Detecting Cyber Attacks in Healthcare IoT Systems

Internet of Things (IoT) has brought significant improvements across a number of industries in recent years notably in healthcare. However, it also exhibited some vulnerabilities that have been exploited by various cyber attacks such as the recent breach that compromised smart thermostat data in US hospitals. These incidents represent the tip of the iceberg in terms of the number of cyber attacks that may target IoT healthcare systems. In this paper, we propose an attack detection approach in healthcare IoT systems based on Gradient Recurrent Unit (GRU) classifier using the IoT-Flock dataset. We also experimented with various other deep learning and machine learning models for comparison purposes. We conducted a thorough evaluation by collecting several metrics such as the accuracy, precision, sensitivity, F1-score and Area under the Curve - Receiver Operating Characteristic (AUC-ROC). Our findings show potential success of our approach since it indicated that the GRU model performed best in all classic evaluation metrics. These results are quite encouraging when comparing them with ones from related works, which suggests that our approach holds the promise of revolutionizing the detection of cyber attacks targeting healthcare systems.