Skip to main content
. 2016 Sep 7;3(11):866–875. doi: 10.1002/acn3.348

Table 2.

Random forest variable importance of preliminary model

Variable Importance
Time from Baseline 26.57%
Baseline ALSFRS‐R 23.19%
Baseline ALSFRS slope 13.74%
Baseline Trunk Sub‐score 6.17%
Time Since Symptom Onset 3.66%
Baseline FVC 2.89%
Baseline Fine Motor Sub‐score 2.80%
Time Since Diagnosis 2.64%
Age 2.40%
Baseline Bulbar Sub‐score 2.33%
Systolic BP 2.10%
Baseline Gross Motor Sub‐score 2.07%
Weight 2.05%
Pulse 1.90%
Height 1.85%
Diastolic BP 1.56%
Baseline Respiratory Sub‐score 0.98%
Limb Onset 0.29%
Bulbar Onset 0.29%
Gender 0.27%
Riluzole Use 0.26%

FVC, forced vital capacity.

This table shows the variable importance list for a preliminary 21 variable model. To reduce the possibility of overfitting, variables whose importance was less than 2.00% were deleted from the final 13 variable model that was used in all the analyses.