Smart Healthcare: Machine Learning Enabled WBAN for Early Detection of Chronic Diseases
S. Kumaran, I. Riya Evangeline Princy, J. Princy Agnes · 2024
Chronic illness diagnosis is delayed by the reactive strategies used by existing healthcare systems. This research study proposes a novel Machine Learning (ML) technique for early chronic illness diagnosis and proactive monitoring within a Wearable Body Area Network (WBAN). Unlike the existing system, the proposed system continuously monitors physiological markers in real-time by using WBAN. The proposed system facilitates early diagnosis by analysing data for small variations in usual health patterns. The proposed system integrates data collection, pre-processing, and ML techniques including Support Vector Machines (SVM) and Random Forests (RF), including Convolutional Neural Networks (CNNs), and real-time anomaly detection. The proposed model has resulted in 95% accuracy, 90% sensitivity, 97% specificity, and 97% AUC, together with better performance, fewer false positive rates and more computing efficiency when compared with the existing system. Reduced false alarms and facilitated quick actions are two ways with which the proposed system improves patient outcomes and lowers medical costs. With such a novel strategy, proactive health care interventions and improved patient care can be implemented.