L2D2: A Novel LSTM Model for Multi-Class Intrusion Detection Systems in the Era of IoMT

• Published in: IEEE access 2025, Vol.13, p.7002-7013
• Main Authors: Akar, Gokhan, Sahmoud, Shaaban, Onat, Mustafa, Cavusoglu, Unal, Malondo, Emmanuel
• Format: Article
• Language: English
• Subjects: Accuracy; Algorithms; Biomedical monitoring; Computer crime; Devices; Internet of Medical Things; Internet of Medical Things (IoMT); Internet of Things; Internet of Things Security; Intrusion detection; intrusion detection system; Intrusion detection systems; Long short term memory; Machine learning; Monitoring; Real-time systems; Security; security of healthcare systems; Temperature sensors
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2025.3526883

The rapid growth of IoT has significantly changed modern technology by allowing devices, systems, and services to connect easily across different areas. Due to the growing popularity of Internet of Things (IoT) devices, attackers focus more and more on finding new methods, ways, and vulnerabilities to penetrate IoT networks. Although IoT devices are utilized across a wide range of domains, the Internet of Medical Things (IoMT) holds particular significance due to the sensitive and critical nature of medical information. Consequently, the security of these devices must be treated as a paramount concern within the IoT landscape. In this paper, we propose a novel approach for detecting various intrusion attacks targeting Internet of Medical Things (IoMT) devices, utilizing an enhanced version of the LSTM deep learning algorithm. To evaluate and compare the proposed algorithm with other methods, we used the CICIoMT2024 dataset, which encompasses various types of equipment and corresponding attacks. The results demonstrate that the proposed novel approach achieved an accuracy of 98% for 19 classes, which is remarkably high for classifications and presents a significant and promising outcome for IoMT environments.