O023 Predicting Dim Light Melatonin Onset in elite athletes without actigraphy

INTRODUCTION: Elite athletes are exposed to irregular schedules, jetlag and intensive physical activity that is sometimes completed late at night, all of which can misalign circadian rhythms and therefore negatively affect recovery and performance. The aim of this study was to predict circadian phase in athletes without actigraphy. METHODS: A regularised random forest model comprising of 17 questionnaire and sleep diary measures was used to predict dim light melatonin onset (DLMO) times from 128 DLMO times collected from 98 (age = 24.3 ± 3.7 years) Australian Football League and National Rugby League male (n = 72, age = 24.6 ± 3.4 years) and female (n = 26, age = 23.3 ± 4.3 years) athletes. RESULTS: DLMO time was predicted with an out-of-sample root mean square error of 52 minutes, capturing 75% with ± 1 hour of actual DLMO time. Actual DLMO time (20.65 ± 1.00) was significantly correlated with predicted DLMO time (20.69 ± 0.29; r2 = 0.36, p < 0.001). DISCUSSION: DLMO time can be predicted without actigraphy in elite athletes. This approach enables the possibility of immediate circadian phase and sleep insights, and the development of interventions necessary for the fast-paced environment of elite sport where optimising athlete recovery and performance is critical.