Identifying a Digital Phenotype of Allostatic Load: Association Between Allostatic Load Index Score and Wearable Physiological Response During Military Training

Allostatic load – a maladaptive biological process wherein physiological stability (‘allostasis’) fails owing to chronic stress exposure - is traditionally measured by the Allostatic Load Index (ALI). Whether ALI is associated with wearable-assessed physiological responses remains unknown. We aimed to determine the association between ALI and wearable-assessed physiological responses to a 10-week military training course. Twenty-five participants (12 women) with ALI and suitable wearable data (84.31% complete data [range: 64.71%−97.56%]) were included. ALI (0–8) was calculated using biomarker components from neuroendocrine, autonomic, and immune systems. Device variables included total energy expenditure (TEE), energy expenditure during physical activity (PAEE), daytime heart rate (HR), sleeping HR, nonlinear HR variability (DFA-α 1 ), and sleep architecture. Flux was calculated as raw (Δ) or absolute difference (|Δ|) in average values between days or nights. Generalized linear mixed effect models assessed the association between high allostatic load (ALI>4) and responses (α=0.05). Twelve (4 women) participants experienced ALI>4. High allostatic load was associated with TEE (β=0.658, SE=0.002, OR=1.931, p<0.001), Δ in relative PAEE (β=0.472, SE=0.002, OR=1.602, p<0.001), daytime HR (β=0.189, SE=0.002, OR=1.208, p<0.001), |Δ| in relative daytime HR (β=0.262, SE=0.001, OR=1.298, p<0.001), and |Δ| in relative sleeping HR (β=−0.048, SE=0.001, OR=0.953, p<0.001). Every one-standard-deviation increase in absolute TEE, flux in relative PAEE, daytime HR, flux in daytime HR, and reduced flux in sleeping HR increased the risk of high allostatic load by 5-90%. Chronically elevated and variable cardiometabolic activity with blunted night-to-night variation in sleeping HR may be a digital phenotype of high allostatic load in military personnel.