applications in various fields, especially health- care. This development has helped formulate the concept known as the Internet of Med- ical Things (IoMT), which has tremendous possibilities in improving healthcare services, especially in remote patient monitoring, real-time vital signs tracking, and automated medical decision processes. This study focuses on the impact of artificial intelligence, par- ticularly deep learning models, in increasing the security of IoMT environments, which face the brunt of serious cyber threats. Attack detection was modeled with four different deep learning models: two centralized (CL-CNN and CL-LSTM) and two federated (FL- CNN and FL-LSTM) using the CIC-BCCC-NRC-IoMT-2024 dataset. Evaluation results demonstrate that federated learning achieves commendable results while maintaining data confidentiality, thereby serving as a powerful method for protecting intelligent medical systems.