Scholarlink

Purpose <p>Prehabilitation prior to colorectal cancer (CRC) surgery aims to improve functional capacity and postoperative outcomes, although results remain inconsistent due to variable programs. This study aimed to develop a predictive model for improvement in functional capacity during prehabilitation, using 6-min walk distance (6MWD) as the primary outcome. Identifying patients most likely to improve functional capacity during prehabilitation could support a more personalized use of prehabilitation.</p> Methods <p>We conducted a retrospective observational study in CRC patients who underwent prehabilitation. Linear mixed-effects (lmer), linear (lm), generalized linear (glm) models, and machine learning algorithms (logistic regression, decision tree, bagged tree, boosted tree, random forest, and support vector machines) were compared to predict absolute and relative improvements and 14-, 20-, and 400-m pass in 6MWD outtake values. Model performance was assessed with ROC-AUC, confusion matrices, and Matthews correlation coefficient.</p> Results <p>An lmer-based prediction model containing baseline 6MWD meters walked, sex, frailty, age, Charlson comorbidity index, and weight loss as fixed effects including interaction between frailty and age showed the best fit for predicting baseline and outtake 6MWD meters walked. A glm-based prediction model containing baseline 6MWD and age showed the best fit for predicting reaching the 400-m outtake threshold. Smaller improvements were not reliably predicted.</p> Conclusion <p>Baseline and outtake 6MWD, as well as passing the 400-m threshold at outtake, can be reasonably estimated. Combining baseline 6MWD with the predicted probability of achieving the 400-m threshold identifies patients who are most likely to improve their functional capacity, thereby supporting personalized prehabilitation care.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Predicting which colorectal cancer patients are most likely to improve their functional capacity with pre-surgery prehabilitation: a retrospective study based on the 6-min walk distance

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月27日

查看全文
  • M. de Klerk,
  • M. J. W. van der Linden,
  • A. P. M. Kerckhoffs,
  • B. R. Meijboom,
  • E. G. G. Verdaasdonk,
  • E. de Vries

摘要

Purpose

Prehabilitation prior to colorectal cancer (CRC) surgery aims to improve functional capacity and postoperative outcomes, although results remain inconsistent due to variable programs. This study aimed to develop a predictive model for improvement in functional capacity during prehabilitation, using 6-min walk distance (6MWD) as the primary outcome. Identifying patients most likely to improve functional capacity during prehabilitation could support a more personalized use of prehabilitation.

Methods

We conducted a retrospective observational study in CRC patients who underwent prehabilitation. Linear mixed-effects (lmer), linear (lm), generalized linear (glm) models, and machine learning algorithms (logistic regression, decision tree, bagged tree, boosted tree, random forest, and support vector machines) were compared to predict absolute and relative improvements and 14-, 20-, and 400-m pass in 6MWD outtake values. Model performance was assessed with ROC-AUC, confusion matrices, and Matthews correlation coefficient.

Results

An lmer-based prediction model containing baseline 6MWD meters walked, sex, frailty, age, Charlson comorbidity index, and weight loss as fixed effects including interaction between frailty and age showed the best fit for predicting baseline and outtake 6MWD meters walked. A glm-based prediction model containing baseline 6MWD and age showed the best fit for predicting reaching the 400-m outtake threshold. Smaller improvements were not reliably predicted.

Conclusion

Baseline and outtake 6MWD, as well as passing the 400-m threshold at outtake, can be reasonably estimated. Combining baseline 6MWD with the predicted probability of achieving the 400-m threshold identifies patients who are most likely to improve their functional capacity, thereby supporting personalized prehabilitation care.