What Do LLMs Know About Alzheimer's Disease? Multi-loss Fine-Tuning and Probing for AD Detection
Researchers have explored the potential of large language models (LLMs) for detecting Alzheimer's disease (AD) from text, particularly in scenarios with limited labeled data. They fine-tuned models like BERT, T5, and Llama-1B on various transcript corpora, achieving new state-of-the-art results on some datasets. The study also analyzed how AD-related information is encoded within the models' internal representations using linear probing, showing that fine-tuning shifts token representations to reflect AD signals. AI
IMPACT Demonstrates LLMs' potential for medical diagnosis from text, potentially improving early detection rates.