Table 2.

Results on the decision tree analyses, including the importance per feature in the model of the test data (ranking variable order from highest relative importance to lowest) and the area under the curve (AUC) per feature in the models of both the train and test dataset

	Importance		AUC
	Relative importance	Rank of importance within the model	Train data	Test data
Cognitive profiles
Model 1			0.969	0.910
Attention	0.181	2
Inhibition	0.112	5
IPS	0.053	6
Verbal fluency	0.173	3
Verbal memory	0.168	4
Visuospatial memory	0.313	1
Model 2			0.754	0.629
Sex	0.035	5
Age	0.382	1
Education	0.050	4
MS type	0.026	6
Disease duration	0.286	2
EDSS	0.221	3
Model 3			0.785	0.661
Sex	0.019	8
Age	0.250	2
Education	0.042	7
MS type	0.011	9
Disease duration	0.228	3
EDSS	0.097	4
HADS-A	0.049	6
HADS-D	0.048	5
CIS20-R	0.256	1
Cognitive status
Model 1			0.999	0.997
Attention	0.244	2
Inhibition	0.112	4
IPS	0.301	1
Verbal fluency	0.064	6
Verbal memory	0.173	3
Visuospatial memory	0.107	5
Model 2
Sex	0.055	5	0.741	0.633
Age	0.172	3
Education	0.056	4
MS type	0.027	6
Disease duration	0.373	1
EDSS	0.317	2
Model 3			0.789	0.646
Sex	0.031	8
Age	0.174	3
Education	0.048	7
MS type	0.008	9
Disease duration	0.301	1
EDSS	0.177	2
HADS-A	0.051	6
HADS-D	0.087	5
CIS20-R	0.122	4

Abbreviations: IPS information processing speed, EDSS Expanded Disability Status Scale, HADS-A Hospital Anxiety and Depression Scale (HADS)-Anxiety subscale, HADS-D HADS-Depression subscale, CIS20-R Checklist Individual Strength 20-Revised