Feasibility of Using Passive Digital Phenotyping Data from Smartphones as Emerging Health Sensors in Medical Students: A Pilot Exploratory Study

Abstract

Background: Noncommunicable diseases and mental-health–related risks are rising among young adults in low- and middle-income countries. With near-universal smartphone use among Indonesian students, passively captured digital traces may offer low-cost signals of lifestyle balance relevant to student wellbeing.  Objective: To characterize smartphone-derived behavioral signatures of daily balance (screen exposure, sleep regularity, mobility, and notification–pickup dynamics) among Indonesian medical students and test their group-level associations as an exploratory foundation for future early-warning models.    Methods: Cross-sectional study of undergraduate medical students (UNISSULA, Indonesia) using iPhones ≥6 months. Anonymized screenshots from Apple Health and Screen Time (July–August 2024) provided daily steps, sleep duration, screen time, notifications, pickups, and dominant app category (social vs entertainment; two independent raters). Normality was assessed; non-parametric tests (Spearman’s rho, Kruskal–Wallis) were applied where appropriate. Results: Forty-three students participated (n=43). Means (±SD): steps 3,546 ± 1,987/day, screen time 7.0 ± 2.3 h/day, sleep 4.2 ± 1.6 h/night, notifications 304 ± 151/day, pickups 143 ± 57/day. App-use distribution showed a polarity: 53.5% social-dominant vs 46.5% entertainment-dominant users. Screen time correlated negatively with sleep (ρ = −0.531, p < 0.001). Notifications correlated positively with pickups (ρ = 0.781, p < 0.001). Entertainment dominance application usage was associated with fewer steps than social dominance one (ρ = −0.455, p = 0.002; Kruskal–Wallis p = 0.03). Longer screen time predicted lower step count (p = 0.031).    Conclusions: Smartphone-derived metrics reveal a behavioral signature of imbalance: longer screen exposure, entertainment-heavy use, and high notification; pickup intensity linked to lower mobility and shorter sleep. These exploratory findings support the feasibility of smartphone data as candidate early-warning inputs for student wellbeing dashboards in resource-limited settings. Future longitudinal studies with psychometric/clinical labels, multi-device inclusion, and privacy-preserving pipelines are warranted.