Predicting clinical outcomes in neuroblastoma with genomic data integration

Biology Direct

Table 2 Predicting OS and EFS in Versteeg cohort using models trained from SEQC cohort

The first column displays the details about the training and test sets. All→All indicates that we used the whole SEQC data for training and the whole Versteeg data for testing. All→All with only HR patients corresponds to the same model as All→All; however, here the performance metrics are only calculated for HR patients. In the third row, we used only the HR patients within the SEQC data for training and similarly we tested only on HR patients within the Versteeg data. In the last row, we only consider the patients with no MYCN amplification for training and testing sets. The Type column indicates the details of the chosen model. For gene expression, microarray data and two versions of the RNA-seq data were used. As such, this entry shows the type of the gene expression data used for the best trained model. The same entry also includes the C parameter of the SVM model and the type of the class weights (balanced or uniform)

ISSN: 1745-6150