Post-Operative Outcome Predictions in Vestibular Schwannoma Using Machine Learning Algorithms

Area of Science:

• Neurosurgery
• Medical Informatics
• Machine Learning

Background:

• Vestibular schwannoma (VS) surgery carries risks of unplanned reoperations and complications.
• Predictive models are needed to identify patients at higher risk for adverse outcomes.
• Personalized medicine approaches can improve patient management and outcomes.

Purpose of the Study:

• To develop and validate a machine learning (ML) algorithm for predicting unplanned reoperations, surgical complications, and medical complications following VS surgery.
• To identify key pre- and peri-operative variables influencing these adverse outcomes.
• To provide clinicians with a tool for enhanced risk stratification and personalized patient care.

Main Methods:

• Utilized the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) database, including 110 pre- and peri-operative variables.
• Developed a seven-layer deep neural network model using the Keras library with a 70:30 training-testing split.
• Assessed feature importance using permutation effects to understand variable contributions to the ML model's predictions.

Main Results:

• The study analyzed 1783 patients undergoing VS surgery, with unplanned reoperation rates of 8.5%, surgical complications at 5.2%, and medical complications at 6.2%.
• The deep neural network achieved ROC-AUC values of 0.6315 for reoperation, 0.7939 for medical complications, and 0.719 for surgical complications.
• Key predictive variables included length of stay post-operation, days from operation to discharge, and total hospital length of stay.

Conclusions:

• An effective ML algorithm was developed to predict unplanned reoperations and both surgical and medical complications after VS surgery.
• The model demonstrates potential for guiding physicians in anticipating post-surgical outcomes.
• This predictive capability can facilitate the creation of personalized medical care plans for patients with vestibular schwannoma.