Frontiers in nutrition · 2026
Metabolic characteristics and factors associated with prediabetes in Chinese adults based on real-world health examination data: a cross-sectional study.
СтатьяНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Sun Xinru, Zhou Jun, Chen Zhaoyi, Xin Zeyin, Ren Jiayi, Zheng Yuhong, Wang Jinkun, Zhang Jiarun, Zhang Yuxin, Liu Lu, Li Bin, Wang Yizhan
Аннотация
BACKGROUND: Prediabetes is a common intermediate state of abnormal glucose metabolism, but its metabolic characteristics and associated factors in health examination populations remain insufficiently defined. Based on a health examination population of Chinese adults, this study aimed to characterize the metabolic profile of prediabetes and identify factors associated with prediabetes using real-world health examination data, thereby providing a foundation for future multicenter, multiregional, and broader population-based validation and implementation studies.
METHODS: This cross-sectional study included adults undergoing routine health examinations between January 1, 2018 and December 31, 2024. Metabolic characteristics were compared between normoglycemic individuals and those with prediabetes. Univariable and multivariable logistic regression analyses were performed to identify factors associated with prediabetes. XGBoost combined with SHAP was further used to assess the relative importance of clinical and metabolic indicators.
RESULTS: A total of 20,271 participants were included in the main analysis, including 8,457 with prediabetes and 11,814 with normoglycemia. Compared with the normoglycemia group, the prediabetes group was older (58.48 ± 13.10 vs. 48.10 ± 12.90, P < 0.001), had a higher BMI (26.82 ± 3.78 vs. 22.83 ± 3.95, P < 0.001) and was more likely to have hypertension and fatty liver. The prediabetes group also had higher TC, TG, and LDL-C levels and lower HDL-C and TBIL levels (P < 0.001). In multivariable analyses, BMI (aOR 2.52, 95% CI 2.42-2.61), fatty liver (aOR 2.80, 95% CI 2.60-3.02), TC (aOR 1.08, 95% CI 1.05-1.12), TG (aOR 1.09, 95% CI 1.05-1.12), and LDL-C (aOR 1.17, 95% CI 1.13-1.21), age, hypertension, ALT, and BUN, were positively associated with prediabetes, whereas HDL-C (aOR 0.84, 95% CI 0.81-0.87) and TBIL (aOR 0.75, 95% CI 0.72-0.78) were inversely associated. Among the extended models, the model additionally including BMI showed the best discriminative performance, with the AUC increasing from 0.763 to 0.825. The XGBoost model achieved an AUC of 0.941 (95% CI 0.934-0.948), and BMI, age, TG, fatty liver, and TC ranked among the most informative variables. SHAP dependence analysis showed that the contribution of BMI to model output was nonlinear, becoming positive at approximately 23.9 kg/m2, peaking near 29.4 kg/m2, and attenuating thereafter, whereas age showed a generally monotonic positive relationship with model output and crossed zero at approximately 54 years.
CONCLUSION: In this health examination population, prediabetes was associated with age, hypertension, BMI, fatty liver, lipid profiles, and altered hepatorenal function-related indicators. Among these factors, BMI and fatty liver showed relatively stronger associations, and machine-learning analysis further highlighted BMI, age, TG, fatty liver, and TC as particularly informative variables for the early identification of prediabetes.
METHODS: This cross-sectional study included adults undergoing routine health examinations between January 1, 2018 and December 31, 2024. Metabolic characteristics were compared between normoglycemic individuals and those with prediabetes. Univariable and multivariable logistic regression analyses were performed to identify factors associated with prediabetes. XGBoost combined with SHAP was further used to assess the relative importance of clinical and metabolic indicators.
RESULTS: A total of 20,271 participants were included in the main analysis, including 8,457 with prediabetes and 11,814 with normoglycemia. Compared with the normoglycemia group, the prediabetes group was older (58.48 ± 13.10 vs. 48.10 ± 12.90, P < 0.001), had a higher BMI (26.82 ± 3.78 vs. 22.83 ± 3.95, P < 0.001) and was more likely to have hypertension and fatty liver. The prediabetes group also had higher TC, TG, and LDL-C levels and lower HDL-C and TBIL levels (P < 0.001). In multivariable analyses, BMI (aOR 2.52, 95% CI 2.42-2.61), fatty liver (aOR 2.80, 95% CI 2.60-3.02), TC (aOR 1.08, 95% CI 1.05-1.12), TG (aOR 1.09, 95% CI 1.05-1.12), and LDL-C (aOR 1.17, 95% CI 1.13-1.21), age, hypertension, ALT, and BUN, were positively associated with prediabetes, whereas HDL-C (aOR 0.84, 95% CI 0.81-0.87) and TBIL (aOR 0.75, 95% CI 0.72-0.78) were inversely associated. Among the extended models, the model additionally including BMI showed the best discriminative performance, with the AUC increasing from 0.763 to 0.825. The XGBoost model achieved an AUC of 0.941 (95% CI 0.934-0.948), and BMI, age, TG, fatty liver, and TC ranked among the most informative variables. SHAP dependence analysis showed that the contribution of BMI to model output was nonlinear, becoming positive at approximately 23.9 kg/m2, peaking near 29.4 kg/m2, and attenuating thereafter, whereas age showed a generally monotonic positive relationship with model output and crossed zero at approximately 54 years.
CONCLUSION: In this health examination population, prediabetes was associated with age, hypertension, BMI, fatty liver, lipid profiles, and altered hepatorenal function-related indicators. Among these factors, BMI and fatty liver showed relatively stronger associations, and machine-learning analysis further highlighted BMI, age, TG, fatty liver, and TC as particularly informative variables for the early identification of prediabetes.