Skip to main content
. 2025 Jun 6;6(7):1283–1294. doi: 10.1038/s43018-025-00976-5

Table 2.

Performance of the crossNN pan-cancer model in several cohorts

Cohort Number of cases Metrics crossNN
450K 3,871 Accuracy 0.859
Precision 0.975
Sensitivity 0.654
AUC 0.901
EPICv1 554 Accuracy 0.949
Precision 0.962
Sensitivity 0.926
AUC 0.926
EPICv2 133 Accuracy 0.992
Precision 0.992
Sensitivity 0.977
AUC 0.939

Nanopore

R9

443 Accuracy 0.928
Precision 0.997
Sensitivity 0.772
AUC 0.947

Nanopore

R10

129 Accuracy 0.860
Precision 0.989
Sensitivity 0.713
AUC 0.949
Targeted sequencing 124 Accuracy 0.847
Precision 0.935
Sensitivity 0.806
AUC 0.895
WGBS 125 Accuracy 0.848
Precision 0.941
Sensitivity 0.640
AUC 0.860
Overall 5,379 Accuracy 0.877
Precision 0.978
Sensitivity 0.691
AUC 0.897

MCF-level raw accuracy before the application of cutoffs, precision with platform-specific cutoffs and AUC of the ROC curve for the classification score to predict the correct classification are given. For crossNN, the following cutoffs, as derived above, were used: microarray > 0.3; nanopore/targeted methyl-seq/WGBS > 0.15.