The Joint Effects of Acoustic and Linguistic Markers for Early Identification of Mild Cognitive Impairment.

Fengyi Tang,Jiayu Zhou,Hiroko H Dodge,Jun Chen

doi:10.3389/fdgth.2021.702772

Abstract

In recent years, behavioral markers such as spoken language and lexical preferences have been studied in the early detection of mild cognitive impairment (MCI) using conversations. While the combination of linguistic and acoustic signals have been shown to be effective in detecting MCI, they have generally been restricted to structured conversations in which the interviewee responds to fixed prompts. In this study, we show that linguistic and acoustic features can be combined synergistically to identify MCI in semi-structured conversations. Using conversational data from an on-going clinical trial (Clinicaltrials.gov: NCT02871921), we find that the combination of linguistic and acoustic features on semi-structured conversations achieves a mean AUC of 82.7, significantly (p < 0.01) out-performing linguistic-only (74.9 mean AUC) or acoustic-only (65.0 mean AUC) detections on hold-out data. Additionally, features (linguistic, acoustic and combination) obtained from semi-structured conversations outperform their counterparts obtained from structured weekly conversations in identifying MCI. Some linguistic categories are significantly better at predicting MCI status (e.g., death, home) than others.

Highlights

Detection of dementia at an early mild cognitive impairment (MCI) stage has been of great interest in recent years for effective prevention of dementia as well as clinical trials enrichment
We examine the use of linguistic and acoustic features for MCI classification in semi-structured conversations
As a result, semistructured conversations introduce challenges in the stability of behavioral markers, especially for ones dependent on acoustic signals. By combining both acoustic and linguistic markers, we show that the composite behavioral markers can significantly out-perform any single modality alone

Summary

Introduction

Detection of dementia at an early MCI stage has been of great interest in recent years for effective prevention of dementia as well as clinical trials enrichment. There has been a growing interest in the use of linguistic-based language and acoustic-based behavioral markers—characteristics related to language use [1], speech [2,3,4], cognitive capacity [5], and lexical preferences [6] ( referred as behavioral markers in this paper)—in early detection. This is because behavioral markers can provide easy accessibility and are generally more cost-effective [7] to obtain than biological markers such as PET/CT scans [8, 9]. In semi-structured conversations, there are no set interview questions; instead, conversations are led by the participant themselves

Objectives

Methods

Results

Conclusion