A novel theoretical model for predicting the individuals? thermal sensations based on air temperature and biomarkers measured by wearable devices

Optimal control of indoor thermal environments based on real-time monitoring of occupants' feeling of warmth (a.k.a. thermal sensation) is one of the effective strategies to improve building energy efficiency and occupant comfort. Accordingly, developing an individual (or personal) thermal comfort model based on continuous observation of biomarkers (often measured by wearable devices) has received much research attention in recent years. Whilst almost all the comfort prediction models presented in the literature are based on a data-driven approach and machine learning techniques, this paper presents a novel theoretical model aimed at explaining the causal relationship in the prediction of thermal sensations based on key environmental and physiological parameters, incl. skin temperature, heart rate, and air temperature. The proposed model is grounded on the ASHRAE thermal sensation index (TSENS) and is based on mediating effects of thermoregulatory and cardiac responses. The new model's performance was evaluated under different thermal environmental conditions by comparing it against the PMV values and the experimental data available from the literature. Results showed that the proposed model could accurately predict individual thermal sensations by monitoring the air temperature, overall or wrist skin temperature, and heart rate (or metabolic rate). This model can provide new insight into individual thermal comfort assessment and can be used alone or in combination with data-driven machine learning models for continuous monitoring of thermal comfort conditions.

Automated summary

This paper presents a novel theoretical model for predicting thermal sensations based on key environmental and physiological parameters. The proposed model, grounded on the ASHRAE thermal sensation index (TSENS), offers insight into the causal relationship between thermal sensations and factors such as skin temperature, heart rate, and air temperature. The performance of the model was evaluated under various thermal environmental conditions and showed accurate predictions of individual thermal sensations. It can be used for individual thermal comfort assessment and in combination with data-driven machine learning models for continuous monitoring of thermal comfort conditions.