Recognition of Alzheimer’s Dementia From the Transcriptions of Spontaneous Speech Using fastText and CNN Models

Abstract

Alzheimer’s dementia (AD) is a type of neurodegenerative disease that is associated with a decline in memory. However, speech and language impairments are also common in Alzheimer’s dementia patients. This work is an extension of our previous work, where we had used spontaneous speech for Alzheimer’s dementia recognition employing log-Mel spectrogram and Mel-frequency cepstral coefficients (MFCC) as inputs to deep neural networks (DNN). In this work, we explore the transcriptions of spontaneous speech for dementia recognition and compare the results with several baseline results. We explore two models for dementia recognition: 1) fastText and 2) convolutional neural network (CNN) with a single convolutional layer, to capture the n-gram-based linguistic information from the input sentence. The fastText model uses a bag of bigrams and trigrams along with the input text to capture the local word orderings. In the CNN-based model, we try to capture different n-grams (we use n = 2, 3, 4, 5) present in the text by adapting the kernel sizes to n. In both fastText and CNN architectures, the word embeddings are initialized using pretrained GloVe vectors. We use bagging of 21 models in each of these architectures to arrive at the final model using which the performance on the test data is assessed. The best accuracies achieved with CNN and fastText models on the text data are 79.16 and 83.33%, respectively. The best root mean square errors (RMSE) on the prediction of mini-mental state examination (MMSE) score are 4.38 and 4.28 for CNN and fastText, respectively. The results suggest that the n-gram-based features are worth pursuing, for the task of AD detection. fastText models have competitive results when compared to several baseline methods. Also, fastText models are shallow in nature and have the advantage of being faster in training and evaluation, by several orders of magnitude, compared to deep models.

Explainable Identification of Dementia From Transcripts Using Transformer Networks

“…More specifically, BERT outperforms all the research works, except [15], in terms of Accuracy by 2.08-8.33%, in F1-score by 1.33-8.68%, and in Recall by 2.66-14.99%. Moreover, BERT+Co-Attention surpasses [22], [27], [28] in Accuracy by 2.50%, 0.42%, and 4.58% respectively. Also, it surpasses [22], [27], [28] in Recall by 17.49%, 5.16%, and 9.16% respectively.…”

“…We extend this research work by employing more transformer-based networks with an efficient training strategy, proposing a new interpretable method to detect AD patients based on siamese networks, introducing two models in a multi-task learning framework by regarding the MMSE prediction task as a multiclass classification task and employing explainability techniques. On the other hand, research works [27] & [28] introduced deep learning models including CNNs and LSTM neural networks with feed-forward highway layers respectively. In [27] results suggested that the utterances of the interviewer boost the classification performance.…”

“…On the other hand, research works [27] & [28] introduced deep learning models including CNNs and LSTM neural networks with feed-forward highway layers respectively. In [27] results suggested that the utterances of the interviewer boost the classification performance. A similar methodology with [28] was proposed by [29], where the authors exploited both BERT and LSTMs with gating mechanism and showed that LSTM with gating mechanism outperforms BERT model with gating mechanism.…”

“…Comparison with state-of-the-art approaches: We compare our introduced models with the following research works, since these research works propose single-task learning models and test their proposed approaches on the ADReSS Challenge test set: (1) Text [15], (2) LSTM with Gating (Acoustic + Lexical + Dis) [28], (3) Fusion Maj. (3-best) [30], (4) Logistic Regression (NLP) [20], (5) fastText, bi + trigram [27], (6) Attempt 5 [21], and (7) Fusion of system [22].…”

Multimodal Deep Learning Models for Detecting Dementia From Speech and Transcripts

“…Meghanani et al ( 2021b ) introduced some approaches to detect AD patients and predict the MMSE scores using only text data. Specifically, the authors proposed a Convolutional Neural Network (CNN) and fastText-based classifiers.…”