Stress Detection using Smartwatches with Machine Learning: A Survey

Authors:
Publisher website
Google Scholar
Abstract
In general, stress has become a significant problem in the current lifestyle, where it should be dealt appropriately before it leads to some blunder. To deal with stress, appropriate stress detection techniques should be developed. The different types of data include heart rate variance (HRV), galvanic skin response (GSR); skin temperature, and sleep pattern. Smartwatches contain all the sensors necessary and are to collect data of signals mentioned above. Different machine learning techniques are applied, such as SVM and KNN, to detect the stress level of the particular person. Different machine learning techniques will provide different accuracies. By comparing these accuracies, a better idea can be obtained. This paper analyses different studies to find and compare the accuracies of different signals when using different ML classifiers, such that the signal and classifier are utilized with the best accuracy. It is also checked whether using two or more signals such as HRV and GSR or ST, GSR and sleep pattern will give us a better accuracy on stress detection.

This publication has 13 references indexed in Scilit: