The global burden of overweight and obesity has tripled over the past 50 years, posing an urgent public health challenge (1). More than half the European and North American population was overweight or obese (2). Such trend is mirrored in China, where the prevalence of overweight and obesity among adults has exceeded 50% and is projected to reach 65.3% by 2030, with substantial implications for healthcare costs. This trend was expected to lead to medical costs exceeding 58.52 billion United States Dollar (USD), representing approximately 21.5% of China’s total healthcare expenditure (3). Beyond economic burdens, obesity profoundly compromises cognitive function through reduced concentration and memory linked to poor sleep, and psychological wellbeing by elevating depression and anxiety from weight stigmatization (4); manifesting as elevated clinical depression and anxiety disorders among obese students (5), pathological eating behaviors like binge-restrict cycles, and social withdrawal due to body image dissatisfaction, collectively impairing identity formation during this critical developmental stage and long-term health trajectories. Among college students, 47.3% with late-night snacking habits face 9.8-fold higher obesity risk, while high-fat dietary patterns like aquatic meat directly increase visceral adiposity. These synergistically impair academic performance, social adaptability, and lifelong productivity, underscoring the need for targeted interventions.

The prevalence of overweight and obesity was substantially high among college students (6). Several studies have shown that 1/3 college students have overweight or obesity and the prevalence rate of obesity in men was significantly higher than that in women (7, 8). Obesity is influenced by a complex interplay of genetic, environmental, and behavioral factors. While genetic predisposition plays a role, the rapid increase in obesity rates highlights the critical influence of environmental and lifestyle changes (9). Dietary habits and lifestyle factors may determine the obesity phenotypes in children and adolescents (10). Recent researches have linked overweight and obesity in this group to unhealthy habits such as a strong preference for high-fat and high-sugar foods, insufficient physical activity, inadequate sleep, and cigarette consumption (11–13). Furthermore, sociodemographic factors, including age, education level, household income, and nutritional knowledge, also contribute to the incidence of overweight and obesity (14).

There are various methods for measuring obesity, and different countries and regions may use different measurement indicators. Most studies only used a single indicator (15), while the combination of multiple indicators can comprehensively consider multiple factors, reduce errors, and improve prediction accuracy. The bioelectrical impedance analyzer (BIA) is a portable, efficient, and fast method that can simultaneously measure multiple obesity indicators, such as commonly used body mass index (BMI), PBF, visceral fat level (VFL), waist hip ratio (WHR), waist circumference (WC), and fat mass index (FMI) (16). Among these, PBF represents the proportion of body fat to total body weight, reflecting how much fat is present in the body. In some populations, although their BMI are within the normal range, their PBF may exceed the normal level and lead to hidden obesity. Abnormally high PBF increases the risk of related diseases such as high blood sugar (17). Therefore, PBF is more valuable as a reference for assessing individual health risks compared to other indicators.

Multiple studies have shown that obesity is closely related to diet, dietary factors greatly affect body composition, and dietary pattern analysis could comprehensively explore the impact of food intake on human health (18). Numerous studies have shown that dietary patterns are closely related to non communicable chronic diseases such as obesity, hypertension, and cardiovascular and cerebrovascular diseases. For example, a healthy dietary pattern is positively correlated with good levels of cardiorespiratory health. Numerous studies have demonstrated significant associations between specific dietary patterns and obesity risk in young adults. For instance, research among college populations indicates that inadequate vegetable consumption doubles obesity risk, while high intake of processed snacks correlates with higher visceral adiposity. Our study builds on this evidence by identifying four distinct dietary patterns in Chinese college students, including the obesity-promoting snack mode and protective vegetable meat grain patterns, quantifying population-specific risk (19, 20). Poor diet could become one of the main causes of overweight and obesity in adolescents (21).

Health risk assessment models played a crucial role in identifying individuals at higher risk of obesity and guiding targeted interventions (22). Classic models usually included BMI and Box-Cox power exponential (BCPE) model to assess obesity risk (23). However, important behavioral modifiable factors such as smoking, physical activity, and diet are often overlooked in these models. Therefore, developing a non-invasive, behavior-based risk assessment model that incorporates modifiable lifestyle factors was essential to enable the early and individualized interventions suggestions for college students at risk of obesity (24).

Hence, this study aimed to investigate the distribution and epidemiological characteristics of overweight and obesity among college students in Xuzhou, and to develop risk assessment models for different obesity types with visualization, providing a theoretical basis for obesity prevention and its practical application.

The investigation revealed a lack of research on risk assessment for overweight and obesity in college students based on dietary patterns. Our study found that the prevalence of overweight and obesity among college students is relatively high and overall obesity prevalence analysis revealed 42.5% of participants (376 out of 885) had PBF-defined obesity. PBF-defined obesity was defined as > 25% body fat for males and > 30% for females; significantly more males were affected (55.0%) than females (32.7%). Supplementary Table 1 demonstrates varying results across the six obesity indicators, further highlighting the differences between these metrics. A comprehensive analysis incorporating multiple indicators can help reduce measurement errors. The “vegetable meat grain dietary pattern” and “milk egg dietary pattern” were associated with a reduced risk of PBF, while the “snack mode dietary pattern” and “aquatic meat dietary pattern” dietary patterns were linked to an increased risk of PBF. Binary logistic regression analysis identified five key predictors of body fat percentage obesity in college students: gender, exercise habits, late-night snacking, regular meal patterns, and adherence to a healthy diet.

The observed associations between dietary patterns and obesity phenotypes were mediated through distinct metabolic mechanisms, where the meat and aquatic product dietary pattern—rich in saturated and monounsaturated fats—upregulated de novo lipogenesis via SREBP-1c activation, promoted visceral adiposity and hepatic insulin resistance that reduced Very Low-Density Lipoprotein clearance (29); conversely, the vegetable meat grain pattern provided polyunsaturated fatty acids (PUFAs) activating PPAR-α pathways to enhance β-oxidation and reduce FMI and VFL. The Snack mode pattern’s high sugar load induced postprandial hyperinsulinemia, promoting adipocyte differentiation through PI3K or Akt signaling. Fructose metabolism specifically stimulated hepatic de novo lipogenesis via ChREBP activation (30). The milk egg pattern delivered bioactive peptides that increased thermogenesis through UCP1 upregulation in brown adipose tissue, and its calcium content inhibited calcitriol-mediated adipocyte Ca²⁺ influx to suppress lipogenesis. Additionally, plant-based fibers in the vegetable pattern were fermented to SCFAs, such as butyrate, which improved insulin sensitivity via GLP-1 secretion and AMPK activation, explaining its protective effect against PBF accumulation (31).

The prevalence of overweight and obesity among college students has become a growing concern. According to the BMI evaluation standard, 30.7% of students are classified as overweight or obese. However, BMI may underestimate the true prevalence of obesity (15). When assessed using the body fat percentage standard, the prevalence increases to 42.5%, reflecting improved predictive accuracy (32). This perspective was supported by other studies (33). This situation can be attributed to the heavy academic pressure faced by young students and the widespread use of electronic devices, which contribute to sedentary behavior, insufficient physical activity, and unhealthy dietary habits. These factors collectively led to abdominal fat accumulation and relatively low muscle mass (33).

Abdominal fat is commonly assessed using indices such as VFL, WHR, and WC. In our study, the prevalence of central obesity varied significantly depending on the measure used: 21.0% based on the VFL standard, 35.6% using WHR, and 40.2% using WC. The correlation and consistency among these measures were not statistically significant, likely due to their inherent strengths and limitations, as well as variations in body composition, muscle mass, and gender-related fat distribution (34).

This study also explored the relationship between dietary patterns and various obesity indicators. The first dietary pattern, characterized by a traditional Chinese plant-based diet rich in rice, wheat, whole grains, vegetables, bean products, fruits, and mushrooms, was negatively associated with BMI and PBF. These findings aligned with previous research suggesting that plant-based diets, similar to the Eastern dietary pattern, were linked to a lower risk of obesity (35). Furthermore, evidence indicated that higher vegetable consumption was negatively associated with body weight, while fruit and vegetable intake is positively associated with a lower body fat percentage (36).

In contrast, the second dietary pattern, primarily composed of pork, poultry, livestock, aquatic products, and oil-based foods, showed positive associations with PBF, FMI, and VFL. This pattern is closer to a Western dietary style and is more prevalent among students from high-income households. Due to its higher fat content compared to other patterns, it is more likely to contribute to abdominal obesity, as evidenced by elevated FMI and VFL. These results were consistent with previous studies linking high-fat diets to increased risks of central obesity (37).

Research indicates that dietary patterns rich in saturated and monounsaturated fatty acids are associated with a 25% increase in the odds of developing metabolic disorders. A study of Filipino adults aged 20 and above found that a dietary pattern primarily consisting of meat and sugary drinks was significantly positively associated with diabetes, overweight, and obesity (38).

The third dietary pattern, characterized by a high intake of beverages, snacks, and sweets, represents a common junk food diet among students. It is positively correlated with PBF, WHR, and WC. This pattern’s high proportion of sugary drinks aligns with evidence showing that unhealthy foods, including sweets and sugary beverages, may increased the risk of abdominal obesity (39). Sugar intake, in particular, has been strongly linked to visceral fat accumulation and other body composition variables (40).

In contrast, the fourth dietary pattern, dominated by dairy and egg products, was negatively associated with PBF and BMI. Dairy products are rich in essential vitamins and minerals, while eggs provide high-quality protein and calcium. Adherence to this pattern ensured sufficient nutrient intake and reduced the risk of obesity (41).

To further investigate, a simplified healthy dietary pattern score was developed using posterior inference to extract dietary patterns. This approach has been widely applied in studies examining the relationship between diet and metabolic diseases such as non-alcoholic fatty liver disease, hypertension, and hyperlipidemia (42). Based on the four identified dietary patterns, a simple dietary score was constructed and found to be significantly correlated with PBF. The food items included in the score reflect the eating habits of the study population and emphasize food groups with a notable impact on obesity. This healthy dietary pattern showed a stronger association with PBF risk, thereby enhancing the potential for targeted obesity prevention strategies.

Our study identified several key factors influencing overweight and obesity among college students, including gender, physical activity, sleep duration, fluid intake, alcohol consumption, regular meals, and adherence to a healthy diet. The prevalence of PBF-defined obesity among male students was 55.0%, significantly higher than the 32.7% observed in females. Other obesity indicators also revealed higher prevalence rates among males, consistent with global and domestic trends (43). This disparity can largely be attributed to lifestyle differences: males are more likely to engage in smoking and alcohol consumption, whereas females tend to adopt healthier habits, such as higher fruit and vegetable intake, reduced alcohol and meat consumption, and increased physical activity (44).

Alcohol consumption was confirmed as a significant risk factor for overweight and obesity among college students. A prospective study demonstrated that reducing weekly alcohol intake and avoiding excessive drinking can help control waist circumference and BMI in men (45). Excessive and frequent alcohol consumption during youth has been shown to contribute to weight gain (46).

Physical activity emerged as one of the most effective ways to improve obesity-related metrics. Increasing moderate-intensity exercise while reducing low-intensity activities significantly improved obesity indicators (47). Additionally, more physical activity, less screen time, and sufficient sleep were positively associated with lower BMI, reduced cardiometabolic risk, and improved cognitive development, such as enhanced motor skills and academic performance (48).

Sleep duration also played a crucial role. A study of Chinese adolescents found that short sleep duration was associated with increased overweight or obesity, particularly central obesity in boys (49). Similarly, adequate hydration has been linked to healthier body composition. Research suggested that sufficient water intake might help prevent abdominal obesity, particularly in young adults (50).

Moreover, higher meal frequency was associated with slightly lower body fat percentage (51). Furthermore, a higher eating frequency could be associated with diet quality improvement, lower adiposity, and lower risk of developing MetS or its components (52).

Finally, our study developed a non-invasive obesity risk assessment model using PBF as the primary indicator, incorporating demographic, dietary, and lifestyle factors. The use of the InBody 570 analyzer provided accurate body composition data within a short timeframe (53), and its reliability has led to frequent application in obesity research (54). The resulting model demonstrated strong discriminatory power and calibration, with an AUC exceeding 0.7. In addition to traditional measures such as BMI and WC for defining overall and abdominal obesity, this study included comprehensive body composition indicators such as FMI, VFL, WHR, and PBF. While our PBF-based nomogram (AUC = 0.805) demonstrates robust predictive performance for college-specific obesity screening, we acknowledge the need for contextualizing these findings against existing models. In direct comparison to Liu et al. (24)—whose machine learning model predicted BMI-defined obesity in Chinese adolescents with an AUC of 0.72—our approach achieves key advancements through outcome specificity. We target PBF-defined obesity, a more sensitive adiposity indicator than BMI, capturing a 42.5% prevalence rate versus BMI’s 30.7% prevalence. Behavioral granularity emerges when incorporating dietary pattern scores such as the healthy diet extract (OR = 0.961, p < 0.002), enabling precise behavioral interventions, whereas Liu et al.’s (24) model relied primarily on demographic or anthropometric factors. Visualizing the results through line charts enhances the model’s practical utility, making it a valuable tool for college health programs. However, the model’s applicability may be limited to the local population, necessitating further validation in diverse cohorts.

Despite its contributions, this study has several limitations. First, as a cross-sectional survey, it cannot establish causal relationships or determine temporal sequences. Second, the sample was limited to college students from a specific region, which limits the generalizability of the findings. Third, the lack of parental body shape data in the questionnaire design prevented the inclusion of hereditary factors in the regression analysis. Future research should focus on larger and more diverse samples, employ longitudinal designs, and explore genetic influences on body composition to enhance the predictive accuracy of obesity risk models. Additionally, validating the model in external populations is crucial to improve its applicability beyond the local context.

In this study, the non-invasive college body fat obesity risk assessment model, combined with the ROC curve, offers valuable support for the primary prevention and management of obesity in college students. Furthermore, we found a significant association between dietary patterns and obesity among college students. Four distinct dietary patterns were identified: the vegetable meat grain dietary pattern, the aquatic meat dietary pattern, the snack mode dietary pattern and the milk egg dietary pattern. A healthy dietary pattern, constructed based on these four patterns, was negatively associated with obesity risk in college students. Binary logistic regression analysis highlighted four key predictors of body fat percentage obesity: gender, physical activity, late-night snacking, regular meals, and adherence to a healthy diet.