Artificial Intelligence and IoT for E-Health Discovery and Management

Abstract

The burgeoning landscape of the Internet of Medical Things (IoMT) presents a critical and escalating challenge to cybersecurity. Traditional Intrusion Detection Systems (IDS) are demonstrably inadequate for this domain, as they are unable to effectively manage the unique characteristics of IoMT network traffic, including its high dimensionality, temporal dependencies, and inherent heterogeneity. This research addresses this deficiency by introducing a novel hybrid deep learning framework, built upon a synergistic integration of a Bidirectional Long Short-Term Memory (BiLSTM) network with a Deep Neural Network (DNN).This innovative architecture is meticulously designed to overcome the limitations of existing models by concurrently capturing both the temporal sequences and spatial features embedded within IoMT network traffic. A core contribution of this work is the framework’s intrinsic transparency. It moves beyond the opaqueness of traditional "black box" models by incorporating Explainable AI (XAI) principles, leveraging SHapley Additive exPlanations (SHAP). This integration provides human-interpretable justifications for every security alert, revealing the specific features and data patterns that drive the model’s predictions. This not only cultivates trust and ensures accountability but also facilitates an effective human-in-the-loop analysis, empowering security personnel to confidently validate alerts and respond without compromising patient care. Rigorous evaluation on the comprehensive CICIoMT2024 benchmark dataset validates the framework’s exceptional performance, demonstrating high accuracy in discriminating between 18 distinct attack types and normal traffic. By seamlessly combining advanced detection capabilities with essential operational transparency, this research marks a significant advancement in both cybersecurity and trustworthy AI, contributing a robust and auditable solution vital for safeguarding sensitive healthcare data and ensuring the resilient operation of critical medical services.