Explainable Quantum Machine Learning for IoMT Security: Evaluating Quantum Advantage
Vincenzo Spadari, Alfredo Nascita, Idio Guarino +2 more
Network security is an even felt theme, due to the continuing evolution of the Internet. In particular, Internet of Medical Things (IoMT) contexts are sensitive areas, as cyber threats can compromise patient safety. In this paper, we investigate Quantum Machine Learning (QML) applied to a IoMT scenario. QML approaches are inherently complex due to the mathematical principles that guide their operations. Hence, as understanding the reasoning behind these algorithms within such environments is decisive, we adopt eXplainable Artificial Intelligence (XAI) techniques to shed light on models’ decisions. Specifically, we consider two base approaches, namely AMPe and ANGe, and explore some data-encoding strategies, namely data re-uploading and chunking. We compare this with a classical Deep Learning (DL) baseline, i.e., CNN+LSTM, examining the impact of the models on CIC-IoMT2024 dataset in different scenarios (Binary, Cat, and Attack). We investigate the behavior of the aforementioned models through (i) an experimental evaluation by varying the training set proportion, (ii) a time-to-insight analysis for time-constrained setup robustness, and (iii) XAI methods for details on how models produce their outcomes. Our findings provide a systematic assessment of the current limits of quantum advantage in IoMT security.
IEEE International Conference on Smart Computing Workshops and Other Affiliated Events (SMARTCOMP Companion)•Jun 1, 2026•0 citations