Intrusion detection system for IoMT in smart hospitals: A systematic literature review

With the quick development of smart healthcare, the Internet of Medical Things (IoMT) has emerged as a paradigm-changing technology. It connects wearable devices, smart sensors, and medical systems to enable real-time, patient-centric care. To protect these life-saving systems, their security is essential and often requires the use of specific Intrusion Detection Systems (IDS). However, such solutions have yet to address complex medical networks. A systematic literature review is conducted to reveal the current state and future trends in the IDS framework for IoMT. The 32 studies identified in prominent digital databases were selected for in-depth qualitative and quantitative analysis. The total number of datasets used in the selected studies is 11; some of these achieve high accuracy (98%-99.9%). The frequently used models are XG Boost (XGB), K-Nearest Neighbors (KNN), Artificial Neural Network (ANN), and Convolutional Neural Network. (CNN). IDS with AI methods, primarily ML, DL, and hybrid models, are very much popular. High accuracy and low latency are the priorities of such systems. However, most are poorly adapted to IoMT energy and resource constraints. This review aims to bridge the gap between algorithmic innovations and clinical utilities. Numerous frameworks also overlook privacy, real-time implementation, and regulatory compliance. We advocate for context-aware systems that center on patient safety.