Pauses for Detection of Alzheimer’s Disease

Pauses for Detection of Alzheimer’s Disease

Blog Article

Pauses, disfluencies and language problems in Alzheimer’s disease can be naturally modeled by fine-tuning Transformer-based pre-trained language models such as BERT and ERNIE.Using this method with pause-encoded transcripts, we achieved 89.6% accuracy on the test set of the ADReSS (Alzheimer’s Dementia Recognition through Spontaneous Speech) Challenge.The best accuracy was obtained Camera Bags with ERNIE, plus an encoding of pauses.Robustness is a challenge for large models and small K Boots training sets.

Ensemble over many runs of BERT/ERNIE fine-tuning reduced variance and improved accuracy.We found that um was used much less frequently in Alzheimer’s speech, compared to uh.We discussed this interesting finding from linguistic and cognitive perspectives.