Cognitive Digital Biomarkers from Automated Transcription of Spoken Language
The Journal of Prevention of Alzheimer's Disease · 2022
Abstract
Voice-based digital biomarkers derived from automated processing methods can effectively predict cognitive status. This study compared acoustic and linguistic features from 200 automated transcriptions and 127 manual transcriptions of neuropsychological assessments from 146 Framingham Heart Study participants. Models incorporating transcription-related features achieved an area under the receiver-operator characteristic curve of 0.90, substantially outperforming demographic-only models (0.60). Combining voice-based digital biomarkers with standard patient screening might constitute a scalable way to enable early detection of dementia.
BibTeX
@article{cognitive-digital-biomarkers-automated-transcription},
title = {Cognitive Digital Biomarkers from Automated Transcription of Spoken Language},
author = {Nazli Tavabi and David Stuck and Alessio Signorini and Cody Karjadi and Tuka Al Hanai and Maggie Sandoval and Christine Lemke and James Glass and Scott Hardy and Michelle Lavallee and Bryan Wasserman and Ting Fang Alvin Ang and Christiane M. Nowak and Raghu Kainkaryam and Luca Foschini and Rhoda Au},
journal = {The Journal of Prevention of Alzheimer's Disease},
month = {10},
year = {2022},
doi = {10.14283/jpad.2022.66},
url = {https://alessiosignorini.com/publications/cognitive-digital-biomarkers-automated-transcription/},
}