Advanced machine learning outperform traditional models in gastric cancer prognosis

A new machine learning framework combined with a nomogram has been developed to predict recurrence after radical gastrectomy in patients with non-metastatic gastric cancer and “double invasion.” Among the 559 patients studied, the random survival forests model achieved the highest c-index of 0.791, outperforming other machine learning approaches. This hybrid model effectively integrates multiple variables to enhance predictive accuracy and guide postoperative care, marking a significant improvement over conventional prognostic models.