Quantifying the Impact of Urban Air Quality on Individual Well-being: A Multimodal Approach for Stress Classification

Urban air pollution contributes to respiratory diseases, cardiovascular conditions, and mental stress. This study explores the relationship between individual exposure to air pollution and well-being, by integrating quantitative and qualitative methods, focusing on psychological stress levels through multimodal data collection. It integrates real-time physiological measurements (e.g., heart rate), environmental sensors (e.g., PM2.5 sensor), and self-reported psychological stress to develop a classification framework for stress episodes based on exposure patterns. The study employs Geographic Information Systems (GIS), Pearson’s R correlation, and machine learning techniques such as Random Forest and k-Nearest Neighbors (kNN) to analyze spatial-temporal exposure data and assess the correlations between urban air quality and individual well-being. Stationary air quality monitoring nodes provide confidence in the measured data through the wearable air quality monitor. Findings show that environmental and physiological data can classify momentary psychological stress, showing 98.56% accuracy with Random Forest model, which can be built on top of emotion analysis for expansion of well-being classification. This study fills research gaps by combining real-time data collection and machine learning to quantify the complex relationship between urban air quality and individual well-being.