On BalanceVar data, model (ten-fold cross validation) developed on training data achieved maximum MCC of 0.42, 0.61 and 0.39 while MCC of 0.37, 0.63 and 0.49 were achieved on evaluation datasets of IgG, IgE and IgA classes respectively.
MCC of the right third finger and metastatic tumor of the liver were suspected.
The average MCC of all algorithms is 0.49 (±0.04), underlining the predictive power of most methods.
Surprisingly LR performed above the average (MCC of 0.51 ± 0.06 and BEDROC 0.88 ± 0.06).
To evaluate the clinical and histopathologic characteristics of Merkel cell carcinoma (MCC) of the eyelid.
Moreover, judged on MCC the QSAR implementation of RF outperforms SVM (MCC of 0.56 ± 0.07 vs. 0.50 ± 0.07).
SVM (MCC of 0.22 ± 0.07 and BEDROC 0.69 ± 0.07) performed in between the DNNs and RF models.
The best method overall is the DNN_MC with an MCC of 0.57 (±0.07), and a BEDROC score of 0.92 (±0.05).
BetaTPred2 had an even bigger decrease in MCC of 0.06.
As a result, the MCC of BLAST was only ∼0.3.
The predictor assigned correctly 22 out of 24 organisms (91.7%), with a MCC of 0.837.
