Optimized Intrusion Detection for IoMT Networks with Tree-Based Machine Learning and Filter-Based Feature Selection

Area of Science:

• Cybersecurity
• Health Informatics
• Machine Learning

Background:

• The Internet of Medical Things (IoMT) generates vast amounts of health data, necessitating secure transmission and processing.
• IoMT devices have limited storage and computation, increasing vulnerability to security and privacy risks during data transfer.
• Existing intrusion detection systems (IDS) may not be optimized for the unique challenges of IoMT environments.

Purpose of the Study:

• To develop and enhance an efficient intrusion detection system (IDS) specifically for Internet of Medical Things (IoMT) networks.
• To improve the accuracy and efficiency of detecting unauthorized or malicious activities within IoMT devices and networks.
• To address security and privacy concerns arising from data transmission in IoMT systems.

Main Methods:

• Leveraged tree-based machine learning classifiers for intrusion detection.
• Employed filter-based feature selection techniques, including Mutual Information (MI) and XGBoost, to optimize performance and reduce computation costs.
• Utilized a mathematical set intersection to refine feature selection, enhancing detection accuracy and efficiency at the network's edge.

Main Results:

• Achieved a high accuracy of 98.79% on the CICIDS2017 dataset.
• Demonstrated a low false alarm rate (FAR) of 0.007, indicating reliable intrusion detection.
• Successfully detected intruders during data transfer, enabling secure and efficient healthcare data analysis.

Conclusions:

• The proposed IDS effectively enhances the security and privacy of IoMT networks.
• The combination of advanced feature selection and machine learning provides a robust solution for real-time threat detection.
• Future work includes developing a multi-classification approach for attack categorization and evaluating performance in real-world IoMT scenarios.