Automated Speech Analysis in FTD Spectrum Disorders
Abstract
Frontotemporal degeneration (FTD) is an understudied form of focal dementia. Its public health impact is immense because clinical FTD is the most common neurodegenerative disease in individuals <65 years old. FTD presents with a specific language deficit (Primary Progressive Aphasia, PPA). A careful analysis of everyday speech can help identify variants of PPA. This proposal fills a major gap by providing an objective, replicable, fully automated approach to discerning speech characteristics of PPA. FTD may co-occur with a motor disorder, including Amyotrophic Lateral Sclerosis (ALS) and Chronic Traumatic Encephalopathy (CTE) which are directly relevant to the military. Detailed analyses of speech in FTD spectrum disorders with associated motor impairments are rare, and we propose to extend our analyses to FTD patients with motor disorders. Finally, longitudinal analyses of speech can play an important role in prognosis and in treatment trials, but longitudinal studies are rare. This study pursues these issues with three Specific Aims: 1. Develop an automated algorithm to analyze lexical semantic word-level content and grammatical category in FTD; 2. Develop automated algorithms to align lexical content with acoustic signal in connected speech samples of FTD speakers; and 3. Develop algorithms to automatically characterize the properties of the complex (acoustic and lexical) signals that are associated with sentence boundaries and syntactic units in FTD speech.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 2022
- Accession Number
- AD1200144
Entities
People
- Murray Grossman
- Naomi Nevler
Organizations
- University of Pennsylvania