Automatic Detection of Horner Syndrome by Using Facial Images

Table 3

The performances of machine learning classifiers.

Classifiers	Sen	Spe	PPV	NPV	Acc
Decision tree	0.432	0.970	0.879	0.773	0.790
K-neighbors	0.483	0.940	0.803	0.784	0.788
XgBoost	0.39	0.983	0.920	0.762	0.785
Gradient boosting	0.373	0.987	0.936	0.758	0.782
Logistic regression	0.364	0.987	0.935	0.756	0.779
Support vector classifier	0.356	0.987	0.933	0.753	0.776
Light GBM	0.322	0.979	0.884	0.742	0.759
Random forest	0.254	0.996	0.968	0.727	0.748
AdaBoost	0.237	0.996	0.966	0.722	0.742
Bernoulli naïve Bayes	0.331	0.902	0.629	0.729	0.711

Sen: sensitivity, Spe: specificity, PPV: positive predictive value, NPV: negative predictive value, and Acc: accuracy.