Risk stratification and pathway analysis based on graph neural network and interpretable algorithm

BMC Bioinformatics

Table 1 Comparison of predictive performance with benchmark methods

Cancer	PathGNN	PGDNN	DNN	RF	LR
LUAD	0.752 ± 0.046	0.592 ± 0.026	0.536 ± 0.059	0.566 ± 0.043	0.586 ± 0.057
SKCM	0.702 ± 0.041	0.626 ± 0.035	0.590 ± 0.031	0.601 ± 0.055	0.588 ± 0.056
LGG	0.862 ± 0.058	0.852 ± 0.082	0.834 ± 0.080	0.808 ± 0.083	0.755 ± 0.068
KIRC	0.754 ± 0.071	0.683 ± 0.045	0.584 ± 0.088	0.608 ± 0.042	0.604 ± 0.027

The predictive performance of PathGNN was compared with pathway-guided deep neutral network (PGDNN), deep neural network (DNN), random forest (RF) and logistic regression (LR). The area under the curve (AUC: mean ± standard deviation) was recorded
Bold indicated the best predictive performance in each case study

ISSN: 1471-2105