Artificial image objects for classification of breast cancer biomarkers with transcriptome sequencing data and convolutional neural network algorithms

Table 3 Cross validation and sample testing results for NHG

	Accuracy^a	AUC^b	Precision	Recall	F1 Score
Cross validation
Grade I	0.838 ± 0.085	0.974 ± 0.005	0.913 ± 0.028	0.838 ± 0.085	0.871 ± 0.037
Grade II	0.809 ± 0.072	0.880 ± 0.012	0.686 ± 0.07	0.811 ± 0.072	0.738 ± 0.026
Grade III	0.825 ± 0.038	0.938 ± 0.004	0.863 ± 0.031	0.756 ± 0.072	0.803 ± 0.031
Weighted average	0.820 ± 0.012	0.931 ± 0.006	0.820 ± 0.012	0.802 ± 0.033	0.804 ± 0.030
Sample testing on GSE81538
Grade I	0.406 ± 0.081	0.873 ± 0.025	0.608 ± 0.116	0.408 ± 0.082	0.475 ± 0.059
Grade II	0.743 ± 0.069	0.833 ± 0.005	0.710 ± 0.026	0.745 ± 0.070	0.725 ± 0.026
Grade III	0.872 ± 0.029	0.928 ± 0.015	0.848 ± 0.022	0.873 ± 0.030	0.858 ± 0.015
Weighted average	0.764 ± 0.052	0.882 ± 0.012	0.762 ± 0.035	0.765 ± 0.052	0.758 ± 0.025
Sample testing on GSE163882
Grade I	0 ± 0	0.622 ± 0.189	0 ± 0	0 ± 0	0 ± 0
Grade II	0.016 ± 0.006	0.564 ± 0.044	0.268 ± 0.133	0.016 ± 0.006	0.030 ± 0.010
Grade III	0.974 ± 0.012	0.596 ± 0.070	0.589 ± 0.003	0.974 ± 0.012	0.734 ± 0.004
Weighted average	0.580 ± 0.006	0.587 ± 0.038	0.437 ± 0.045	0.330 ± 0.003	0.443 ± 0.004

ISSN: 1465-542X