Development of a machine learning-based model to predict prognosis of resected invasive pulmonary adenocarcinoma.

Background: Invasive pulmonary adenocarcinoma (IPA) poses a significant threat to global health and patients still experience tumor recurrence and metastasis. This study aimed to construct an optimized prognosis model using machine learning to predict the disease-free survival (DFS) of IPA patients. Methods: A total of 670 resected IPA patients from 2015 to 2020 were enrolled. Clinicopathological information was collected and the outcomes of patients were followed up. Patients were divided into a training set and a test set at a ratio of 4:1. Four machine learning models were compared to build the DFS models and 5-fold cross validation was performed. The area under the receiver operating characteristic curve (AUC), C-index, calibration curves, and decision curve analysis (DCA) were used to evaluate the model. Results: Among the four models, the least absolute shrinkage and selection operator (Lasso) model showed the best performance in predicting DFS at 2-year (training set: AUC =0.906, test set: AUC =0.862), at 3-year (training set: AUC =0.894, test set: AUC =0.879), at 4-year (training set: AUC =0.901, test set: AUC =0.902), and at 5-year (training set: AUC =0.927, test set: AUC =0.887). The calibration curves and DCA exhibited a good predictive performance. Conclusions: Our study successfully constructed a machine-learning based prognostic model to predict DFS, which may provide oncologists with an effective tool for early medical intervention and survival improvement.