Functional data analysis of college students’ sleep patterns and their relationships with academic performance and social networks: A four-year longitudinal study

Background: College students are subject to insufficient sleep and irregular sleep patterns. Most existing studies regarding sleep behaviors have relied on static measures or discrete time points for analyzing sleep data, potentially missing the dynamic and continuous nature of sleep behavior. Objectives: To examine sleep pattern evolution and their relationships with academic performance and social networks among college students using functional data analysis. Methods: This study introduces functional data analysis to examine sleep patterns and their relationships with academic performance and social networks among college students throughout their four-year undergraduate experience. Using data from the NetHealth Project, we analyzed daily sleep records from Fitbit devices worn by 76 undergraduate students, along with their academic records and social network data, comprising 61,225 daily observations. We employed functional regression to model time-varying relationships between sleep and GPA, and functional t-tests to compare sleep patterns between students with different social activity levels. Results: Sleep duration increased significantly across the undergraduate years, with pronounced seasonal fluctuations corresponding to academic cycles. The relationship between sleep and academic performance remained consistently positive throughout college, with each GPA point associated with 27.4 additional minutes of sleep on average. This relationship exhibited a U-shaped temporal pattern, strongest during freshman year (54 minutes/GPA point), weakest during junior year (5 minutes/GPA point), and recovering during senior year (48 minutes/GPA point). Social network characteristics showed no statistically significant associations with sleep patterns, though students with larger networks consistently slept slightly less than those with smaller networks. Conclusions: This study demonstrates the utility of functional data analysis in sleep research, revealing dynamic patterns in sleep behavior and time-varying relationships with academic performance that traditional discrete-time analyses would not capture. The consistently positive association between sleep duration and academic performance was maintained throughout the four-year undergraduate experience, with temporal variations suggesting the relationship is strongest during freshman and senior years. These findings provide longitudinal evidence about sleep patterns and their correlates among college students, with potential implications for the timing of sleep-related support services.