Table 9 Performance assessment of using the random forests (RF) models.

From: Machine learning-based prediction of drug response in ischemia reperfusion animal model

Model

Random forest (RF)

Metric

Complete

(all features)

Reduced model

(SFS)

Molecular

Accuracy

0.8765 ∓ 0.0155

0.8709 ∓ 0.0148

 

Precision

0.9402 ∓ 0.0288

0.9464 ∓ 0.0209

 

Recall

0.8786 ∓ 0.0503

0.8709 ∓ 0.0267

 

Specificity

0.8778 ∓ 0.0595

0.8944 ∓ 0.0388

 

MCC

0.7348 ∓ 0.0261

0.7244 ∓ 0. 0314

Biochemical

Accuracy

0.8877 ∓ 0.0210

0.8990 ∓ 0.0130

 

Precision

0.9183 ∓ 0.0300

0.9333 ∓ 0.0119

 

Recall

0.9182 ∓ 0.0096

0.9182 ∓ 0.0096

 

Specificity

0.8241 ∓ 0.0571

0.8574 ∓ 0.0233

 

MCC

0.7410 ∓ 0.0467

0.7680 ∓ 0.0315

Molecular-biochemical

Accuracy

0.8990 ∓ 0.0130

0.8990 ∓ 0.0130

 

Precision

0.9504 ∓ 0.0398

0.9333 ∓ 0.0119

 

Recall

0.9026 ∓ 0.0327

0.9182 ∓ 0.0096

 

Specificity

0.8963 ∓ 0.0818

0.8574 ∓ 0.0233

 

MCC

0.7794 ∓ 0.0371

0.7680 ∓ 0.0315

  1. Results are reported in the (average accuracy ∓ variance) format for all the 3-folds used to cross-validate the models. Note that recall is also the sensitivity metric. MCC, Matthews correlation coefficient.
Back to article page