Machine Learning Enhanced Cybersecurity Framework for Protecting Health Care Data

Abstract

The study advances hospital network security by its hybrid machine learning system that identifies cyber-attacks. The system applies K-best feature selection and XGBoost and LSTM and Stacking Classifier to attain high detection accuracy and effective intrusion detection. The system employs Explainable AI approaches (SHAP, LIME) for transparent decision-making and TPOT for hyperparameter optimization. The system identifies various types of threats that include ARP Spoofing and MQTT-based DDoS/DoS attacks and Reconnaissance and TCP/IP attacks. The system offers better accuracy and real-time detection capabilities along with increased interpretability that guards sensitive patient data effectively.