Skip to main content

Table 1 Comparison of alternative machine learning methods. (LGB: Light Gradient Boosting, XGB: Extreme Gradient Boosting, LR: Logistic Regression, SVM: Support Vector Machines, RF: Random Forest, KNN: K-nearest Neighbor, LASSO: Least Absolute Shrinkage and Selection Operator)

From: Gradient boosting for Parkinson’s disease diagnosis from voice recordings

Metrics

 

Accuracy Metrics with 95% CI

 

LGB

XGB

LR

SVM

RF

KNN

LASSO

F1

0.839 [0.831–0.847]

0.810 [0.802–0.819]

0.771 [0.762–0.780]

0.730 [0.721–0.739]

0.810 [0.800–0.819]

0.744 [0.735–0.753]

0.763 [0.755–0.7723]

AUC

0.898 [0.892–0.905]

0.891 [0.885–0.898]

0.839 [0.830–0.847

0.838 [0.830–0.846]

0.884 [0.876–0.892]

0.841 [0.834–0.848]

0.870 [0.863–0.877]

Accuracy

0.841 [0.833–0.849]

0.816 [0.809–0.823]

0.771 [0.762–0.780]

0.744 [0.735–0.752]

0.818 [0.810–0.826]

0.760 [0.752–0.768]

0.761 [0.753–0.769]

Sensitivity

0.839 [0.827–0.850]

0.801 [0.789–0.813]

0.777 [0.765–0.790]

0.704 [0.691–0.716]

0.795 [0.782–0.808]

0.712 [0.699–0.725]

0.782 [0.769–0.794]

Specificity

0.844 [0.832–0.855]

0.830 [0.819–0.841]

0.764 [0.750–0.778]

0.784 [0.771–0.798]

0.841 [0.831–0.852]

0.807 [0.796–0.818]

0.741 [0.729–0.754]

PPV

0.853 [0.843–0.863]

0.835 [0.825–0.845]

0.780 [0.769–0.791]

0.780 0.769–0.791]

0.844 [0.834–0.854]

0.796 [0.786–0.806]

0.762 [0.753–0.772]