Wearable Biosensors for Continuous Monitoring of Physiological Signals: A Review and Future Perspectives
Abstract
Together with a wide range of smart devices, big data, cloud services, wireless connectivity, and small electronics and intelligent sensors, these developments have not just made wearable biosensors possible but also extended human lifespans. This is accomplished through a variety of apps, such as those that monitor physiological signals, provide personalized medical treatment, and offer suggestions for enhancing user expertise. Nevertheless, several issues pertaining to low energy usage, accuracy, ease, and dependability have constrained the use of wearable biosensors in medical settings to monitor physiological signals. This study introduced a new wearable sensor that measures physiological signals continuously: blood pressure (BP), heart rate (HR), and electrodermal activity (EDA). The design of the suggested sensor is explained in depth, and then the assessment that was conducted using a biosensor on participants to collect data is discussed. The stress categorization technique reduces feature collection redundancy by utilizing mutual data and genetic algorithms in a coupled manner. To additionally adjust machine learning hyperparameters, Bayesian optimization is used. The findings show that, when utilizing the NB method for 2-stage and 3-stage stress categorization, a blend of EDA, BP, and HR produces the maximum categorization accuracy of 98.3% and 97.1%, respectively. However, for 2-stage and 3-stage stress categorization, EDA and HR alone exhibit similar accuracy of 97.1% and 95.3%, respectively. Additionally, the research reveals proof that stress rates are lowered when engaging in meditative audio. These results demonstrate ways wearable sensors and ML may be used to detect and regulate stress in educational settings in real time.