Table 4 Sensitivity and specificity of each model at optimal cutoff point.

From: Using deep learning to predict temporomandibular joint disc perforation based on magnetic resonance imaging

Model

Sensitivity

Specificity

Cutoff point of probability

MLP

85.2%

84.8%

0.508

Random forest

96.3%

75.8%

0.196

Disc shape alone

80.8%

63.0%

0.307

  1. Optimal cutoff was considered the point maximizing the sum of sensitivity and specificity.
  2. MLP multilayer perceptron.
Back to article page