Classification of Stress using Machine Learning Based on Physiological and Psychological Data from Wearables

Stress is a common psychological issue in today's world, causing individuals to experience physiological and behavioral changes. When someone experiences stress for a prolonged period, it can have an impact on their body and thoughts, leading to severe behaviors. Stress-related problems can be prevented if identified early. Nowadays, researchers' use of machine learning in early disease detection is increasing, offering a solution to stress-related issues. This study focuses on developing a machine learning predictive model for classifying stress levels using physiological data. Physiological data is obtained from an open-source dataset containing biosignal data such as Electrodermal (EDA), Skin Temperature, Heart Rate (HR), Blood Volume Pulse (BVP), and others. The dataset is then used to train a machine learning model using Support Vector Machine (SVM), Random Forest (RF), Adaboost, and Bootstrapping algorithms. Accuracy, precision, recall, and F1-score are used to evaluate the model's performance. As a result, the predictive model using the Bootstrapping algorithm achieves the best performance with an accuracy of 95.82%, a precision of 99.33%, a recall of 98.02%, and F1-score of 94.69% compared to other machine learning classifiers. Creating such models can aid in early stress detection once they are integrated into a wearable device for continuous monitoring purposes.