Speech pause distribution as an early marker for Alzheimer’s disease

Abstract

BackgroundPause duration analysis is a common feature in the study of discourse in Alzheimer’s disease (AD) and may also be helpful for its early detection. However, studies involving patients with amnestic mild cognitive impairment (aMCI) have yielded varying results.ObjectivesTo characterize the probability density distribution of speech pause durations in AD, two multi-domain amnestic MCI patients (with memory encoding deficits, a-mdMCI-E, and with retrieval impairment only, a-mdMCI-R) and healthy controls (HC) in order check whether there are significant differences between them.Method112 picture-based oral narratives were manually transcribed and annotated for the automatic extraction and analysis of pause durations. Different probability distributions were tested for the fitting of pause durations while truncating shorter ranges. Recent findings in the field of Statistics were considered in order to avoid the inherent methodological uncertainty that this type of analysis entails.ResultsA lognormal distribution (LND) explained the distribution of pause duration for all groups. Its fitted parameters (µ,σ) followed a gradation from the group with shorter durations and a higher tendency to produce short pauses (HC) to the group with longer pause durations and a considerably higher tendency to produce long pauses with greater variance (AD). Importantly, a-mdMCI-E produced significantly longer pauses and with greater variability than their a-mdMCI-R counterparts (α = 0.05).ConclusionWe report significant differences at the group level in pause distribution across all groups of study that could be used in future diagnostic tools and discuss the clinical implications of these findings, particularly regarding the characterization of aMCI.

Symptomatic, Genetic, and Mechanistic Overlaps between Autism and Alzheimer’s Disease

“…However, the severity of this symptom depends on the stage of disease and is positively correlated with AD progression [ 45 , 46 , 47 ]. The distribution of speech pauses has been described as a marker for AD diagnosis [ 48 ]. Studies have shown that identification of language and diarization of speakers provide promising results for the diagnosis of speech loss in the case of AD [ 49 ].…”

Alzheimer’s Disease Detection from Spontaneous Speech Through Combining Linguistic Complexity and (Dis)Fluency Features with Pretrained Language Models

“…These indicators encompass acoustic features, such as vocalisation features (i.e. speech-silence patterns) [4], paralinguistic features, such as fluency features [5] and speech pause distributions [6], as well as syntactic and lexical features extracted from speech transcripts [7].…”