Table 8 Top seven features (*) are selected by both methods (RF and KO): gaitSpeed_Off, PIGD_score, partII_sum, BMI, X2.11, H_and_Y_OFF, X3.10gait_off.

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

Random Forests

Knockoff

Features

Frequency

Features

Frequency

gaitSpeed_Off*

0.992

X2.11*

0.822

PIGD_score*

0.992

PIGD_score*

0.784

partII_sum*

0.878

Gender

0.742

TUG_OFF

0.856

X3.10gait_off*

0.621

BMI*

0.806

H_and_Y_OFF*

0.579

X2.11*

0.788

partII_sum*

0.566

R_middle_temporal_gyrus

0.632

gaitSpeed_Off*

0.544

H_and_Y_OFF*

0.586

X2.12

0.394

R_inferior_temporal_gyrus

0.558

X1.8

0.355

R_middle_orbitofrontal_gyrus

0.406

BMI*

0.346

partI_sum

0.404

X2.8

0.333

L_middle_temporal_gyrus

0.392

MoCA

0.256

L_gyrus_rectus

0.384

X2.9

0.246

X3.10gait_off*

0.376

X3.17d

0.240

L_middle_occipital_gyrus

0.354

X1.9

0.211

R_fusiform_gyrus

0.354

X3.12pull_test_off

0.202

L_lateral_orbitofrontal_gyrus

0.352

X1.10

0.195

L_middle_orbitofrontal_gyrus

0.326

X2.13

0.192

R_angular_gyrus

0.290

L_middle_frontal_gyrus

0.157

L_superior_occipital_gyrus

0.282

X2.10

0.154

Back to article page