A novel methodology for human thermal comfort decoding via physiological signals measurement and analysis

Personal comfort models (PCM) represent the most promising paradigm for human-centric thermal comfort in buildings. Several data sources can be used to generate a PCM: environmental data, physiological data, occupants' response. Advances in wearable sensing suggest that the use of physiological data for real time comfort measurement can be the start-up of the next generation of building design and operation with PCMs. However, proof of evidence about the adoption of non-invasive but accurate measurement methods and about correlations between physiological features and thermal sensation, are still required. This study presents the results from a large original experimental campaign aiming at human thermal comfort decoding via physiological signal. Two non-invasive wearables were used to simultaneously measure four key physiological signals (electroencephalography (EEG), Heart Rate Variability (HRV), electrodermal activity (EDA) and skin temperature (ST) on 52 subjects exposed to three different thermal conditions (namely cold, warm, and neutral) in a controlled environment. Data acquired from 219 tests were therefore analysed to determine the statistical importance of physiological features. Results showed that cold and warm thermal sensations can be uniquely identified by each physiological signal; while neutral sensation is the less distinguishable. More specifically, statistical differences (p-value <0.01) between cold and warm conditions were detected for the first time among EEGs features (Beta TP10, Gamma TP10 relative alpha TP9), time- and frequency-domain features of HRV, EDA tonic component and mean ST. Experimental results finally demonstrated that physiological measurements can identify specific thermal sensation, of crucial importance for the most advanced PCMs and for disclosing novel energy saving opportunities, accounting for people's diversities.