Recognition and prediction of elderly thermal sensation based on outdoor facial skin temperature

The global ageing issues and climate extremes are becoming increasingly prevalent. Increasingly, elderly people may have difficulty expressing their feelings or needs for outdoor thermal comfort accurately and promptly. Therefore, it is crucial to find a reliable method for them to evaluate their thermal comfort. This study aims to explore the use of infrared images for this purpose. Experiments were conducted outdoors in the Home for the Aged Guangzhou in summer through thermal environment measurements and questionnaire surveys. The results show that the elderly are unresponsive to hot environments. The preference for wind speed changes was significant, compared to air temperature, relative humidity and solar radiation. Experiments were conducted to construct six machine learning models by inputting the local skin temperature (forehead, eye, nose, cheek and chin) of the face of an elderly person and Thermal Sensation Vote as the output parameter. The experimental results showed that the facial skin temperature of an individual can be used as an indicator of their thermal sensation. And the best performing model is Random Forest, an area under the curve value of 0.889 was achieved. This paper also discusses the selection of measurement site. The nose, a key area on the face, plays a crucial role in operating the proposed method. The results of this work may provide a theoretical basis for the dynamic monitoring of thermal sensations using facial skin temperature, which may contribute to the development of useful strategies for improving the thermal comfort of elderly populations in harsh environments.