Skip to main content

Table 1 Performances of ICA-based feature extraction. Mean values of sensitivity and specificity are reported as well as class probability originated from random forest voting

From: Deconvolution of transcriptomes and miRNomes by independent component analysis provides insights into biological processes and clinical outcomes of melanoma patients

Predicted variables

Groups

Accuracy (st.dev.)

Sensitivity specificity

P2PM (prob.)

P4PM (prob.)

P6PM (prob.)

P4NS (prob.)

NHEM (prob.)

Gender

female: 179

0.996 (< 0.001)

0.994

0.994

female (0.73)

female (0.66)

male (0.79)

female (0.68)

female (0.67)

male: 293

Sample type

primary: 105

0.871 (0.003)

0.733

0.733

primary (0.68)

primary (0.55)

primary (0.65)

primary (0.59)

meta-static (0.51)

metastatic: 367

Subtype (RNA cluster)

immune: 170

0.902 (0.006)

0.877

0.945

keratin (0.64)

keratin (0.48)

keratin (0.61)

keratin (0.64)

keratin (0.55)

keratin: 102

MITF-low: 59