Scholarlink

Background <p>This study aimed to develop and evaluate machine learning (ML) models for the perioperative prediction of orchiectomy in patients with testicular torsion.</p> Methods <p>We conducted a retrospective analysis of 204 patients with intraoperatively confirmed testicular torsion who underwent surgical exploration between January 2003 and April 2025. The patient cohort was partitioned into a training set (70%) and a testing set (30%) using stratified sampling. Least Absolute Shrinkage and Selection Operator (LASSO) regression was employed to identify seven predictors. Subsequently, five ML models were developed using these predictors. Five-fold cross-validation was applied during model development in the training set, and the final models were evaluated in the testing set. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, and specificity. Shapley Additive Explanations (SHAP) were used to assess model interpretability and quantify the contribution of each feature to model predictions.</p> Results <p>Among the five ML models, the Light Gradient Boosting Machine (LightGBM) model demonstrated the best overall predictive performance in the testing set, with an AUC of 0.8902 (95% confidence interval [CI]: 0.7937–0.9866) and an accuracy of 0.8710. SHAP analysis identified symptom duration as the most influential predictor, followed by fibrinogen, markedly reduced or absent testicular blood flow on initial ultrasonography, degree of torsion, manual detorsion at first presentation, lymphocyte-to-monocyte ratio (LMR), and abdominal pain.</p> Conclusion <p>We developed and internally validated interpretable ML models for predicting orchiectomy risk in patients with testicular torsion. The LightGBM model, incorporating seven predictors, showed promising performance for perioperative risk assessment. Further external validation is required before routine clinical implementation.</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

Using interpretable machine learning model to predict orchiectomy after testicular torsion

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月24日

查看全文
  • Pengfeng Gong,
  • Xuan Wen,
  • Jun Zhou,
  • Cheng Chen,
  • Zinong Tian,
  • You Zhao,
  • Tianwei Zhang

摘要

Background

This study aimed to develop and evaluate machine learning (ML) models for the perioperative prediction of orchiectomy in patients with testicular torsion.

Methods

We conducted a retrospective analysis of 204 patients with intraoperatively confirmed testicular torsion who underwent surgical exploration between January 2003 and April 2025. The patient cohort was partitioned into a training set (70%) and a testing set (30%) using stratified sampling. Least Absolute Shrinkage and Selection Operator (LASSO) regression was employed to identify seven predictors. Subsequently, five ML models were developed using these predictors. Five-fold cross-validation was applied during model development in the training set, and the final models were evaluated in the testing set. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, and specificity. Shapley Additive Explanations (SHAP) were used to assess model interpretability and quantify the contribution of each feature to model predictions.

Results

Among the five ML models, the Light Gradient Boosting Machine (LightGBM) model demonstrated the best overall predictive performance in the testing set, with an AUC of 0.8902 (95% confidence interval [CI]: 0.7937–0.9866) and an accuracy of 0.8710. SHAP analysis identified symptom duration as the most influential predictor, followed by fibrinogen, markedly reduced or absent testicular blood flow on initial ultrasonography, degree of torsion, manual detorsion at first presentation, lymphocyte-to-monocyte ratio (LMR), and abdominal pain.

Conclusion

We developed and internally validated interpretable ML models for predicting orchiectomy risk in patients with testicular torsion. The LightGBM model, incorporating seven predictors, showed promising performance for perioperative risk assessment. Further external validation is required before routine clinical implementation.