A Random Forest model using wearable sensor data outperformed other machine learning techniques for stress detection, achieving F1-scores of 83.34 for binary and 65.73 for three-class classification.
A Random Forest machine learning model using multimodal wearable sensor data can effectively detect stress states.
Mental states like stress, depression, and anxiety have become a huge problem in our modern society. The main objective of this work is to detect stress among people, using Machine Learning approaches with the final aim of improving their quality of life. We propose various Machine Learning models for the detection of stress on individuals using a publicly available multimodal dataset, WESAD. Sensor data including electrocardiogram (ECG), body temperature (TEMP), respiration (RESP), electromyogram (EMG), and electrodermal activity (EDA) are taken for three physiological conditions - neutral (baseline), stress and amusement. The F1-score and accuracy for three-class (amusement vs. baseline vs. stress) and binary (stress vs. non-stress) classifications were computed and compared using machine learning techniques like k-NN, Linear Discriminant Analysis, Random Forest, AdaBoost, and Support Vector Machine. For both binary classification and three-class classification, the Random Forest model outperformed other models with F1-scores of 83.34 and 65.73 respectively.
Garg et al. (Tue,) conducted a other in Stress. Machine Learning models (Random Forest) vs. Other ML models (k-NN, LDA, AdaBoost, SVM) was evaluated on F1-score and accuracy for three-class (amusement vs. baseline vs. stress) and binary (stress vs. non-stress) classifications. A Random Forest model using wearable sensor data outperformed other machine learning techniques for stress detection, achieving F1-scores of 83.34 for binary and 65.73 for three-class classification.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: