Table 1_Interpretable prediction of coronary heart disease risk in adults over 50 with accelerated aging using 45 dietary nutrients.docx

BackgroundThe relationship between dietary nutrient intake and coronary heart disease (CHD) risk among older adults with accelerated aging remains inadequately understood.
MethodsThis study analyzed data from seven cycles of the National Health and Nutrition Examination Survey (NHANES) conducted in the United States between 2005 and 2018. Weighted Quantile Sum (WQS) regression was employed to evaluate the association between dietary nutrient mixtures and CHD risk in individuals aged 50 and older with accelerated aging. Additionally, six machine learning models were developed, with SHAP and LIME algorithms applied to assess the contribution of individual nutrients to CHD risk.
ResultsIn the fully adjusted model, dietary nutrient mixtures were inversely associated with CHD risk in older adults experiencing accelerated aging (adjusted OR = 0.90, 95% CI: 0.81–0.99, p = 0.048). Both SHAP and LIME analyses consistently identified vitamin B12 and lutein + zeaxanthin as protective nutrients, independent of demographic adjustments.
ConclusionAmong adults aged 50 and older with accelerated aging, higher intake of specific dietary nutrients was associated with reduced CHD risk. Of the machine learning models tested, the random forest algorithm demonstrated the strongest predictive performance. SHAP and LIME analyses jointly highlighted vitamin B12 and lutein + zeaxanthin as key contributors to the reduced CHD risk in this high-risk population.

背景：加速衰老大人群的膳食营养素摄入与冠心病（coronary heart disease, CHD）发病风险之间的关联，目前仍未得到充分阐明。
方法：本研究分析了2005年至2018年间美国开展的7轮美国国家健康与营养检查调查（National Health and Nutrition Examination Survey, NHANES）数据。本研究采用加权分位数和（Weighted Quantile Sum, WQS）回归，评估加速衰老的50岁及以上人群的膳食营养素组合与冠心病发病风险之间的关联；此外还构建了6种机器学习模型，并应用SHAP（SHapley Additive exPlanations）与LIME（Local Interpretable Model-agnostic Explanations）算法，以评估单一营养素对冠心病发病风险的贡献度。
结果：在完全校正模型中，加速衰老大人群的膳食营养素组合与冠心病发病风险呈负相关（校正后优势比（Odds Ratio, OR）=0.90，95%置信区间（Confidence Interval, CI）：0.81~0.99，P=0.048）。SHAP与LIME分析均一致证实，维生素B12及叶黄素+玉米黄质为保护性营养素，且该结果不受人口统计学因素校正的影响。
结论：在50岁及以上的加速衰老大人群中，特定膳食营养素的更高摄入与更低的冠心病发病风险相关。在所测试的6种机器学习模型中，随机森林算法展现出最优的预测性能。SHAP与LIME分析共同证实，维生素B12及叶黄素+玉米黄质是该高危人群冠心病发病风险降低的关键影响因素。