Clinical Application of Artificial Intelligence in Prediction of Intraoperative Cerebrospinal Fluid Leakage in Pituitary Surgery: A Systematic Review and Meta-Analysis

Abstract

Background

Postoperative cerebrospinal fluid (CSF) leakage is the leading adverse event in transsphenoidal surgery. Intraoperative CSF (ioCSF) leakage is one of the most important predictive factors for postoperative CSF leakage. This systematic review and meta-analysis aimed to evaluate the effectiveness of artificial intelligence (AI) models in predicting ioCSF.

Methods

Literature records were retrieved on June 13, 2024, using the relevant key terms without filters in PubMed, Embase, Scopus, and Web of Science. Records were screened according to the eligibility criteria, and the data from the included studies were extracted. The quality assessment was performed using the Quality Assessment of Diagnostic Accuracy Studies–2 tool. The meta-analysis, sensitivity analysis, and meta-regression were conducted using R software.

Results

Our results demonstrate that the AI models achieved a pooled sensitivity of 93.4% (95% confidence interval [CI]: 74.8%–98.6%) and specificity of 91.7% (95% CI: 75%–97.6%). The subgroup analysis revealed that the pooled sensitivities in machine learning and deep learning were 86.2% (95% CI: 83%–88.8%) and 99% (95% CI: 93%–99%), respectively (P < 0.01). The subgroup analysis demonstrated a pooled specificity of 92.1% (95% CI: 63.1%–98.7%) for machine learning and 90.6% (95% CI: 78.2%–96.3%) for deep learning models (P = 0.87). The diagnostic odds ratio meta-analysis revealed an odds ratio 114.6 (95% CI: 17.6–750.9). The summary receiver operating characteristic curve demonstrated that the overall area under the curve of the studies was 0.955, which is a considerable performance.

Conclusions

AI models have demonstrated promising performance for predicting the ioCSF leakage in pituitary surgery and can optimize the treatment strategy.