How Your Body Shape Shifts Over Two Years: Sleep and Exercise Emerge as Key Drivers in Korean Adults

For decades, the number on a scale and the ratio of weight to height have dominated how medicine thinks about body fat. A new longitudinal study from South Korea argues that this approach misses something crucial: the way fat distributes itself across the body, and the way that distribution changes over time, follows distinct patterns that body mass index alone cannot capture. By tracking more than 1,400 adults over two years and applying a machine learning technique to their changing measurements, researchers have identified three separate trajectories of body shape change, and they have tied two modifiable lifestyle factors, sleep quality and physical activity, to the trajectory most likely to threaten long-term metabolic health.

The study, published in PLOS One by Jaeho Lee, Kyoungsik Jeong, Youngseop Lee, and Younghwa Baek, drew its data from the Korean Medicine Daejeon Citizen Cohort, a community-based longitudinal study that follows adults in the city of Daejeon. Rather than treating each measurement as an isolated number, the team asked whether people move along recognizable paths as their bodies change. To do this, they applied k-means cluster analysis, an unsupervised algorithm that groups observations based on similarity. The clustering was performed on two-year changes in six body-shape measures: body mass index, waist circumference, hip circumference, waist-to-hip ratio, waist-to-height ratio, and total body weight. The method is simple in principle but powerful in effect, because it allows patterns of change to emerge from the data rather than being imposed by predetermined categories such as normal weight, overweight, or obese.

Three clusters emerged from the analysis, and their proportions are striking. The largest group, comprising 48 percent of participants, was labeled Stable, showing minimal change across all six measures over the two-year window. The second group, 31 percent of the cohort, fell into an Adiposity-Increasing pattern, characterized by growth in waist circumference and related markers of central fat accumulation. The remaining 21 percent formed an Adiposity-Decreasing cluster, in which waist and hip measurements moved downward together. The fact that nearly a third of a general adult population shifted toward increasing adiposity in just two years underscores how dynamic body shape truly is, and how quickly the distributions of fat that matter for metabolic risk can evolve in midlife.

Central adiposity, the accumulation of fat around the abdomen, has become a focal point of cardiometabolic research because it behaves differently from fat stored elsewhere. Visceral fat, which wraps around internal organs, is metabolically active in ways that promote insulin resistance, elevated blood pressure, and chronic low-grade inflammation. This is why waist circumference, waist-to-hip ratio, and waist-to-height ratio have repeatedly outperformed body mass index in predicting type 2 diabetes, cardiovascular disease, and mortality. Body mass index, being a simple ratio of weight to height squared, cannot distinguish between a person whose weight reflects dense skeletal muscle and one whose weight reflects abdominal fat, nor can it detect the gradual loss of muscle that accompanies aging. The Korean study was designed precisely to address these blind spots by monitoring shape and composition simultaneously.

The researchers went a step further by connecting the clusters to underlying body composition. At baseline, cluster membership was significantly associated with body fat mass, meaning that the amount of fat a person carried before the study began already helped predict which trajectory their body shape would follow. More importantly, the changes in body fat mass and skeletal muscle mass over the follow-up period showed strong associations with cluster assignment. This dual signal, fat gain and muscle change moving in tandem with shape change, supports a model in which adiposity-related trajectories are not merely about excess energy storage but about the shifting balance between fat and lean tissue. A person can gain waist circumference while losing muscle mass, a combination that is metabolically worse than fat gain alone, and the cluster approach makes such patterns visible in a way that single measures cannot.

The lifestyle findings are where the study acquires its most practical relevance. Using multinomial logistic regression, a statistical method for modeling outcomes with more than two categories, the researchers tested which baseline and follow-up variables predicted membership in each cluster. Two factors stood out for the Adiposity-Increasing group. The first was a clinically meaningful worsening of sleep quality, defined as an increase of at least 2.5 points on the Pittsburgh Sleep Quality Index, a validated questionnaire that scores sleep across seven dimensions including duration, latency, disturbances, and daytime dysfunction. The second was a reduction in physical activity as measured by the Global Physical Activity Questionnaire, an instrument that captures work-related, transport-related, and recreational activity. Both associations remained significant after accounting for other variables, pointing to sleep and movement as modifiable levers that may help steer people away from the harmful trajectory.

The link between sleep and adiposity has accumulated considerable mechanistic support in recent years. Short or fragmented sleep disrupts the hormonal regulation of appetite, raising ghrelin, the hormone that stimulates hunger, and lowering leptin, the hormone that signals satiety. Poor sleep also increases insulin resistance, alters food preferences toward energy-dense options, and may reduce the energy expended through spontaneous physical movement the following day. Sleep quality deterioration is often a gradual process tied to stress, irregular schedules, and aging, which makes it easy to overlook in clinical practice. The finding that even a moderate worsening on a standardized sleep index tracks with increasing central adiposity suggests that clinicians monitoring waist circumference should also be asking patients about how they are sleeping.

Physical activity operates through more obvious but equally important channels. Skeletal muscle is the body’s largest site of glucose disposal, and regular activity maintains muscle mass and insulin sensitivity while increasing total energy expenditure. When activity declines, muscle mass tends to erode, resting metabolic rate falls, and the same caloric intake produces a greater fat surplus. The co-occurrence of reduced activity and increasing adiposity in the same cluster therefore reflects a coherent physiological story: less muscle-stimulating movement, more fat accumulation, and a body shape that shifts outward at the waist. The Korean data reinforce the idea that interventions aimed at preventing central obesity should treat exercise not simply as a weight-loss tool but as a body-composition intervention that protects lean tissue.

The study has limitations worth keeping in view. It was conducted in a single Korean city, and body-shape trajectories may differ across populations with different diets, occupational structures, and genetic backgrounds. The observational design means that associations between sleep, activity, and cluster membership do not prove causation; it is possible, for example, that early metabolic changes drive both poor sleep and reduced activity rather than the reverse. Even so, the two-year follow-up, the use of multiple correlated shape measures, and the integration of body-composition data give the findings a depth that cross-sectional studies of a single measure cannot match. Because sleep and activity are modifiable, the results translate readily into intervention targets even under the more conservative causal reading.

The broader message is one that resonates with a growing movement in obesity science: replacing the single-number mentality with an integrative view of body shape, body composition, and lifestyle. Two people with identical body mass indexes may sit on entirely different health trajectories depending on where their fat is stored, how much muscle they carry, how well they sleep, and how much they move. The Korean cohort’s three clusters, stable, adiposity-increasing, and adiposity-decreasing, provide a template for identifying people early, before their changing waistlines harden into diabetes or cardiovascular disease. And they offer reassurance of a kind: the trajectory most tied to risk is also the one most tied to behaviors that individuals can, with support and attention, change. Body shape, in other words, is not destiny written in a BMI number, but a dynamic process shaped nightly by sleep and daily by movement.

Subject of Research: Longitudinal clustering of body shape change and its association with body composition and lifestyle factors in Korean adults

Article Title: Lifestyle factors and clustered patterns of body shape change in Korean adults: A longitudinal observational study

Article References: Lee, J., Jeong, K., Lee, Y., & Baek, Y. (2026). Lifestyle factors and clustered patterns of body shape change in Korean adults: A longitudinal observational study. PLOS One, 21(10), e0360853. https://doi.org/10.1371/journal.pone.0360853

Image Credits: AI Generated

DOI: 10.1371/journal.pone.0360853

Keywords: obesity, body shape, waist circumference, body mass index, sleep quality, physical activity, body composition, skeletal muscle mass, cluster analysis, Korean adults, longitudinal study, adiposity