Table 19 Performance of model-based and model-free methods (using selected features) for Tel-Aviv dataset to predict no fall or at least two falls, contrast to results in Table 11 (falls/no-fall).

From: Model-based and Model-free Machine Learning Techniques for Diagnostic Prediction and Classification of Clinical Outcomes in Parkinson’s Disease

Method

acc

sens

spec

ppv

npv

lor

auc

Random Forests

0.811

0.714

0.855

0.690

0.869

2.689

0.880

AdaBoost

0.822

0.750

0.855

0.700

0.883

2.872

0.886

XGBoost

0.811

0.643

0.887

0.720

0.846

2.649

0.885

SVM

0.833

0.714

0.887

0.741

0.873

2.978

0.881

Neural Network

0.722

0.607

0.774

0.548

0.814

1.667

 

Super Learner

0.800

0.643

0.871

0.692

0.844

2.497

 
Back to article page