Thermal comfort prediction of building occupants can be remarkably instrumental for both ensuring a comfortable living environment, and improving building energy-efficiency through strategic ambient-control based on the occupant's thermal state index (TSI: Cool-Discomfort/Comfort/Warm-Discomfort). This paper proposes a personal TSI prediction method termed as the enhanced Predicted Thermal State (ePTS) method by sensing physiological parameters namely, hand skin temperature and pulse rate, along with the ambient air temperature. The ePTS method is developed through significant enhancements over our previous skin temperature based PTS model. First, we investigate pulse rate as a potential TSI predictor. Second, we propose a logic-gated normalization process to address the individual differences in pulse rate. Third, we develop a spectral analysis approach to formulate a new pulse rate feature. Consequently, six significant TSI predictive features are curated that map the TSI based on an optimized Support Vector Machine algorithm. The method is compared with 5 different input cases and 3 state-of-the-art methods. Pulse rate is revealed to be a significant TSI predictor subject to the occupant's gender and BMI. The ePTS method achieves the highest accuracy at over 97%, outperforming the PTS model (82%), and other physiology based methods (82%–94%). The study is substantiated with extensive real human experiments in regulated thermal environment. The method is deemed adaptive by its machine-learning framework, robust by normalization, and non-intrusive by convenience of measurement. It may be realized by integrating the wearable and ambient sensor networks with the Internet-of-Things for a comfortable and energy-efficient building. Display omitted •Novel TSI prediction method using skin temperature, pulse rate and air temperature.•Novel normalization process addresses individual differences in occupant pulse rate.•Spectral analysis method extracts new pulse rate feature for modeling.•SVM based ePTS method at 97% accuracy outperforms 5 input cases & 3 existing methods.•Occupants' thermal comfort prediction can improve building energy efficiency.