Karataş Levent, Orbak Yenidünya Esra Sena, Demirsoy Nesrin
Department of Physical Medicine and Rehabilitation, Faculty of Medicine, Gazi University, Ankara, Turkiye.
Turk J Med Sci. 2025 Jun 23;55(4):930-939. doi: 10.55730/1300-0144.6046. eCollection 2025.
BACKGROUND/AIM: Existing treadmill-based VO prediction models may not accurately estimate submaximal VO in patients with coronary artery disease (CAD), as they are often derived from healthy populations. This study aimed to develop and validate a submaximal VO prediction model derived from healthy individuals and tested for generalizability in CAD patients by incorporating clinically relevant parameters.
A retrospective analysis was conducted with 101 participants (54 healthy, 47 CAD patients) undergoing cardiopulmonary exercise testing using the modified Bruce protocol. To better represent the submaximal VO reached during exercise, the average VO in the last minute of each stage was used. The model was developed using data from healthy individuals and subsequently validated in the CAD cohort. A linear mixed-effects model was employed to predict VO based on speed, grade, and other confounders, including peak VO, body weight, and body mass index (BMI). The model's performance was evaluated and compared with previously published equations using Bland-Altman plots, mean absolute error (MAE), root mean square error (RMSE), and Lin's concordance correlation coefficient (CCC).
The final model, including speed, grade, and peak VO, achieved an R of 0.83 (95% CI: 0.79, 0.86; f = 4.88). For CAD patients, the predicted-actual VO difference was -0.05 ± 1.8 mL/kg/min, with MAE and RMSE values of 1.4 and 1.8 mL/kg/min, respectively. The model outperformed reference equations, achieving the highest accuracy (CCC = 0.923) and minimal bias. Incorporating peak VO effectively accounted for exercise response differences between healthy individuals and CAD patients.
A submaximal VO estimation model derived from healthy individuals and validated in CAD patients demonstrated high accuracy. Incorporating peak VO effectively bridged physiological differences, supporting individualized exercise prescriptions in cardiac rehabilitation. However, larger prospective cohorts are warranted to confirm external validity.
背景/目的：现有的基于跑步机的最大摄氧量（VO）预测模型可能无法准确估计冠状动脉疾病（CAD）患者的次极量VO，因为这些模型通常来源于健康人群。本研究旨在开发并验证一个源自健康个体的次极量VO预测模型，并通过纳入临床相关参数来测试其在CAD患者中的通用性。
对101名参与者（54名健康者，47名CAD患者）进行回顾性分析，这些参与者采用改良Bruce方案进行心肺运动测试。为了更好地代表运动过程中达到的次极量VO，使用每个阶段最后一分钟的平均VO。该模型使用健康个体的数据开发，随后在CAD队列中进行验证。采用线性混合效应模型，根据速度、坡度和其他混杂因素（包括峰值VO、体重和体重指数（BMI））预测VO。使用Bland-Altman图、平均绝对误差（MAE）、均方根误差（RMSE）和Lin一致性相关系数（CCC）评估模型性能，并与先前发表的方程进行比较。
最终模型包括速度、坡度和峰值VO，R值为0.83（95%CI：0.79，0.86；f = 4.88）。对于CAD患者，预测VO与实际VO的差值为-0.05±1.8 mL/kg/min，MAE和RMSE值分别为1.4和1.8 mL/kg/min。该模型优于参考方程，具有最高的准确性（CCC = 0.923）和最小的偏差。纳入峰值VO有效地解释了健康个体和CAD患者之间的运动反应差异。
源自健康个体并在CAD患者中验证的次极量VO估计模型显示出高准确性。纳入峰值VO有效地弥合了生理差异，支持心脏康复中的个性化运动处方。然而，需要更大的前瞻性队列来确认外部有效性。
