Predictive Model for Preoperative Risk Assessment of Prolonged Pleural Effusion Following Lobectomy

Introduction. To develop a nomogram and a preoperative risk calculator for predicting prolonged pleural effusion (PPE) in patients undergoing lobectomy for lung cancer, aimed at early prevention of this complication. Materials and methods. The study was conducted at the Thoracic Oncology Surgery Department of the Regional Oncology Dispensary in Ulyanovsk, Russia. It included patients who underwent lobectomy between January 1, 2023, and January 1, 2025 (State Registration Certificate № 2025621960 for the database). A total of 215 patients were analyzed, of whom 115 did not develop PPE and 100 experienced the complication. After statistical analysis, bootstrap resampling was applied and logistic regression calibration was performed using artificial intelligence-based machine learning techniques. Results. The incidence of prolonged pleural effusion was 100 out of 215 patients (46,5 %). Multivariate analysis identified significant predictors: preoperative white blood cell, neutrophil, and lymphocyte counts; Tiffeneau index; and video-assisted thoracoscopic access. The developed predictive model demonstrated high sensitivity (85 %) and specificity (61,7 %). Conclusion. The incidence of prolonged pleural effusion in this study was 46.5 %. A predictive model and interactive risk calculator for PPE following lobectomy have been developed.