A Multimodal Approach to Predicting Toe Temperature: Experimental, CFD, and LSTM-Based Methods

The assessment of footwear insulation in cold environments remains a challenge due to the lack of standardized evaluation methods beyond the pass/fail criterion of ISO 20344:2021. To address this gap, this study develops a multi-approach framework combining experimental measurements, computational simulations, and machine learning predictions to evaluate thermal protection in cold-weather footwear. Human trials were conducted in a controlled climatic chamber to examine thermoregulatory responses under varying environmental conditions, insulation levels, and activity intensities. To complement these measurements, computational simulations were performed to estimate the thermal resistance (RcT) of footwear under ISO 15831:2004 conditions. Additionally, a Long Short-Term Memory (LSTM) neural networkwas trained on experimental data to predict big toe temperature based on skin temperature, ambient conditions, and activity levels. This integrated approach enables a more comprehensive evaluation of footwear thermal performance, providing valuable insights for footwear manufacturers and researchers.