Investigation of physiological signals associated with the comfort-to-warm discomfort transition during seat heater operation under cold cabin conditions

Personal heating devices (PHDs) can enhance comfort in cold vehicle cabins. While previous research has emphasized their comfort and energy-saving benefits, limited attention has been given to PHD use after occupants have reached a comfortable state. In particular, it remains unclear whether physiological signals can indicate when occupants no longer prefer continued heating. This study investigated whether heart rate variability (HRV) features and wrist skin temperature contain information associated with the transition from comfort to warm discomfort during seat heater operation. Twelve participants were exposed to cabin temperatures of 16, 13, and 10 °C. Comfort and warm discomfort were identified from subjective ratings collected at 5-min intervals, and corresponding physiological features were analyzed. Preference for continued seat heating exceeded 85% during comfort but declined to 2.9%, 10.5%, and 0% during warm discomfort at 16, 13, and 10 °C, respectively. No significant differences were observed in individual physiological features between the two states. However, repeated random-split machine-learning analyses suggested modest transition-related discriminatory information under subject-overlapping evaluation. The best performance was obtained using combined HRV time-domain features and wrist skin temperature (G-5): macro F1 scores of 0.714, 0.583, and 0.742 at 16, 13, and 10 °C, respectively. Held-out-subject performance was lower across all feature groups, with the best macro F1 per condition ranging from 0.471 to 0.653. These findings do not demonstrate strong physiological separability, stable cross-subject discrimination, or reliable subject-independent prediction but suggest that the examined features may have exploratory value for future personalized or subject-adaptive applications.