The rapid digitization of healthcare systems has increased the risk of cyberattacks targeting sensitive patient data. Protecting smart healthcare data is critical for maintaining confidentiality, integrity, and availability of patient information. This study presents an AI-based Intrusion Detection System (IDS) specifically designed for smart healthcare environments. The system utilizes a synthetic dataset consisting of physiological parameters such as temperature, heart rate, oxygen level, humidity, and ECG data. The IDS classifies inputs into multiple categories, including normal and various cyberattacks like Port Scanning, DDoS, TCP Vulnerability Scans, MITM, Botnet attacks, Data Tampering, and Malware Injection. The proposed framework integrates deep learning and ensemble methods, combining Convolutional Neural Networks (CNN) with XGBoost and a Stacking Classifier to enhance detection accuracy. The CNN captures temporal patterns from vital signals, while XGBoost efficiently handles tabular sensor data. The Stacking Classifier merges the predictions, improving overall performance. A Flask-based back-end handles data processing and inference, while a front-end built with HTML, CSS, and JavaScript provides an interactive interface for healthcare personnel. Experimental evaluation demonstrates the system’s ability to identify diverse attack types accurately. The model also incorporates explainability techniques to provide transparency in decision-making. This IDS framework offers a scalable and robust solution for safeguarding smart healthcare infrastructures against sophisticated cyber threats. The research contributes to improving cybersecurity in medical environments while maintaining data integrity and patient safety.

Keywords

Intrusion Detection, Smart Healthcare, Ensemble Learning, CNN, XGBoost, Stacking Classifier, Cybersecurity, ECG Analysis, Data Tampering, Malware Detection