Predicting Overall Survival in Patients with Nonmetastatic Gastric Signet Ring Cell Carcinoma: A Machine Learning Approach

Table 1

Performance of machine learning models in OS.

Model	Training cohort	Testing cohort
	AUC with 95% CI	AUC with 95% CI
1-year survival
KNN	0.773 (0.747-0.799)	0.715 (0.669-0.760)
Support vector machines	0.784 (0.757-0.810)	0.738 (0.695-0.782)
Random forest	0.998 (0.997-0.999)	0.725 (0.681-0.770)
XGBoost	0.842 (0.819-0.863)	0.749 (0.708-0.791)
Neural network	0.789 (0.764-0.815)	0.706 (0.662-0.751)
3-year survival
KNN	0.801 (0.779-0.823)	0.800 (0.766-0.835)
Support vector machines	0.795 (0.773-0.818)	0.812 (0.779-0.846)
Random forest	0.997 (0.995-0.998)	0.807 (0.773-0.841)
XGBoost	0.831 (0.811-0.852)	0.823 (0.790-0.854)
Neural network	0.814 (0.792-0.836)	0.765 (0.729-0.802)
5-year survival
KNN	0.813 (0.791-0.836)	0.765 (0.725-0.806)
Support vector machines	0.813 (0.789-0.836)	0.774 (0.733-0.815)
Random forest	0.996 (0.994-0.998)	0.774 (0.734-0.814)
XGBoost	0.838 (0.816-0.858)	0.829 (0.793-0.863)
Neural network	0.811 (0.787-0.836)	0.776 (0.737-0.815)