Exploring an occupant-involved closed-loop wearable sensing system and online tuning for individualized thermal preference

This study introduces a novel, human-in-the-loop multimodal sensing system and platform, designed for the data collection and modeling of individualized thermal comfort. We investigated whether incorporating alert-based wearable sensing and online threshold-tuning functions can enhance human interaction with the indoor environment, thereby improving the efficiency and effectiveness of data gathering for predictive modeling. The research findings indicate that the proposed method significantly reduces the number of sampling points needed to achieve equivalent overall accuracy in predictive models by monitoring the physiological and environmental inputs to the system. Likewise, for the same input data quantity, the cross-validation accuracy of the optimized models outperformed that of the baseline model. This system decreases the user's input requirements and boosts autonomous data collection and modeling on an individual basis for personal comfort modeling purposes, which can be also incorporated into long-term indoor environment monitoring and smart building paradigms.