Intrusion Detection in Healthcare 4.0 Internet of Things Systems via Metaheuristics Optimized Machine Learning

Nikola Savanović, Ana Toskovic, Aleksandar Petrovic, Miodrag Zivkovic, Robertas Damaševičius, Luka Jovanovic, Nebojsa Bacanin, Bosko Nikolic

Rapid developments in Internet of Things (IoT) systems have led to a wide integration of such systems into everyday life. Systems for active real-time monitoring are especially useful in areas where rapid action can have a significant impact on outcomes such as healthcare. However, a major challenge persists within IoT that limit wider integration. Sustainable healthcare supported by the IoT must provide organized healthcare to the population, without compromising the environment. Security plays a major role in the sustainability of IoT systems, therefore detecting and taking timely action is one step in overcoming the sustainability challenges. This work tackles security challenges head-on through the use of machine learning algorithms optimized via a modified Firefly algorithm for detecting security issues in IoT devices used for Healthcare 4.0. Metaheuristic solutions have contributed to sustainability in various areas as they can solve nondeterministic polynomial time-hard problem (NP-hard) problems in realistic time and with accuracy which are paramount for sustainable systems in any sector and especially in healthcare. Experiments on a synthetic dataset generated by an advanced configuration tool for IoT structures are performed. Also, multiple well-known machine learning models were used and optimized by introducing modified firefly metaheuristics. The best models have been subjected to SHapley Additive exPlanations (SHAP) analysis to determine the factors that contribute to occurring issues. Conclusions from all the performed testing and comparisons indicate significant improvements in the formulated problem.