The Journal of nutrition · 2026
Dietary Patterns Are Key Predictors of Cardiovascular Risk Beyond Genetic Risk Across Metabolic Syndrome Trajectories in A Korean Prospective Cohort 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.
Yang Hye Jeong, Kim Min Jung, Jang Hyun-Jun, Hur Haeng Jeon, Kim Myung-Sunny, Park Sunmin
Аннотация
BACKGROUND: Metabolic syndrome (MetS) progression involves distinct trajectory states, including newly-developed (ND-MetS), recovered (R-MetS), and persistent MetS (P-MetS), each associated with different cardiovascular disease (CVD) risk profiles. The transition to incident MetS represents a critical window for early dietary intervention, as subclinical metabolic deterioration often precedes clinical recognition. However, the predictive value of dietary quality relative to biochemical and genetic markers within integrated machine learning frameworks remains poorly characterized in longitudinal cohorts.
OBJECTIVES: This study aimed to develop and evaluate machine learning-based prediction models integrating dietary quality indices, biochemical markers, and genetic variants for CVD risk stratification across MetS trajectory groups in a large Korean prospective cohort.
METHODS: We analyzed 58,701 Korean adults from the Korean Genome and Epidemiology Study with a 5-year follow-up, classified into four MetS trajectory groups: N-MetS (no MetS; n=46,647), ND-MetS (n=3,505), R-MetS (n=5,307), and P-MetS (n=3,242). Cox proportional hazards models established CVD risk across trajectory groups, identifying ND-MetS as the highest-risk population. Random forest and extreme gradient boosting(XGBoost) models were then developed to evaluate the predictive contributions of dietary patterns, Modified Korean Healthy Eating Index scores, biochemical markers, and genetic variants.
RESULTS: ND-MetS was positively associated with incident CVD risk (HR=1.87, 95% CI: 1.23-2.67) compared with N-MetS. Excluding MetS component variables to avoid circularity, the random forest model achieved the best predictive performance (AUC=0.756), outperforming XGBoost (AUC=0.731). Key predictors included hepatic enzymes (γ-GTP, ALT, AST), inflammatory markers (hs-CRP), dietary patterns, and anthropometric measures. The genetic risk score of the APOA4-APOA5-ZPR1-SIK3-APOC3 cluster consistently ranked below dietary patterns (reflecting diet quality) and biochemical markers in predictive importance across all machine learning models.
CONCLUSIONS: Machine learning models integrating dietary quality indices with routine biochemical markers improved CVD risk stratification in ND-MetS beyond genetic approaches alone, offering a practical, scalable, and genotype-independent precision nutrition strategy for Korean adults.
OBJECTIVES: This study aimed to develop and evaluate machine learning-based prediction models integrating dietary quality indices, biochemical markers, and genetic variants for CVD risk stratification across MetS trajectory groups in a large Korean prospective cohort.
METHODS: We analyzed 58,701 Korean adults from the Korean Genome and Epidemiology Study with a 5-year follow-up, classified into four MetS trajectory groups: N-MetS (no MetS; n=46,647), ND-MetS (n=3,505), R-MetS (n=5,307), and P-MetS (n=3,242). Cox proportional hazards models established CVD risk across trajectory groups, identifying ND-MetS as the highest-risk population. Random forest and extreme gradient boosting(XGBoost) models were then developed to evaluate the predictive contributions of dietary patterns, Modified Korean Healthy Eating Index scores, biochemical markers, and genetic variants.
RESULTS: ND-MetS was positively associated with incident CVD risk (HR=1.87, 95% CI: 1.23-2.67) compared with N-MetS. Excluding MetS component variables to avoid circularity, the random forest model achieved the best predictive performance (AUC=0.756), outperforming XGBoost (AUC=0.731). Key predictors included hepatic enzymes (γ-GTP, ALT, AST), inflammatory markers (hs-CRP), dietary patterns, and anthropometric measures. The genetic risk score of the APOA4-APOA5-ZPR1-SIK3-APOC3 cluster consistently ranked below dietary patterns (reflecting diet quality) and biochemical markers in predictive importance across all machine learning models.
CONCLUSIONS: Machine learning models integrating dietary quality indices with routine biochemical markers improved CVD risk stratification in ND-MetS beyond genetic approaches alone, offering a practical, scalable, and genotype-independent precision nutrition strategy for Korean adults.