Evaluation of linguistic and prosodic features for detection of Alzheimer’s disease in Turkish conversational speech

Abstract

Automatic diagnosis and monitoring of Alzheimer's disease can have a significant impact on society as well as the well-being of patients. The part of the brain cortex that processes language abilities is one of the earliest parts to be affected by the disease. Therefore, detection of Alzheimer's disease using speech-based features is gaining increasing attention. Here, we investigated an extensive set of features based on speech prosody as well as linguistic features derived from transcriptions of Turkish conversations with subjects with and without Alzheimer's disease. Unlike most standardized tests that focus on memory recall or structured conversations, spontaneous unstructured conversations are conducted with the subjects in informal settings. Age-, education-, and gender-controlled experiments are performed to eliminate the effects of those three variables. Experimental results show that the proposed features extracted from the speech signal can be used to discriminate between the control group and the patients with Alzheimer's disease. Prosodic features performed significantly better than the linguistic features. Classification accuracy over 80% was obtained with three of the prosodic features, but experiments with feature fusion did not further improve the classification performance.

Computer‐based evaluation of Alzheimer's disease and mild cognitive impairment patients during a picture description task

“…There is extensive literature covering the analysis of the linguistic characteristics of AD patients [6,7,24,[28][29][30][31][32][33][34]. As part of our evaluation, we selected those that most authors have found to correlate significantly with the disease and that could be used in picture description tasks (Table 1).…”

Detecting Alzheimer’s Disease Using Gated Convolutional Neural Network from Audio Data

“…From the study, they found that patients often not only uttered incomplete phrases but also interrupted others. These degraded the intelligibility of their speech [15].…”

“…After the utterance-based classification is performed, we make the final verdict for each subject based on the proportion of each class; we classified a subject into AD if he/she has AD percentage above the half. The utterance-level subject classification is expected to give more detailed information about the symptoms while most of the previous studies conducted the subject-level classification [15,3,12,13,9,16].…”

Manual and Automatic Transcriptions in Dementia Detection from Speech

“…Automatic speech processing has been shown to be a promising way forward in the diagnosis of dementia. Approaches have used acoustic or prosodic features [3,4,5,10,11] and linguistic or text-based features [12,6,7,10,13,14,15,16,17] in a classification task that aims to distinguish speakers affected by dementia from cognitively health speakers using just their speech.…”

“…Nonetheless automatic transcripts of elderly speech have successfully been used to detect dementia in standardized texts [12,7,15] although the word error rates for individual subjects reached over 90% [15]. Still, many of the approaches that have used text based features relied on manual transcriptions [10,13]. If we find dementia detection using features extracted from automatic transcriptions to work as well as dementia detection using features extracted from manual transcripts, then we no longer need to rely on manual transcriptions.…”