Physical activity and the outcome of cognitive trajectory: a machine learning approach

Table 4 Performance of the SVM models trained using grid search’s optimal set of hyperparameters on the held-out blind test-set for different cognitive outcomes

Cognitive Variables	Group	Accuracy	AUC
MMSE	CG	62.5	48.8
MMSE	IG	63.6	60.7
RWT_animals	CG	40.6	34.0
RWT_animals	IG	51.5	51.7
CDT	CG	59.4	50.4
CDT	IG	60.6	40.4
TMT	CG	50.0	39.8
TMT	IG	39.4	43.1
DS_for	CG	68.8	37.3
DS_for	IG	51.5	42.6
DS_back	CG	75.0	59.4
DS_back	IG	75.6	30.8
CVLT	CG	73.1	56.4
CVLT	IG	84.8	72.9

AUC Area under the ROC curve, CDT Clock Drawing Test, CVLT California Verbal Learning Test, DS_back Digit Span Test backward, DS_for Digit Span Test forward, MMSE Mini Mental State Examination, RWT_animals Regensburg Word Fluency Test subtest animals

ISSN: 1861-6909