Abstract
Human physiology is known to react to various environmental stimuli over different time frames. Prolonged exposure to elements such as heat, air pollution, and volatile organic compounds negatively affects health, as established in previous research. Our earlier work demonstrated that autonomic responses of the human body, recorded through biometric sensors on a single individual, could empirically predict levels of inhalable particulate matter in their immediate environment. This current study extends this finding to observations from multiple participants. Subjects cycled on stationary bikes outdoors, equipped with a range of biometric sensors, while environmental sensors simultaneously captured data on their surroundings. Using this expanded data set, machine learning models achieved a high degree of accuracy (R2=0.97) in predicting concentrations of particulate matter (PM2.5) using a few readily available biometric features, including skin temperature, heart rate, and respiration rate. This research underscores the importance of physiological responses as markers of exposure to particulate matter, laying the foundation for the use of biometric data in environmental health surveillance and real-time pollution assessment.