Dementia Language Models: a generalizable and controllable representation of cognitive impairment
- 1Technion - Israel Institute of Technology
Abstract
We introduce Dementia Language Models (DeLMs)--generalizable and controllable representations of cognitive impairment through language--alongside an evaluation framework for establishing their validity and clinical grounding. DeLMs created by fine-tuning large language models on a small clinical corpus successfully generated patient-like narratives across unseen tasks, received predicted Mini-Mental State Examination (MMSE) scores in the impaired range, and produced narratives that neurologists identified with accuracy comparable to real transcripts. The models' internal representations, as well as their non-linguistic decision-making, supported mild cognitive impairment detection in unseen cohorts. The effect was controllable: moving from Healthy toward Dementia in weight space progressively worsened language and predicted MMSE scores while increasing dementia probability. DeLMs could support clinician training, hypothesis generation, and scalable experimentation, reserving patient involvement for where it is truly needed.
Related
- Distance-dependent Functional Connectivity Strength Analysis Using Resting-State fMRI: Application to Vascular Cognitive Impairment, No Dementia
- A protocol for a scoping review of language and cultural considerations when screening, assessing, and caring for people living with dementia or mild cognitive impairment from minoritised ethnic groups
- Semaglutide ameliorates neuroinflammation and cognitive impairment in APP/PS1 mice
- Atypical Chemokine Receptor DARC/ACKR1 Is Expressed Selectively by Neurons with Regional Heterogeneity and Protects against Neuroinflammation and Cognitive Impairment
- Marital status and incident screen-positive cognitive impairment in a Taiwanese longitudinal cohort (2005–2016): Associations vary by cognitive screening instrument
BibTeX
@misc{peledcohen2026dementia,
title = {{Dementia Language Models: a generalizable and controllable representation of cognitive impairment}},
author = {L. Peled-Cohen and A. Shmidov and N. Rein and E. Shapira and N. Calderon and R. Tikochinski and E. Zeltzer and T. Nathan and B. Uliel and K. D. Mueller and I. Ganmore and B. Reis and R. Reichart},
howpublished = {bioRxiv},
year = {2026},
month = sep,
doi = {10.64898/2026.09.16.752129},
url = {https://www.biorxiv.org/content/10.64898/2026.09.16.752129v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - Dementia Language Models: a generalizable and controllable representation of cognitive impairment
AU - L. Peled-Cohen
AU - A. Shmidov
AU - N. Rein
AU - E. Shapira
AU - N. Calderon
AU - R. Tikochinski
AU - E. Zeltzer
AU - T. Nathan
AU - B. Uliel
AU - K. D. Mueller
AU - I. Ganmore
AU - B. Reis
AU - R. Reichart
PY - 2026
DA - 2026/09/16
PB - bioRxiv
DO - 10.64898/2026.09.16.752129
UR - https://www.biorxiv.org/content/10.64898/2026.09.16.752129v1
AB - We introduce Dementia Language Models (DeLMs)--generalizable and controllable representations of cognitive impairment through language--alongside an evaluation framework for establishing their validity and clinical grounding. DeLMs created by fine-tuning large language models on a small clinical corpus successfully generated patient-like narratives across unseen tasks, received predicted Mini-Mental State Examination (MMSE) scores in the impaired range, and produced narratives that neurologists identified with accuracy comparable to real transcripts. The models' internal representations, as well as their non-linguistic decision-making, supported mild cognitive impairment detection in unseen cohorts. The effect was controllable: moving from Healthy toward Dementia in weight space progressively worsened language and predicted MMSE scores while increasing dementia probability. DeLMs could support clinician training, hypothesis generation, and scalable experimentation, reserving patient involvement for where it is truly needed.
ER -