Abstract
Objective
To assess the effectiveness of a machine learning framework and nomogram in predicting progression-free survival (PFS) post radical gastrectomy in patients with dMMR.
Method
An observational study conducted at Shanxi Cancer Hospital from 2002 to 2020 focused on developing and evaluating three machine learning models and nomogram to forecast PFS in patients undergoing radical gastrectomy for nonmetastatic gastric cancer with dMMR. Independent risk factors were identified using Cox regression analysis to develop the nomogram. The performance of the models was assessed through C-index, time receiver operating characteristic (T-ROC) curves, calibration curves, and decision curve analysis (DCA) curves in both training and validation cohorts. Subsequently, patients were categorized into high-risk and low-risk groups based on the nomogram's risk scores.
Results
Among the 582 patients studied, machine learning models exhibited higher c-index values compared to the nomogram. RSF demonstrated the highest c-index (0.968), followed by XG boosting (0.945), DST (0.924), the nomogram (0.808), and 8th TNM staging (0.757). Age, positive lymph nodes, neural invasion, and Ki67 were identified as key factors and integrated into the prognostic nomogram. Calibration and DCA curves provided evidence of the accuracy and clinical benefits of both machine learning and nomogram models.
Conclusion
Our study first successfully developed and validated machine learning and nomogram model based on clinical parameters for predicting 3-, 5-year PFS among dMMR gastric patients following gastrectomy. The nomogram exhibited a remarkable capability in identifying high-risk patients, furnishing clinicians with invaluable insights for postoperative surveillance and tailored therapeutic interventions.