Suboptimal Classifier for Dysarthria Assessment

This work is focused on the design and evaluation of a suboptimal classifier for dysarthria assessment. The classification relied on self organizing maps to discriminate 8 types of dysarthria and a normal group. The classification technique provided an excellent accuracy for assessment and enabled clinicians with a powerful relevance analysis of the input features. This technique also allows a bi-dimensional map that shows the spatial distribution of the data revealing important information about the different dysarthric groups.