Detection of Cognitive Impairment Using Bidirectional Long Short‐Term Memory and Speech‐Derived Digital Biomarkers in Native English Speakers
Our results show that our proposed BiLSTM architecture outperforms machine learning and deep learning models, and that acoustic speech features are better predictors of cognitive impairment than linguistic features. The proposed model has the potential to improve the clinical diagnosis of cognitive impairment in an easy and scalable fashion.
Authors
- Pooyan Mobtahej (ORCID: https://orcid.org/0000-0003-1973-8761)
- Maria G. Corona
- David L. Sultzer
- Dennis Le
- Tianchen Qian (ORCID: https://orcid.org/0000-0003-4282-7826)
- Michelle McDonnell (ORCID: https://orcid.org/0000-0001-6811-8708)
- Seyed ahmad Sajjadi (ORCID: https://orcid.org/0000-0001-5003-7889)
- Sam Gouron (ORCID: https://orcid.org/0009-0004-4938-2107)
- Melina vom Saal
Institutions
- University of California, Irvine (US)
Publication Details
- Journal
- Alzheimer s & Dementia
- Published
- 2025-12-01
- DOI
- https://doi.org/10.1002/alz70856_104560
- Primary Topic
- Voice and Speech Disorders
- Type
- article
- Field-Weighted Citation Impact
- 0.00