Heart rate variability as an indicator of fatigue: A structural equation model approach.

• HRV indices predict multidimensional driving-related fatigue. • Lower RMSSD associated with greater fatigue in young drivers. • Higher LF/HF ratio linked to higher fatigue levels. • Findings support HRV metrics as fatigue biomarkers. • Applied research needed to validate real-world use. Driver fatigue significantly impairs performance and is a major risk factor for road crashes. However, fatigue is difficult to objectively measure and quantify. This study aimed to elucidate associations between heart rate variability (HRV) metrics and multidimensional fatigue construct encompassing objective and subjective measures in young drivers with self-reported short sleep. Eighty-four young adults underwent assessments during simulated driving. HRV indices RMSSD and LF/HF ratio were derived from an electrocardiogram, while relative Theta power, oculography drowsiness, lane position variability, and Karolinska Sleepiness Score measured fatigue. RMSSD negatively correlated with Theta power, and LF/HF ratio positively correlated with position variability. Exploratory factor analysis extracted three factors, with RMSSD and LF/HF loading onto one. Structural equation modelling tested the prediction of a latent fatigue construct from HRV. Two models demonstrated an acceptable fit. RMSSD negatively predicted fatigue, explaining 6.4 % of the variance, while the LF/HF ratio positively predicted fatigue, accounting for 3.7 % of the variance. In summary, HRV, particularly RMSSD, showed significant relationships with multidimensional fatigue. Lower RMSSD and higher LF/HF ratio are associated with greater fatigue levels. This study demonstrates that HRV indices exhibit significant relationships within a multidimensional model of fatigue incorporating objective performance, ocular, and subjective measures. The findings provide initial evidence towards using HRV for monitoring complex manifestations of fatigue, rather than single dimensions. Further applied research should investigate translating these findings to naturalistic settings and validating HRV thresholds predictive of on-road impairment. [ABSTRACT FROM AUTHOR]