Automated classification of primary progressive aphasia subtypes from narrative speech transcripts.

摘要: In the early stages of neurodegenerative disorders, individuals may exhibit a decline in language abilities that is difficult to quantify with standardized tests. Careful analysis connected speech can provide valuable information about patient's capacities. To date, this type has been limited by its time-consuming nature. study, we present method for evaluating and classifying primary progressive aphasia using computational techniques. Syntactic semantic features were automatically extracted from transcriptions narrative three groups: dementia (SD), nonfluent (PNFA), healthy controls. Features varied significantly between groups used train machine learning classifiers, which then tested on held-out data. We achieved accuracies well above baseline binary classification tasks. An influential showed contrast controls, both patient tended use words higher frequency (especially nouns SD, verbs PNFA). The SD patients also nouns) familiarity, they produced fewer nouns, but more demonstratives adverbs, than PNFA group be slower incorporate shorter distinguished each other patients' relatively increased are high and/or familiarity.