Development and external validation of a machine learning model for predicting postoperative hydrocephalus in 1,073 posterior fossa tumor patients.

Authors

Ronghao Li, Dan Li, Li Liu, Wenyu Ji, Yuchun Pei, Run Wan, Dong Liu, Yongxin Wang, Suo Liu, Tingqin Huang, Ming Zhang, Xiaobin Liu, Chenghai Zuo, Chong Li, Xu Fang, Changlin Yin, Rong Hu, Zhao Yang, Liang Tan, Jingyu Chen

Published in

Neurosurgical review. Volume 49. Issue 1. Jul 30, 2026. Epub Jul 30, 2026.

Abstract

Postoperative hydrocephalus is a common complication following posterior fossa tumor resection, affecting 7-40% of patients. Although preoperative cerebrospinal fluid (CSF) diversion may be required in selected patients with hydrocephalus, decisions remain individualized in routine neurosurgical practice. We therefore developed and externally validated a model to provide supplementary preoperative risk stratification using routinely available variables. We retrospectively analyzed 1,073 patients following resection of posterior fossa tumors (PFTs) treated at five tertiary centers between 2013 and 2024, dividing them into a development cohort (n = 854) and an external validation cohort (n = 219). We initially screened 30 perioperative variables from the multicenter dataset and then selected a clinically implementable model restricted to variables available before tumor resection. Feature importance was ranked using Shapley additive explanations (SHAP), and seven distinct machine learning (ML) algorithms were evaluated. Model performance was comprehensively measured using the area under the receiver operating characteristic curve (AUC), sensitivity, specificity, precision-recall curves, and decision curve analysis. The final clinically implementable model included three preoperative variables: Evans index, tumor-fourth ventricle relationship, and preoperative cerebrospinal fluid diversion status. In the external validation cohort, the support vector machine model showed good external discrimination, with an area under the receiver operating characteristic curve of 0.877, accuracy of 81.3%, sensitivity of 80.8%, and specificity of 81.7%. Logistic regression achieved comparable discrimination. Calibration assessment suggested dataset shift between the development and external validation cohorts, indicating that absolute predicted probabilities should be interpreted cautiously in populations with different baseline risks. A concise three-variable preoperative model showed good external discrimination for clinically relevant postoperative hydrocephalus after posterior fossa tumor resection. The model may support preoperative risk communication and postoperative surveillance planning, but should not be used as a stand-alone indication for cerebrospinal fluid diversion. Prospective validation and local recalibration are warranted before routine clinical implementation.

PMID:
42530672
Bibliographic data and abstract were imported from PubMed on 30 Jul 2026.

Read full publication at:
Please sign in to see all details.

Stats

1-terrible, 9-excellent. How would you rate this publication? Sign in in to submit your rating.

Post a comment

You need to be signed in to post comments. You can sign in here.

Comments

There are no comments yet.