Researchers have reported validation of an automated digital speech-based measure, which can be used to identify subjects with speech and language changes associated with impaired cognition.
The researchers from Arizona State University, in Tampa, and other centers in the USA, developed a semantic relevance (SemR) metric which is algorithmically extracted from speech. This measures the overlap between a picture’s content and the words that a subject uses to describe the picture.
The SemR algorithm was developed based on transcripts of the “Cookie Theft” picture in the Boston Diagnostic Aphasia Exam (BDAE), completed by 25 participants who provided weekly speech samples.