Machine Learning: A Novel Approach for Predicting Visual Outcomes and Factors Affecting it in Patients with Pituitary...

Area of Science:

• Neurosurgery
• Ophthalmology
• Machine Learning

Background:

• Pituitary adenomas can cause significant visual deficits.
• Predicting visual outcomes after surgery is crucial for patient management.

Purpose of the Study:

• To develop and evaluate a machine learning (ML) model for predicting visual outcomes in patients with pituitary adenomas.
• To identify key factors influencing postoperative visual recovery.

Main Methods:

• A retrospective analysis of 284 pituitary adenoma patients with preoperative visual deficit.
• Utilized Weka software with J48 trees, LMT, REP tree, and Random Forest algorithms for classification.
• Collected patient variables including tumor volume, visual acuity, and symptom duration.

Main Results:

• Vision improved in 89.78% of patients post-surgery.
• Key predictors of visual outcome included extent of resection, preoperative visual acuity, tumor volume, and symptom duration.
• The ML model achieved 88.94% accuracy, with an AUC of 0.846.

Conclusions:

• Machine learning offers a promising tool for predicting visual outcomes after pituitary adenoma surgery.
• Accurate prediction can enhance patient-specific care and counseling.
• The developed ML model demonstrates high accuracy in forecasting visual recovery.