Beyond group-level comparisons: machine learning-based clustering identifies sex-specific heat acclimation phenotypes and response patterns

Heat acclimation (HA) responses show substantial individual variability, yet traditional analyses rely on group-level comparisons that may mask important individual differences. Purpose: This study aimed to identify distinct HA phenotypes using machine learning-based cluster analysis of comprehensive physiological data from heat tolerance tests (HTTs). We hypothesized that multivariate analysis would reveal sex-specific phenotypes despite similar overall adaptation potential. Methods: 40 healthy participants without history of EHI (19M/21F, mean±SD, 23±4 years old) completed two HTTs before (PreHA) and after (PostHA) five days of repeated exercise in (40°C, 40% relative humidity) categorized as isothermal (exercise intensity adjusted to maintain Trec at 38.5-39.5 °C for 60 min) exercise-HA. HTT1: treadmill walking (120 min at 5 km·h -1 , 2% grade), HTT2: treadmill running at 65% v[Formula: see text]2max (5.6 ± 0.76 mph) until reaching T rec 39.5°C or volitional fatigue. Systematic feature selection through correlation and principal component analysis reduced 114 physiological variables to 12 key features. Results: Initial clustering revealed strong sex-based separation primarily driven by whole-body sweat loss (WBSL: p<0.001; WBSL BSA : p<0.001). Male participants clustered into two distinct phenotypes differentiated by sweat response efficiency (WBSL BSA : p<0.001) and baseline rectal temperature (T rec , p<0.001). Males consistently demonstrated two phenotypes across tests, with significant separation in thermal comfort and fatigue responses (END_Thermal p<0.001, END_Fatigue p<0.001). Female participants exhibited three distinct phenotypes, primarily separated by humidity tolerance (p<0.001) and sweat responses (WBSL BSA : p<0.001). Females showed a clear bimodal distribution with the predominant phenotype (n=16) characterized by higher peak heart rates (~190-200 bpm) and increased T rec rate of rise (~0.07 °C·min -1 ), while the smaller Cluster 1 (n=5) showing superior heat tolerance with reduced skin temperature and enhanced cardiovascular stability. Temporal analysis revealed greater phenotype transitions in males from Poor Acclimation at baseline (PreHA: n=11) to Well Acclimated post-intervention (PostHA: n=13), while females maintained more stable classifications throughout. These findings suggest that while standardized HA protocols may induce similar performance improvements, the underlying physiological adaptations follow distinct sex-specific patterns, highlighting the need for more nuanced approaches to heat tolerance assessment and acclimation protocols. DOD BA200299 This abstract was presented at the American Physiology Summit 2025 and is only available in HTML format. There is no downloadable file or PDF version. The Physiology editorial board was not involved in the peer review process.