Deep learning based digital pathology for predicting treatment response to first-line PD-1 blockade in advanced gastric cancer

Table 2 Comparison of Models’ metrics and selection on the internal test dataset

	ACC	AUC	PPV	NPV	SENS	SPEC	F1 score
DenseNet121	0.691	0.776	0.669	0.729	0.816	0.552	0.735
EfficientNet-B4	0.701	0.778	0.699	0.703	0.758	0.638	0.727
Swin Transformer-V2	0.690	0.795	0.656	0.767	0.864	0.496	0.746
Ensemble Model	0.715	0.805	0.689	0.760	0.834	0.583	0.755

ACC: accuracy; AUC: area under the curve; PPV: positive predictive value; NPV: negative prediction value; SENS: sensitivity; SPEC: specificity

ISSN: 1479-5876