Therefore, molecular markers derived from a deeper understanding of pathogenesis are necessary to achieve a more accurate and reproducible classification of MPN according to their risk of fibrotic progression.
Percentage dense area estimation using a VAS appears more reproducible than the BIRADS classification.
Further applications are needed to establish whether the present method that relies on compartment models leads always to more clinically useful classifications and more reproducible diagnostic accuracies than methods that rely on non-biological models.
The authors showed that subnetwork markers are more reproducible and achieve higher accuracy in the classification than individual marker genes selected without network information.
Several studies [ 13, 14] have shown that pathway-based classifiers are more reproducible and often achieve comparable or better classification performance than classifier based on independent gene biomarkers [ 15].
In addition to more reproducible ICRS scoring, cross-sectional OCT images allow detection and classification of subsurface lesions [ 12].
The subnetwork markers were proved to be more reproducible than individual marker genes and achieved higher accuracy in the classification.
As we have demonstrated in this paper, the LLR-based pathway activity inference scheme significantly improves the discriminative power of the pathway markers, increases the overall classification accuracy, and finds reliable pathway markers that are more reproducible across different datasets.
We apply the proposed method to the classification of breast cancer metastasis, and show that it achieves higher accuracy and identifies more reproducible pathway markers compared to several existing pathway activity inference methods.
