Sleep‐digital phenotypes among undergraduate medical students: A latent class and symptom‐network analysis

Abstract Background Studies of student smartphone use usually treat screen time as one linear risk factor and dichotomise sleep quality, hiding how symptoms cluster and connect. Methods In a cross‐sectional survey, 354 undergraduate medical students in India reported sleep quality, sleepiness, sleep hygiene, stress, mood and digital habits. Latent classes were validated against examination percentage and excessive daytime sleepiness, neither of which was used for derivation. A regularised partial‐correlation network of 39 items located cross‐domain bridge symptoms, with stability and domain‐definition sensitivity analyses. Results Three classes formed a severity gradient: low‐burden (36.2%), intermediate‐burden (48.0%) and high‐burden (15.8%). Examination scores (64.4% in high‐burden vs. 68.0% in low‐burden students) and excessive daytime sleepiness (33.9% vs. 10.9%) differed across classes ( p = 0.003; p < 0.001). The strongest and most stable bridges were heavy exercise near bedtime, low mood, difficulty falling asleep and perceived stress (bootstrap top‐10 retention 93%–99%). Under five domain definitions, at most one digital‐habit item ranked in the top 10. The purpose of bedtime phone use was not significantly associated with poor sleep (adjusted p = 0.90). Conclusions A high‐burden minority carries a disproportionate share of sleepiness and academic impairment. Pre‐sleep arousal and stress, rather than screen exposure, are the most strongly connected symptoms and merit longitudinal testing as intervention targets.