Relative performance of gLASSO, sgLASSO and SVM, depended on the dataset, while sLASSO and Random Forest were generally outperformed by the other algorithms.

<p>(a) semantic verbal fluency, (b) phonological verbal fluency and (c) combined datasets. Classification performance was significantly different between all algorithms and significantly higher for each algorithm with the combined dataset (<i>p</i> < 0.001, u-test).