eduzhai > Applied Sciences > Engineering >

Comparing Natural Language Processing Techniques for Alzheimers Dementia Prediction in Spontaneous Speech

  • Save

... pages left unread,continue reading

Document pages: 5 pages

Abstract: Alzheimer s Dementia (AD) is an incurable, debilitating, and progressiveneurodegenerative condition that affects cognitive function. Early diagnosis isimportant as therapeutics can delay progression and give those diagnosed vitaltime. Developing models that analyse spontaneous speech could eventuallyprovide an efficient diagnostic modality for earlier diagnosis of AD. TheAlzheimer s Dementia Recognition through Spontaneous Speech task offersacoustically pre-processed and balanced datasets for the classification andprediction of AD and associated phenotypes through the modelling of spontaneousspeech. We exclusively analyse the supplied textual transcripts of thespontaneous speech dataset, building and comparing performance across numerousmodels for the classification of AD vs controls and the prediction of MentalMini State Exam scores. We rigorously train and evaluate Support VectorMachines (SVMs), Gradient Boosting Decision Trees (GBDT), and ConditionalRandom Fields (CRFs) alongside deep learning Transformer based models. We findour top performing models to be a simple Term Frequency-Inverse DocumentFrequency (TF-IDF) vectoriser as input into a SVM model and a pre-trainedTransformer based model `DistilBERT when used as an embedding layer intosimple linear models. We demonstrate test set scores of 0.81-0.82 acrossclassification metrics and a RMSE of 4.58.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×