LoRA-Tuned Large Language Models for Dementia Detection via Multi-View Speech-Derived Features
Abstract
Early detection of dementia enables timely intervention, and reflecting cognitive impairment, spontaneous speech offers a non-invasive screening modality. Conventional approaches often focus on a single representational dimension -- such as acoustic descriptors, pause modeling, automatic speech recognition (ASR) transcripts, or multimodal fusion -- limiting integrative reasoning across heterogeneous cognitive symptoms. We propose a low-rank adaptation (LoRA)-tuned large language model (LLM) that performs structured multi-view reasoning over four complementary speech-derived signals: ASR transcripts with pause markers, discourse-level topic cues, temporal fluency statistics, and phonological sequences. These cues are encoded within a unified prompt, enabling a single LLM to learn a coherent decision function without modality-specific encoders or late-stage fusion. On ADReSSo, our best model achieves an F1-score of 90.14%, and ablation confirms the complementary contribution of each view.
Cite
@article{arxiv.2606.28445,
title = {LoRA-Tuned Large Language Models for Dementia Detection via Multi-View Speech-Derived Features},
author = {Jonghyeon Park and Olivier Jiyoun Jung and Myungwoo Oh},
journal= {arXiv preprint arXiv:2606.28445},
year = {2026}
}
Comments
Accepted at INTERSPEECH 2026