Association Between Physical Activity Timing and Metabolic Syndrome in Korea: A Functional Principal Component Approach

Abstract

Background: Metabolic syndrome (MetS), characterized by the co-occurrence of obesity, hypertension, hyperglycemia, and dyslipidemia, substantially increases the risk of cardiovascular disease and type 2 diabetes. In South Korea, the prevalence of MetS is steadily increasing. While physical activity is known to mitigate this risk, recent evidence suggests that the timing of activity, not just its volume, may also be important. Methods: We analyzed accelerometer data from Korean adults who participated in the 2014-2016 Korea National Health and Nutrition Examination Survey (KNHANES). Functional principal component analysis (FPCA) was applied to minute-level physical activity trajectories to extract key temporal patterns. Logistic regression models assessed associations between the resulting principal component (PC) scores and MetS, adjusting for demographic, behavioral, and occupational factors, as well as total moderate-to-vigorous physical activity (MVPA). Results: Among the four extracted components, the third principal component (PC3)-reflecting higher morning and evening activity with reduced afternoon variability-was significantly associated with increased risk of MetS in the fully adjusted model (adjusted OR = 1.117; 95% CI: 1.003-1.244). Conclusions: These findings suggest that temporal patterns of physical activity, particularly reduced variability in the afternoon, may be linked to adverse metabolic outcomes. Beyond overall activity volume, the timing and distribution of daily physical activity should be considered in metabolic health research and interventions.

Keywords: functional principal component analysis; metabolic syndrome; temporal pattern.

Figure 1

Flowchart of study participant selection from the 2014–2016 Korea National Health and Nutrition Examination Survey (KNHANES) accelerometer dataset.

Figure 2

Mean function of minute-level physical activity intensity over seven consecutive days.

Figure 3

Estimated eigenfunctions (PC1–PC4) from functional principal component analysis of daily physical activity. PC1 reflects overall movement intensity across the day; PC2 captures evening activity variation; PC3 contrasts afternoon activity with morning and evening peaks; and PC4 represents irregular weekend-related fluctuations.

Figure 4

Estimated eigenfunctions of the four principal components (PC1–PC4) from FPCA. Red lines show the estimated eigenfunctions, and blue dots in panels (b,c) highlight the local maxima, indicating time periods with the greatest variation.