AUTHOR=Chlasta Karol , Wołk Krzysztof 

TITLE=Towards Computer-Based Automated Screening of Dementia Through Spontaneous Speech

JOURNAL=Frontiers in Psychology

VOLUME=Volume 11 - 2020

YEAR=2021

URL=https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2020.623237

DOI=10.3389/fpsyg.2020.623237

ISSN=1664-1078

ABSTRACT=Dementia, a prevalent disorder of the brain, has negative effects on individuals and society. This paper concerns using Spontaneous Speech (ADReSS) Challenge of Interspeech 2020 to classify Alzheimer’s Dementia. We used (1) VGGish, a deep, pretrained, Tensorflow model as an audio feature extractor, and Scikit-learn classifiers to detect signs of dementia in spoken language. Three classifiers (LinearSVM, Perceptron, 1NN) were 59% accurate, which was up to 6% above baseline models using MRCG or ComParE features. We also proposed (2) a custom PyTorch audio CNN model that was more accurate than (1) and was 62.5% accurate, similar to the best performing baseline LDA classifier on ComParE features. We discovered that audio transfer learning with a pretrained VGGish feature extractor performs better than the baseline approach using automatically extracted acoustic features.  Both methods presented in this paper offer progress towards new, innovative, and more effective computer-based screening of dementia through spontaneous speech.