Accurately predicting personalized thermal sensation (TS) at a localized level is crucial for the development of human-centric environments and the advancement of personal environmental control systems. Conventional TS models primarily rely on skin temperature, with limited exploration of alternative physiological predictors. This study investigates a novel approach by incorporating the body’s overall heat generation (energy expenditure) and localized heat dissipation (local heat flux) as predictive factors, shifting the focus from thermoregulatory outcomes (skin temperature) to underlying thermophysiological processes.

Using controlled human experiments and data-driven modelling, we evaluated TS predictions across 16 body regions. Our findings indicate that models integrating energy expenditure and heat flux perform comparably to skin temperature-based models (mean absolute error range for different body parts: 0.1–0.55 vs. 0.08–0.44). Notably, skin temperature provided better accuracy in extremities (i.e., hands), likely due to vasomotor complexities, movement artifacts, and the limitations of whole-body energy expenditure data rather than segment-specific measurements. Despite these challenges, the proposed predictor has the potential for better performance. To achieve this, future research should focus on integrating segment-specific energy generation with local dissipation to better capture underlying physiological processes and enhance the accuracy of personalized local TS models.