Associations of Free-Living Physical Activity, Sleep, and Wellness with Neuromuscular, Postural, and Visuomotor Performance in University Students: A 14-Day Multimodal Study

Background and Objectives: Habitual physical activity, sleep, and psychological wellness may relate to physical and cognitive–motor performance in young adults, but these relationships are seldom examined within one free-living monitoring framework. This study examined associations of daily physical activity, sleep, and wellness with neuromuscular, postural, and visuomotor performance in university students. Methods: In a 14-day prospective observational design, Samsung Health provided daily step totals, while WHO-5, PSS-10, PSQI, and daily wellness measures characterized recovery. OptoJump, GYKO, and BlazePod assessments were repeated across three laboratory sessions. The prespecified analysis cohort comprised 242 students with ≥10 valid monitoring days (including ≥1 weekend day) and ≥2 valid laboratory sessions per outcome. Sex-adjusted linear mixed-effects models included 237 participants and 696 repeated observations, adjusting for session, age, BMI, weekly training volume, academic period, and testing time. Results: Each additional 1000 steps/day was associated with greater countermovement jump height (β = 1.31 cm, 95% CI 0.97–1.65), lower eyes-closed L5 movement velocity (β = −0.65 mm/s, 95% CI −1.00 to −0.30), and faster choice-response time (β = −6.22 ms, 95% CI −9.37 to −3.08; all Holm-adjusted p < 0.001). Higher perceived stress and poorer PSQI-defined sleep quality were independently associated with less favorable performance across all three domains, whereas mean self-reported sleep duration was not. Conclusions: Higher habitual step count, lower perceived stress, and better sleep quality were associated with a more favorable multidimensional performance profile. Because step count was summarized at the participant level, these findings should be interpreted as between-participant associations rather than causal or day-to-day readiness effects. Device-source attribution and postural-measure validation remain limitations.