Dynamics of Sleep Architecture and Nocturnal Oxygen Saturation During Acclimatization at Intermediate Altitude: A Longitudinal Observational Study in Elite Athletes

Background: Athletes typically undergo training camps living and sleeping at intermediate altitudes (1500–2500 m). Hypoxia reshapes sleep through central circuits regulating arousal, chemoreflex gain and sympathetic outflow. However, the relationship between peripheral oxygen saturation (SpO2) and sleep architecture and how it co-evolves during acclimatization, is unclear. Methods: We tested whether SpO2 differs across sleep stages at intermediate altitude and whether its recovery is stage specific. Seventeen mogul-ski athletes (seven women) travelled from Montréal (Canada) to Tignes (France) for a 2-week altitude training camp (AltitudeCamp; 2100 m) in June 2025. During the first week (6 nights) and the last night of AltitudeCamp, athletes were provided with a validated ring pulse oximeter and an electroencephalography (EEG) headband. We evaluated the changes in sleep stage distribution across successive nights, scored on 30-second epochs and synchronized with oximetry data. Model estimates are reported as mean ± Standard Error (SE), raw data as mean ± Standard Deviation (SD). Results: Average total sleep time was 429 ± 47 min (mean ± SD) and did not change during AltitudeCamp (p = 0.939). Across the first week, %N3 decreased logarithmically (β = −2.35 ± 0.72%/ln (night), p = 0.002) and %Rapid eye movement sleep (REM) increased logarithmically (β = 3.13 ± 0.88%/ln (night), p < 0.001), with no changes in %Wake, %N1, or %N2. SpO2 was lower during AltitudeCamp (95.2 ± 0.3%) than measures obtained in normoxia (98 ± 0.4%; p < 0.001), and overnight heart rate (HR) was 7 ± 2 beats per minute (bpm) higher than the habitual 58 ± 3 bpm observed in normoxia (p = 0.005). Wake SpO2 increased by 0.13 ± 0.01 percentage points per night (p < 0.001), with faster recovery in N1 (0.19 ± 0.02 percentage points/night; p = 0.006) and REM (0.23 ± 0.01 percentage points/night; p < 0.001), while N2 and N3 did not differ from Wake. HR declined by 0.79 bpm/night during Wake, similar to N3 (p = 0.26), but larger than the decline in N1 (p = 0.024) and REM (p < 0.001). In N2, HR did not evolve during the first week at altitude. Conclusions: At intermediate altitude, nocturnal SpO2 and HR show sleep-stage-specific dynamics that mirror central ventilatory and autonomic adaptation to hypoxia. Our data suggest that acclimatization is not uniform across sleep stages, and that the first 2 nights after ascent may capture the most rapid adjustments despite similar total sleep duration. Sleep therefore offers a heterogeneous and ecologically valid index of central nervous system (CNS) acclimatization, which can be monitored longitudinally with wearable EEG–oximetry.