Improved prediction of postoperative pediatric cerebellar mutism syndrome using an artificial neural network

• Published in: Neuro-oncology advances Vol. 4; no. 1; p. vdac003
• Main Authors: Sidpra, Jai, Marcus, Adam P, Löbel, Ulrike, Toescu, Sebastian M, Yecies, Derek, Grant, Gerald, Yeom, Kristen, Mirsky, David M, Marcus, Hani J, Aquilina, Kristian, Mankad, Kshitij
• Format: Journal Article
• Language: English
• Published: US Oxford University Press 01-01-2022
• Subjects: Clinical Investigations; artificial neural network; magnetic resonance imaging; complications; postoperative pediatric cerebellar mutism syndrome; posterior fossa tumor
• ISSN: 2632-2498; 2632-2498
• DOI: 10.1093/noajnl/vdac003

Abstract
Background
Postoperative pediatric cerebellar mutism syndrome (pCMS) is a common but severe complication that may arise following the resection of posterior fossa tumors in children. Two previous studies have aimed to preoperatively predict pCMS, with varying results. In this work, we examine the generalization of these models and determine if pCMS can be predicted more accurately using an artificial neural network (ANN).
Methods
An overview of reviews was performed to identify risk factors for pCMS, and a retrospective dataset was collected as per these defined risk factors from children undergoing resection of primary posterior fossa tumors. The ANN was trained on this dataset and its performance was evaluated in comparison to logistic regression and other predictive indices via analysis of receiver operator characteristic curves. The area under the curve (AUC) and accuracy were calculated and compared using a Wilcoxon signed-rank test, with P < .05 considered statistically significant.
Results
Two hundred and four children were included, of whom 80 developed pCMS. The performance of the ANN (AUC 0.949; accuracy 90.9%) exceeded that of logistic regression (P < .05) and both external models (P < .001).
Conclusion
Using an ANN, we show improved prediction of pCMS in comparison to previous models and conventional methods.