Combined use of wearable devices and Machine Learning for the measurement of thermal sensation in indoor environments

The measurement of subjects’ thermal sensation and, in particular, the accurate knowledge of this quantity are fundamental to properly design and control living environments, promoting the subject’s well-being and indoor environmental quality. In this study, the authors used wearable sensors (the Empatica E4 wristband and the MUSE headband) to measure the thermal sensation of subjects experiencing different environmental thermal conditions (i.e. hot, neutral, and cold). Results show that the analysis of multimodal physiological parameters provides operational features suitable as input for Machine Learning algorithms able to predict the subject’s thermal sensation with accuracies up to 80%. These results could be further exploited in view of the development of Personal Comfort Models, which can be employed within indoor environmental control systems centered on the occupants’ needs and well-being status, also optimizing the building energy consumption.