There exist few maximal oxygen uptake (VO) non-exercise-based prediction equations, fewer using machine learning (ML), and none specifically for older adults. Since direct measurement of VO is infeasible in large epidemiologic cohort studies, we sought to develop, validate, compare, and assess the transportability of several ML VO prediction algorithms. The Baltimore Longitudinal Study of Aging (BLSA) participants with valid VO tests were included ( = 1080). Least absolute shrinkage and selection operator, linear- and tree-boosted extreme gradient boosting, random forest, and support vector machine (SVM) algorithms were trained to predict VO values. We developed these algorithms for: (a) for the overall BLSA, (b) by sex, (c) using all BLSA variables, and (d) for variables common in aging cohorts. Finally, we quantified the associations between measured and predicted VO and mortality. The age was 69.0 ± 10.4 years (mean ± SD) and the measured VO was 21.6 ± 5.9 mL/kg/min. Least absolute shrinkage and selection operator, linear- and tree-boosted extreme gradient boosting, random forest, and support vector machine yielded root mean squared errors of 3.4 mL/kg/min, 3.6 mL/kg/min, 3.4 mL/kg/min, 3.6 mL/kg/min, and 3.5 mL/kg/min, respectively. Incremental quartiles of measured VO showed an inverse gradient in mortality risk. Predicted VO variables yielded similar effect estimates but were not robust to adjustment. Measured VO is a strong predictor of mortality. Using ML can improve the accuracy of prediction as compared to simpler approaches but estimates of association with mortality remain sensitive to adjustment. Future studies should seek to reproduce these results so that VO, an important vital sign, can be more broadly studied as a modifiable target for promoting functional resiliency and healthy aging.

中文翻译：

---

几种针对老年人的机器学习、非运动型 VO2max 预测模型的开发、验证和可移植性

几乎不存在非基于运动的最大摄氧量 (VO) 预测方程，使用机器学习 (ML) 的预测方程较少，而且没有专门针对老年人的预测方程。由于在大型流行病学队列研究中直接测量 VO 是不可行的，因此我们试图开发、验证、比较和评估几种 ML VO 预测算法的可移植性。巴尔的摩老龄化纵向研究 (BLSA) 参与者进行了有效的 VO 测试 (= 1080)。训练最小绝对收缩和选择算子、线性和树增强极端梯度增强、随机森林和支持向量机 (SVM) 算法来预测 VO 值。我们开发这些算法的目的是：(a) 总体 BLSA，(b) 按性别，(c) 使用所有 BLSA 变量，以及 (d) 老龄群体中常见的变量。最后，我们量化了测量和预测的摄氧量与死亡率之间的关联。年龄为 69.0 ± 10.4 岁（平均值 ± SD），测量的 VO 为 21.6 ± 5.9 mL/kg/min。最小绝对收缩和选择算子、线性和树增强极端梯度增强、随机森林和支持向量机产生的均方根误差为 3.4 mL/kg/min、3.6 mL/kg/min、3.4 mL/kg/min 、 3.6 mL/kg/min 和 3.5 mL/kg/min 分别。测量的 VO 增量四分位数显示死亡风险呈反梯度。预测的 VO 变量产生了类似的效果估计，但调整后不稳健。测量的摄氧量是死亡率的有力预测指标。与更简单的方法相比，使用机器学习可以提高预测的准确性，但与死亡率相关的估计仍然对调整敏感。未来的研究应该寻求重现这些结果，以便更广泛地研究摄氧量这一重要的生命体征，将其作为促进功能弹性和健康衰老的可修改目标。