LLMCARE: early detection of cognitive impairment via transformer models enhanced by LLM-generated synthetic data
Integrating transformer embeddings with handcrafted linguistic features enhances ADRD detection from speech. Distributionally aligned LLM-generated narratives provide effective but bounded augmentation, while current multimodal models remain limited. Crucially, validation on the Delaware corpus demonstrates that the proposed pipeline generalizes to early-stage impairment, supporting its potential as a scalable approach for clinically relevant early screening. All codes for LLMCARE are publicly available at: GitHub.
Authors
- James M. Noble (ORCID: https://orcid.org/0000-0003-0648-6702)
- M. Nezhad (ORCID: https://orcid.org/0000-0002-9242-8543)
- Masoud Rouhizadeh (ORCID: https://orcid.org/0000-0002-9006-6112)
- Yadollah Yaghoobzadeh (ORCID: https://orcid.org/0000-0003-0646-0852)
- Suzanne Bakken (ORCID: https://orcid.org/0000-0001-6202-6001)
- Masoud Khani (ORCID: https://orcid.org/0000-0003-3131-0661)
- Maryam Zolnoori (ORCID: https://orcid.org/0000-0003-4484-2990)
- Abdol‐Hossein Vahabie (ORCID: https://orcid.org/0000-0003-1603-8866)
- Ali Zolnour (ORCID: https://orcid.org/0009-0007-0668-5169)
- Samin Mahdizadeh Sani
- Amirsajjad Taleban
- Hossein Azadmaleki (ORCID: https://orcid.org/0009-0008-4284-9869)
- Yasaman Haghbin
- Sina Rashidi
- Maryam Dadkhah
- Fatemeh Taherinezhad
Institutions
- University of Tehran (IR)
- Columbia University Irving Medical Center (US)
- University of Florida (US)
- Alzheimer’s Disease Neuroimaging Initiative (US)
- University of Wisconsin–Milwaukee (US)
- Columbia University (US)
Publication Details
- Journal
- Frontiers in Artificial Intelligence
- Published
- 2025-11-06
- DOI
- https://doi.org/10.3389/frai.2025.1669896
- Primary Topic
- Neurobiology of Language and Bilingualism
- Type
- article
- Field-Weighted Citation Impact
- 0.00
Funders
- Centre de Coopération Internationale en Recherche Agronomique pour le Développement