Scholarlink

Background <p>Myocardial injury after non-cardiac surgery (MINS) is a common complication following lung cancer surgery and is associated with adverse prognostic outcomes.&#xa0;Preoperative cardiac risk assessment is essential for optimizing perioperative management and improving patient outcomes. This study aimed to develop and validate an interpretable machine learning (ML) model using preoperative clinical features to predict MINS following lung cancer surgery.</p> Methods <p>Patients undergoing elective lung cancer surgery from two medical centers were included. Preoperative clinical data were retrospectively collected. Multiple feature selection methods were applied to identify the final predictors. Prediction models were constructed using eight ML algorithms and compared with the Revised Cardiac Risk Index (RCRI). Model performance was evaluated by the area under the receiver operating characteristic curve (AUC), calibration curves, and decision curve analysis. The SHapley Additive exPlanations (SHAP) framework was used to enhance model interpretability.</p> Results <p>A total of 1644 patients were included. Five preoperative predictors were identified: male sex, age, neoadjuvant therapy, leukocyte count, and high-sensitivity cardiac troponin T (hs-cTnT). In the external validation cohort, the extreme gradient boosting (XGB) model achieved the best generalization performance (AUC = 0.795), significantly outperforming the RCRI (AUC = 0.595). In the high-risk subgroup (RCRI score ≥ 2), the random forest model showed further improved discriminative ability (AUC = 0.837). SHAP analysis identified hs-cTnT and age as the most important predictors. The model was deployed as a user-friendly web-based calculator for clinical application.</p> Conclusions <p>This study successfully developed and validated an interpretable ML model using routinely available preoperative indicators, which significantly outperformed the conventional RCRI in predicting MINS after lung cancer surgery. This tool provides effective decision support by improving preoperative cardiac risk stratification.</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

Development and validation of a preoperative interpretable machine learning model for predicting myocardial injury following lung cancer surgery

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年8月4日

查看全文
  • Jingya Yuan,
  • Feng Chen,
  • Zongxiao Jia,
  • Weipeng Shao,
  • Hongbo Guo,
  • Zhan Liu

摘要

Background

Myocardial injury after non-cardiac surgery (MINS) is a common complication following lung cancer surgery and is associated with adverse prognostic outcomes. Preoperative cardiac risk assessment is essential for optimizing perioperative management and improving patient outcomes. This study aimed to develop and validate an interpretable machine learning (ML) model using preoperative clinical features to predict MINS following lung cancer surgery.

Methods

Patients undergoing elective lung cancer surgery from two medical centers were included. Preoperative clinical data were retrospectively collected. Multiple feature selection methods were applied to identify the final predictors. Prediction models were constructed using eight ML algorithms and compared with the Revised Cardiac Risk Index (RCRI). Model performance was evaluated by the area under the receiver operating characteristic curve (AUC), calibration curves, and decision curve analysis. The SHapley Additive exPlanations (SHAP) framework was used to enhance model interpretability.

Results

A total of 1644 patients were included. Five preoperative predictors were identified: male sex, age, neoadjuvant therapy, leukocyte count, and high-sensitivity cardiac troponin T (hs-cTnT). In the external validation cohort, the extreme gradient boosting (XGB) model achieved the best generalization performance (AUC = 0.795), significantly outperforming the RCRI (AUC = 0.595). In the high-risk subgroup (RCRI score ≥ 2), the random forest model showed further improved discriminative ability (AUC = 0.837). SHAP analysis identified hs-cTnT and age as the most important predictors. The model was deployed as a user-friendly web-based calculator for clinical application.

Conclusions

This study successfully developed and validated an interpretable ML model using routinely available preoperative indicators, which significantly outperformed the conventional RCRI in predicting MINS after lung cancer surgery. This tool provides effective decision support by improving preoperative cardiac risk stratification.