Heat stress, cognitive performance, and temporal metabolomic changes: Evidence from a prospective panel study

Heat stress (HS) may impair cognitive function, yet evidence from real-world settings involving combined environmental and exertional exposure remains limited, and the underlying mechanisms are unclear. We conducted a prospective panel study with repeated measurements among 46 first-year university students during an outdoor military training program in Guangzhou, China. Environmental HS was assessed using the Wet-Bulb Globe Temperature (WBGT), while exertional HS was characterized using individual-level physical activity (PA). Cognitive performance was evaluated using forward/backward total scores and correct reaction times in the Corsi block-tapping test (CBTT). Tympanic temperature, blood pressure, heart rate, and repeated serum samples for untargeted metabolomics were collected. Linear mixed-effects models assessed associations, with exploratory mediation analyses evaluating potential indirect associations through physiological responses. Across 396 daytime observations, higher WBGT was associated with lower forward and backward total scores and longer backward correct reaction times across the 10-, 20-, and 30-minute exposure windows after adjustment for step rate measured during the corresponding exposure window and false discovery rate (FDR) correction. The association between WBGT and backward correct reaction time was weaker during Week 2 than during Week 1, whereas formal interaction analyses did not support week related differences for forward correct reaction time or either total score outcome. Step rate was not significantly associated with any cognitive outcome, and no statistically significant interaction between step rate and WBGT on cognitive outcomes was observed. Exploratory mediation findings suggested an indirect association between WBGT and backward total score through tympanic temperature, which did not remain statistically significant after FDR correction. Metabolomics analysis indicated a temporal shift from amino acid-related perturbations in Week 1 to more prominent lipid metabolic alterations in Week 2. Overall, higher WBGT was associated with poorer cognitive performance, with outcome-specific temporal heterogeneity across training weeks.