Advancing Healthcare Monitoring: Integrating Machine Learning With Innovative Wearable and Wireless Systems for Comprehensive Patient Care
Malik Muhammad Arslan, Xiaodong Yang, Zhiya Zhang, Saeed Ur Rahman, Muneeb Ullah, Qammer H. Abbasi
Abstract
This article introduces an advanced vital sign monitoring system that seamlessly integrates sensor technology, wireless communication, and machine learning (ML) algorithms to enhance patient care. The system accurately records and interprets vital physiological metrics such as oxygen saturation (SpO2), heart rate (HR), and body temperature. The integration forms the foundation of our comprehensive approach to patient health status classification through the ML algorithm. We employed three distinct ML algorithms—K-nearest neighbors (KNNs), support vector machines (SVMs), and stochastic gradient descent (SGD) models to analyze health states, each precisely trained and validated using data collected from sensors on vital sign. This meticulous training aims to enable the models to accurately identify various patient health states, with the KNN model achieving an average accuracy of 92%, while the SVM and SGD models attained even higher accuracies of 94% and 97%, respectively. These outcomes were rigorously validated against established health scoring systems, showcasing the models’ exceptional performance in multiclass classification scenarios. This is further evidenced by their area under the receiver operating characteristic (AUROC) scores: KNN (0.9021), SVM (0.9841), and SGD (0.9933). The integration of these technologies into our health monitoring system represents a significant advancement in the continuous and comprehensive assessment of patient health, demonstrating the potential of modern technologies to revolutionize healthcare monitoring and management.