Abstract

The increasing reliance on IoT-driven healthcare systems has revolutionized patient care but also introduced significant cybersecurity challenges, with threats to data confidentiality, system integrity, and patient safety. To address these challenges, this study proposes a novel framework that integrates an Autoencoder for dimensionality reduction and feature extraction with ensemble methods such as Random Forest (bagging) and XGBoost (boosting) for robust and precise threat detection. By leveraging PCA for preprocessing, SMOTE for handling imbalanced data, and advanced feature engineering, the framework ensures scalability and adaptability for real-time threat mitigation. The Autoencoder extracts meaningful latent features, which enhance the robustness of Random Forest and the precision of XGBoost, creating a synergistic approach that significantly outperforms traditional methods. Achieving a perfect classification accuracy of 100%, this innovative model demonstrates exceptional performance in identifying normal and attack patterns, setting a new benchmark for securing IoT healthcare systems against evolving cybersecurity threats.

Authors

Mohammed Ismail¹, A. Ramesh Babu²
Auroras PG College, India¹, Chaitanya-Deemed to be University, India²

Keywords

IoT Healthcare, Autoencoder, Ensemble Learning, Random Forest, XGBoost, Threat Detection