Towards Secure IoMT: Attack Detection Using Deep Q-Learning in Healthcare Networks
Layal Abu Daher · 2023
With the wide advent of IoT systems in the health-care industry, not to mention that medical data is confidential and critical, any alteration in the data could affect the patients’ treatment. Statistics show that cyber-attacks are rapidly increasing, thus an efficient intrusion detection technique is needed for integration in the healthcare sector. In this work, we explore the most frequent threats that target sensitive health data collected by IoMT devices. We introduce reinforcement learning in the context of medical IoT systems, and after presenting relevant literature, we conduct a study on a healthcare dataset and build models that constitute different intrusion detection systems using classical machine learning techniques together with deep reinforcement learning models using Q-learning. We performed extensive simulations based on the constructed models and compared the results using different performance metrics with a level of accuracy exceeding 92%.