[TD‐P‐023]: FEASIBILITY OF AUTOMATED VOICE‐BASED COGNITIVE ASSESSMENT ON A CONSUMER VOICE PLATFORM

• DOI: 10.1016/j.jalz.2017.06.2619
• Publication Date: 2017-10-20T08:32:30Z
• Authors: Nick Taptiklis; Francesca K. Cormack; Pasquale Dente; Rosa Backx; Jennifer H. Barnett
• Keywords: Memory span; Cognitive test

Speech is a complex and uniquely human behaviour including both motor cognitive components. Neurological neurodegenerative diseases impact on different aspects of speech language. response modality in large number established tests tapping non-language cognition episodic memory, working memory executive function. In clinical studies, accuracy consistency scoring administration are still problematic; computerised batteries such as CANTAB can eliminate this source error by automating test scoring. Computerising verbal has not been feasible to date due errors introduced the relatively poor performance recognition technology. However, state-of-the-art incorporating deep learning other recent advances have prompted us revisit prior feasibility assumptions. We developed short voice-based battery with tasks assessing function using off-the-shelf The was composed automated versions Digit Span, Auditory-Verbal Learning Verbal Paired Associates tests. assessed twenty participants simultaneous independent audio. Participants were also standard measuring similar domains. Analysis focused 1) reliability 2) construct validity compared Expected patterns covariance found between voice battery. software varied tasks, digit during span task superior unconstrained word paired associates task. This corresponds intrinsic difficulty Automated assessment may now be practical accurate means administering remotely, for example population screening or home based monitoring. Further effectiveness required.