X-NET: Explainable AI-Based Network Data Security Framework for Healthcare 4.0

Abstract

Remote patient monitoring aspect of Healthcare 4.0 has revolutionized the way we pursue medical treatment. Integration of Medical Internet of Things (MIoT), patient centric healthcare devices and remote robotic equipment has vanished the separative line between physical onsite monitoring and remote monitoring. Secure use of available network to handle this critical patient data is one of the major concern raised due to this methodology. To mitigate this paramount concern Machine Learning (ML) algorithms can be utilized to detect in malicious activity or manipulation occurred on incoming data. Though use of ML algorithms along with conventional feature extraction technique brings performance saturation challenge and creates a black-box in this highly complex communication environment. To overcome this drawback, we propose X-NET, an eXplainable Artificial Intelligent (X-AI) based network data security approach for Healthcare 4.0 applications. For comparison purpose we have used five different types of tradition feature extraction techniques along with Naive Bayes (NB), Logistic Regression (LR) and Perception. Utilization of X-AI algorithms such as LIME and SHAP has significantly boosted the overall performance and reliability of X-NET. Performance evolution of our proposed approach has been done through various metrics such as accuracy, precision, recall, f1-score, Receiver Operating Characteristic and Precision-Recall Trade-off. Thus, X-NET contributes in defining secured, reliable and robust remote patient monitoring scheme for Healthcare 4.0.