Walrus Optimization-Enhanced Machine Learning Intrusion Detection for the Internet of Medical Things

While Internet of Things (IoT) systems increase efficiency by enabling devices to exchange data by connecting with each other and the internet, they also require continuous monitoring and protection of security vulnerabilities. In this context, since the Internet of Medical Things (IoMT), a branch of IoT for the healthcare sector, enables medical devices and sensors to collect data over the network, detecting and preventing security vulnerabilities with artificial intelligence techniques is critical for patient safety and data privacy. To address these challenges, this study evaluated the performance of machine learning models based on the Walrus Optimization Algorithm (WaOA) using the IoMT-TrafficData, a recent and comprehensive dataset specifically designed for benchmarking intrusion detection in IoMT networks, consisting of IP-based packet data and IP-based flow data. The experiments demonstrated that, for IP-based packet data, the Decision Tree (DT) algorithm achieved 99.89% in binary classification, while the Random Forest (RF) algorithm reached $\mathbf{9 9. 4 8 \%}$ in multiclass classification, as measured by the F1 score. For IP-based flow data, XGBoost achieved a consistent score of $\mathbf{9 9. 9 3 \%}$ across all metrics in binary classification, and a $\mathbf{9 9. 5 8 \% F 1}$ score in multiclass classification. These results indicate that the proposed method provides high accuracy in detecting security threats within IoMT networks. In future research, incorporating more advanced optimization techniques and exploring deep learning models could further enhance the performance and robustness of intrusion detection systems in IoMT environments.