Smart Computing- Book ISBN- 978-81-953658-1-4

Authors:-  Dércio Anselmo Bobo, Ankita Gandhi, Jay Gandhi

Abstract :-On a daily basis, people can experience a moment of stress for different reasons. It can become dangerous to mental health when this moment of stress is too frequent. Previous papers have laboratory environment, which can lead to bias output because of the ground truth. This paper studies the possibility of detecting if a person is under stress or no, using machine learning classification algorithms and an unobtrusive wearable device where the person will use it on a daily basis, and the data will be collected and sent to our system. We can use various physiological signals to detect stress, such as Electrocardiogram (ECG), Galvanic Skin Response (GSR), and Blood Volume (BV). This physiological signal is used as input in different machine learning techniques to measure a person's stress. Most papers show that support vector machines have the best average classification. Applying the support vector machine, KNN and random forest were able to compare the three algorithms, and the SVM recorded the highest accuracy.indicated that we can use physiological signals to detect if a person is under stress or not. However, researchers have conducted experiments in the traditional method in a 

Keywords: - Stress Detection, Machine Learning (ML), Human-computer Interaction (HCI), Wearable Devices, Physiological Signals