Convolutional Graph Isomorphism Network to Detect Glaucomatous Visual Field Defects.

Although the difference in AUC compared to GIN TD (AUC: 0.976, P = 0.339) and GIN PD (AUC: 0.975, P = 0.330) did not reach statistical significance, the full GIN model achieved superior sensitivity at 95% specificity (94.1% vs. 93.5% and 93.9%, respectively), higher F1-score (0.896 vs. 0.806 and 0.894, respectively), and better repeatability than GIN TD (93.0% vs. 92.1%), indicating better balanced performance for clinical application.

also in 111,027 other papers