Automated Lexical Dysfluency Analysis to Differentiate Primary Progressive Aphasia Variants

Differentiating nfvPPA and lvPPA relies on detailed phoneme-level error characterization, traditionally requiring extensive expertise and manual effort. Our novel approach automates this process using a quick and easy-to-administer reading task, providing an objective, scalable, and clinically viable solution. By capturing dysfluencies missed by ASR, this method enhances diagnostic precision and reduces reliance on highly specialized clinicians, facilitating broader clinical adoption for differential diagnosis.