Intelligent Intrusion Detection System Using Improved Osprey Optimization and Stacked Ensemble Learning for IoT ‐Based Healthcare Systems
Ashok Kumar, Rahul Gupta, Sunil Kumar, Kamlesh Dutta, Rajender Kumar · Security and Privacy · 2025
ABSTRACT Internet of Things (IoT)‐enabled smart healthcare systems aim to improve patient care by providing advanced treatments at home, in hospitals, and in remote areas. However, these systems are the prime target of cyber attackers due to the sensitivity of the patient data. According to IBM's 2022 report, the healthcare sector has incurred the highest data breach costs among all industries for 12 consecutive years. These cyber‐threats pose severe risks to human lives by disrupting critical medical services and compromising sensitive health data. To address these security concerns, this paper proposes an intelligent intrusion detection system that integrates the Improved Osprey Optimization Algorithm (IOOA) with a stacked ensemble learning approach. The proposed system is named the Improved Osprey Optimization Algorithm and Stacked Ensemble Learning‐based Intrusion Detection System (IOS‐IDS). IOOA enhances the standard Osprey Optimization Algorithm by integrating Sobol sequences, adaptive weighting, Weibull distribution, and Lévy flight for faster and more efficient convergence. The proposed IOS‐IDS system operates in two stages: first, IOOA is applied for efficient feature selection, achieving faster convergence compared to conventional optimization techniques; second, a stacked ensemble learning model optimized by IOOA classifies the malicious and benign activities efficiently. To ensure robustness and generalizability, the performance of IOS‐IDS is evaluated on three widely recognized datasets: UNSW‐NB15, WUSTL‐EHMS‐2020, and TON‐IoT. The evaluation is performed using standard metrics, including accuracy, precision, recall, F1‐score, ROC score, memory (inference), delay (per sample), and inference complexity. The experimental results highlight the feasibility and effectiveness of the proposed intrusion detection system in the resource‐constrained environment of an IoT‐enabled smart healthcare system.