Verma Shreya, Holthaus Tori A, Martell Shelby, Holscher Hannah D, Zhu Ruoqing, Khan Naiman A
Health and Kinesiology, University of Illinois Urbana-Champaign, Urbana, IL, United States.
Division of Nutritional Sciences, University of Illinois Urbana-Champaign, Urbana, IL, United States.
J Nutr. 2025 Jul;155(7):2144-2153. doi: 10.1016/j.tjnut.2025.05.003. Epub 2025 May 12.
Machine learning (ML) use in health research is growing, yet its application to predict cognitive outcomes using diverse health indicators is underinvestigated.
We used ML models to predict cognitive performance based on a set of health and behavioral factors, aiming to identify key contributors to cognitive function for insights into potential personalized interventions.
Data from 374 adults aged 19-82 y (227 females) were used to develop ML models predicting cognitive performance (reaction time in milliseconds) on a modified Eriksen flanker task. Features included demographics, anthropometric measures, dietary indices (Healthy Eating Index, Dietary Approaches to Stop Hypertension, Mediterranean, and Mediterranean-Dietary Approaches to Stop Hypertension Intervention for Neurodegenerative Delay), self-reported physical activity, and systolic and diastolic blood pressures. The data set was split (80:20) for training and testing. Predictive models (decision trees, random forest, AdaBoost, XGBoost, gradient boosting, linear, ridge, and lasso regression) were used with hyperparameter tuning and crossvalidation. Feature importance was calculated using permutation importance, whereas performance using mean absolute error (MAE) and mean squared error.
Random forest regressor exhibited the best performance, with the lowest MAE (training: 0.66 ms; testing: 0.78 ms) and mean squared error (training: 0.70 ms; testing: 1.05 ms). Age was the most significant feature (score: 0.208), followed by diastolic blood pressure (0.169), BMI (0.079), systolic blood pressure (0.069), and Healthy Eating Index (0.048). Ethnicity (0.005) and sex (0.003) had minimal predictive effect.
Age, blood pressure, and BMI show strong associations with cognitive performance, whereas diet quality has a subtler effect. These findings highlight the potential of ML models for developing personalized interventions and preventive strategies for cognitive decline.
机器学习（ML）在健康研究中的应用正在增加，但其使用多种健康指标预测认知结果的应用研究不足。
我们使用ML模型基于一组健康和行为因素预测认知表现，旨在确定认知功能的关键影响因素，以洞察潜在的个性化干预措施。
来自374名年龄在19 - 82岁的成年人（227名女性）的数据用于开发预测改良埃里克森侧翼任务中认知表现（以毫秒为单位的反应时间）的ML模型。特征包括人口统计学、人体测量指标、饮食指数（健康饮食指数、终止高血压膳食方法、地中海饮食以及用于神经退行性延迟的地中海 - 终止高血压膳食方法干预）、自我报告的身体活动以及收缩压和舒张压。数据集按80:20分割用于训练和测试。使用预测模型（决策树、随机森林、AdaBoost、XGBoost、梯度提升、线性、岭回归和套索回归）并进行超参数调整和交叉验证。使用排列重要性计算特征重要性，而使用平均绝对误差（MAE）和均方误差评估性能。
随机森林回归器表现最佳，MAE最低（训练：0.66毫秒；测试：0.78毫秒），均方误差也最低（训练：0.70毫秒；测试：1.05毫秒）。年龄是最显著的特征（得分：0.208），其次是舒张压（0.169）、BMI（0.079）、收缩压（0.069）和健康饮食指数（0.048）。种族（0.005）和性别（0.003）的预测作用最小。
年龄、血压和BMI与认知表现密切相关，而饮食质量的影响较细微。这些发现凸显了ML模型在开发针对认知衰退的个性化干预措施和预防策略方面的潜力。
