Learning Models on Physiological and Behavioral Data

Abstract

The rising incidence of stress-related mental health issues, particularly among college students, has highlighted the need for effective, real-time detection methods. Traditional self-reported assessments are subjective and often unreliable. In this study, we present a data-driven approach using machine learning (ML) models to predict stress levels from physiological and behavioral indicators such as heart rate variability, skin conductance, and sleep patterns. We evaluate and compare the performance of multiple ML algorithms—Support Vector Machines (SVM), Logistic Regression (LR), Random Forest (RF), K-Nearest Neighbors (KNN), and Convolutional Neural Networks (CNN). Among these, Random Forest achieved the highest accuracy of 92%, followed by CNN at 90%, demonstrating strong precision (91% and 89%, respectively) and F1-scores (92% and 94%). These results affirm the potential of AI-powered stress monitoring systems for early mental health intervention, particularly when integrated with wearable technologies.