Predicting Cognitive Performance from Circadian-Relevant Lifestyle Factors

Circadian rhythm regulates sleep and alertness through a 24-hour cycle, and disruption of thiscycle has been associated with reduced cognitive performance. As night-shift work and irregularschedules become more common, identifying which sleep and lifestyle factors are most closelyrelated to cognition could provide more specific guidance than general sleep recommendations.This study examined whether lifestyle and sleep variables, including circadian-related factorssuch as shift work and chronotype, could predict a continuous cognitive performance score. Aninitial dataset of 60 participants produced unreliable results (R² = -0.85), so the analysis used adataset of 100,000 participants. Linear Regression, Ridge Regression, and Random Forestmodels were trained on 80,000 participants and tested on the remaining 20,000. Linear andRidge Regression each reached an R² of 0.90, and Random Forest reached an R² of 0.92 witha Mean Absolute Error of 5.10 points on the 100-point scale. Sleep quality score accounted for73.6% of the Random Forest’s feature importance, while shift work and chronotype togetheraccounted for about 0.15%. Shift workers scored about 20 points lower than other participantson average, but most of that gap disappeared when sleep quality was included in a linearmodel. These results suggest that circadian disruption may affect cognition mainly through sleepquality, which should be tested with causal methods and independent data.