New methods improve Alzheimer's detection via speech analysis, but data cleaning may hurt generalization

Researchers have developed a new method called LLM-Anchored Paralinguistic Enrichment (LAPE) to improve the detection of Alzheimer's disease using speech analysis. LAPE integrates linguistic content with paralinguistic cues like pauses and word elongations, which are affected by Alzheimer's. The method was evaluated on the ADReSS and ADReSSo datasets, achieving state-of-the-art performance. Separately, a study revisited existing benchmarks for speech-based Alzheimer's detection, finding that while speech enhancement can improve in-domain performance, it may reduce the generalization capabilities of deep learning models and large audio-language models, suggesting that "cleaner" speech data isn't always more reliable for real-world detection. AI

IMPACT New speech analysis techniques show promise for Alzheimer's detection, but researchers caution that data preprocessing choices can impact model generalization.

RANK_REASON The cluster contains two research papers discussing methods and benchmarks for Alzheimer's disease detection using speech analysis.

Read on Hugging Face Daily Papers →

• Alzheimer's disease
• Hugging Face
• Large Audio-Language Models
• Pitt Corpus
• Pitt-Derived Benchmarks
• ADReSSo
• arXiv
• LAPE
• LLM-Anchored Paralinguistic Enrichment

• paper
• model release
• other

AI-generated summary · Google Gemini · from 2 sources. How we write summaries →