Using the optimal threshold (0.74) accurate class predictions were made for 77% of the test cases with sensitivity = 0.70, specificity = 1.00.By using Bayesian probit regression to develop a model, we show that it is entirely possible to make a diagnosis of IPF from the peripheral blood with gene signatures.
The total areas of mitochondrial intensity in 300 400 cells were extracted using the optimal threshold parameters and calculated with LuminaVision image analysis software.
Using the optimal threshold of 0.30 determined by the ROC curve (Figure S5), 76.2% of individuals were correctly classified (79.6% sensitivity, 70.7% specificity, Table 8).
The total areas of nitrotyrosine in 300 400 cells were extracted using the optimal threshold parameters and calculated with LuminaVision image analysis software.
The total areas of MitoSOX-Red or DHR123 intensity in 300 400 cells were extracted using the optimal threshold parameters and calculated with LuminaVision image analysis software.
GENN performed better than logistic regression using the arbitrary 0.5 threshold with 63.0% of individuals correctly classified, but was not as successful as logistic regression using the optimal threshold.
Using the optimal threshold value of the Whitehead dataset, 68 out of 100 true peaks were successfully predicted by Splitter.
Using the optimal threshold (0.74) accurate class predictions were made for 77% of the test cases with sensitivity = 0.70, specificity = 1.00.
It is important to note that the results presented in this section were obtained by using the optimal threshold values (I0) for all algorithms (C3NET, ARACNE and MRNET).
On p. 7 the authors state "The DPI tolerance parameter of ARACNE, when used for comparison purposes, was chosen as 0.1 [ 30]," but later they state "the results presented in this section were obtained by using the optimal threshold values (I0) for all algorithms (C3NET, ARACNE and MRNET)." These statements seem inconsistent.
