Machine Learning Model for Predicting Operative Duration in Transsphenoidal Neuroendoscopic Pituitary Adenomectomy

Abstract Aim To develop and validate a machine learning model predicting surgical duration for transsphenoidal endoscopic pituitary tumor resection, and assess its utility for nursing management. Background Accurate surgical duration prediction is vital for nursing resource allocation and OR scheduling, yet traditional methods often lack precision, leading to workflow inefficiencies and increased nursing burden. Methods Retrospective data from 100 patients (2016–2025) yielded 22 preoperative variables. Artificial neural network (ANN) and random forest models were developed and evaluated using R², MAE, RMSE, and clinical accuracy (± 30/±45 min). Results The ANN model outperformed random forest, achieving a test set MAE of 0.636 hours (~ 38 min). Clinically, 63.3% of predictions were within ± 30 min, and 76.7% within ± 45 min. Key predictors included tumor recurrence, distance from the third ventricle floor, and the cinch sign. Conclusion ANN enables accurate surgical duration prediction, providing objective evidence to optimize nursing schedules, reduce uncertainty, and alleviate perioperative nursing burden.