Alzheimer's Disease Detection from Spontaneous Speech through Combining Linguistic Complexity and (Dis)Fluency Features with Pretrained Language Models
This work addresses Alzheimer's detection for medical diagnosis, but it is incremental as it builds on existing methods and datasets.
The paper tackled Alzheimer's disease detection from spontaneous speech by combining linguistic complexity and disfluency features with pretrained language models, achieving an accuracy of 83.1% on a test set, which is a 4.23% improvement over the baseline.
In this paper, we combined linguistic complexity and (dis)fluency features with pretrained language models for the task of Alzheimer's disease detection of the 2021 ADReSSo (Alzheimer's Dementia Recognition through Spontaneous Speech) challenge. An accuracy of 83.1% was achieved on the test set, which amounts to an improvement of 4.23% over the baseline model. Our best-performing model that integrated component models using a stacking ensemble technique performed equally well on cross-validation and test data, indicating that it is robust against overfitting.