Artificial Intelligence-Assisted Risk Prediction of Postoperative Pulmonary Complications in Non-Small Cell Lung Cancer Surgery

Background: Pulmonary complications are the most frequent adverse events following surgery for non-small cell lung cancer (NSCLC), influencing both short-term clinical outcomes and long-term prognosis. This study aimed to develop and evaluate artificial intelligence (AI) models to predict postoperative pulmonary complications in patients undergoing surgical treatment for NSCLC. Material and Methods: A total of 953 patients who underwent lung resection and mediastinal lymph node dissection for NSCLC between 2001 and 2023 were retrospectively analyzed. Clinical, laboratory, respiratory function, tumor-related radiological, surgical, and pathological parameters served as input variables, while the occurrence of postoperative pulmonary complications constituted the output variable. A fully connected deep neural network was employed, using 10-fold cross-validation. Results: For the training dataset, the model achieved a sensitivity of 66.4%, a positive predictive value (PPV) of 89.8%, and an accuracy of 88.6%. The algorithm's F1 1 score for the training data was 76.3%, its F1 0 score was 92.5%, and its F1 average score was 84.4%. The algorithm's sensitivity for the test data was 65.4%, its positive predictive value was 100%, and its accuracy was 90.4%. The F1 1 score was 79.1%, the test F1 0 score was 93.8%, and its F1 average score was 86.4%. The area under the receiver operating characteristic (ROC) curve (AUC) for the test dataset was 0.84. Conclusions: Accurate prediction of postoperative pulmonary complications in NSCLC surgery is crucial for optimizing perioperative care and reducing morbidity. The proposed deep learning model demonstrates promising predictive performance, enabling stratification of patients into high- and low-risk groups, and may serve as a valuable decision-support tool for clinicians.