Machine learning for adaptive intrusion detection system towards secure internet of things enabled healthcare systems

Abstract

Internet of medical things (IoMT) ties medical sensors, actuators, wearable, and communication technologies to provide an improved healthcare delivery process by means of constant patient monitoring and decision-making. The trend has resulted in an increase in its use because of the increasing prevalence of chronic diseases and aging, and the need to have cost-effective healthcare services. Nonetheless, the networks provided by the IoMT are prone to cyberattacks under the considerable threat as both process huge volumes of sensitive health data, and do not have the resources to do it. In this paper, the author suggests a new adaptive intrusion detection system (IDS) as a way to enhance the security of IoMT-enabled healthcare environments. The proposed model the pre-treats NSL-KDD dataset, filling gaps and transmitting nominal values. Particle swarm optimization (PSO) is used in the optimal selection of features and artificial neural network (ANN), K-nearest neighbors (KNN), and random forest (RF) classifiers are used in intrusion detection. Experimental performance proves that the PSO-ANN model can perform better with a high accuracy of 99.54%, sensitivity of 99.31% and specificity of 97.32% on average with high protection of intelligent healthcare systems.