Evaluating the Clinical Utility of a Brief Remote Speech Assessment for Screening and Triage of Speech Disorders

Purpose: Speech abnormalities can often be early or clinically important manifestations of neurologic disease, yet access to clinicians with specialized expertise required for comprehensive motor speech evaluation may limit timely assessment. Brief remote speech assessment could provide a scalable approach for identifying patients who warrant more comprehensive evaluation, but its utility across heterogeneous speech presentations remains unclear. This study evaluated the validity and clinical utility of the Mayo Speech Test (MST), a brief (~5-minute) audio-recorded remote speech assessment, relative to a comprehensive in-person motor speech examination (~1-hour). Specifically, we examined whether MST could support clinically meaningful screening and triage decisions by distinguishing normal from abnormal speech and identifying the presence of a speech disorder. Method: One hundred participants completed a comprehensive in-person speech evaluation and a remote MST. Participants were excluded when the interval between assessments exceeded 90 days (n = 13) or either rater judged audio quality insufficient for diagnostic interpretation (n = 6), with overlapping exclusions resulting in a final analytic sample of 82 participants. Remote recordings were independently evaluated by two speech-language pathologists (SLPs) who were blinded to participant identity. In-person evaluation classifications served as the clinical reference standard and were grouped as normal (n = 33), dysphonia only (n = 5), or speech disorder (n = 43). One participant could not be classified. Performance was evaluated for two clinically motivated binary decisions: distinguishing normal from abnormal (such as dysphonia or speech disorder; Tier 0) and identifying normal speech or dysphonia only versus speech disorder (Tier 1). Sensitivity, specificity, predictive values, agreement, and error patterns were examined. Results: Both remote raters demonstrated similar overall performance despite differing error profiles. For Tier 0 screening, Rater 1 demonstrated 93.6% sensitivity and 71.9% specificity for identifying abnormal voice/speech, whereas Rater 2 demonstrated 78.7% sensitivity and 93.9% specificity. For Tier 1 identification of speech disorder, sensitivity and specificity were 83.3% and 86.5%, respectively, for Rater 1 and 71.4% and 97.4% for Rater 2. Interrater agreement was 79.7% for Tier 0 (κ = 0.60) and 86.1% for Tier 1 (κ = 0.72). Missed speech disorders represented an important source of error, with differences between raters indicating distinct thresholds for assigning abnormality and speech-disorder classifications. Conclusions: A brief remote speech assessment showed promising performance for broad clinical classification, particularly when used to address screening and triage questions rather than reproduce the diagnostic granularity of a comprehensive examination. Similar overall performance across raters masked meaningful sensitivity-specificity tradeoffs, highlighting the importance of examining error profiles in addition to aggregate accuracy. MST may therefore have clinical utility for identifying patients who warrant comprehensive motor speech evaluation, but should not be considered a substitute for a full diagnostic assessment.

Authors

Institutions

Publication Details

Journal
medRxiv
Published
2026-11-18
DOI
https://doi.org/10.64898/2026.11.16.26368191
Primary Topic
Voice and Speech Disorders
Type
preprint

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Evaluating the Clinical Utility of a Brief Remote Speech Assessment for Screening and Triage of Speech Disorders

Heather M. Clark, Joseph R. Duffy, Rene L. Utianski, Hugo Botha et al.
medRxiv
Voice and Speech Disorders
preprint

Evaluating the Clinical Utility of a Brief Remote Speech Assessment for Screening and Triage of Speech Disorders

Heather M. Clark, Joseph R. Duffy, Rene L. Utianski, Hugo Botha, Leland R Barnard, Nathan P. Staff, Julie A. G. Stierwalt, Nicole V. Dickerson
preprint en

Abstract

Purpose: Speech abnormalities can often be early or clinically important manifestations of neurologic disease, yet access to clinicians with specialized expertise required for comprehensive motor speech evaluation may limit timely assessment. Brief remote speech assessment could provide a scalable approach for identifying patients who warrant more comprehensive evaluation, but its utility across heterogeneous speech presentations remains unclear. This study evaluated the validity and clinical utility of the Mayo Speech Test (MST), a brief (~5-minute) audio-recorded remote speech assessment, relative to a comprehensive in-person motor speech examination (~1-hour). Specifically, we examined whether MST could support clinically meaningful screening and triage decisions by distinguishing normal from abnormal speech and identifying the presence of a speech disorder. Method: One hundred participants completed a comprehensive in-person speech evaluation and a remote MST. Participants were excluded when the interval between assessments exceeded 90 days (n = 13) or either rater judged audio quality insufficient for diagnostic interpretation (n = 6), with overlapping exclusions resulting in a final analytic sample of 82 participants. Remote recordings were independently evaluated by two speech-language pathologists (SLPs) who were blinded to participant identity. In-person evaluation classifications served as the clinical reference standard and were grouped as normal (n = 33), dysphonia only (n = 5), or speech disorder (n = 43). One participant could not be classified. Performance was evaluated for two clinically motivated binary decisions: distinguishing normal from abnormal (such as dysphonia or speech disorder; Tier 0) and identifying normal speech or dysphonia only versus speech disorder (Tier 1). Sensitivity, specificity, predictive values, agreement, and error patterns were examined. Results: Both remote raters demonstrated similar overall performance despite differing error profiles. For Tier 0 screening, Rater 1 demonstrated 93.6% sensitivity and 71.9% specificity for identifying abnormal voice/speech, whereas Rater 2 demonstrated 78.7% sensitivity and 93.9% specificity. For Tier 1 identification of speech disorder, sensitivity and specificity were 83.3% and 86.5%, respectively, for Rater 1 and 71.4% and 97.4% for Rater 2. Interrater agreement was 79.7% for Tier 0 (κ = 0.60) and 86.1% for Tier 1 (κ = 0.72). Missed speech disorders represented an important source of error, with differences between raters indicating distinct thresholds for assigning abnormality and speech-disorder classifications. Conclusions: A brief remote speech assessment showed promising performance for broad clinical classification, particularly when used to address screening and triage questions rather than reproduce the diagnostic granularity of a comprehensive examination. Similar overall performance across raters masked meaningful sensitivity-specificity tradeoffs, highlighting the importance of examining error profiles in addition to aggregate accuracy. MST may therefore have clinical utility for identifying patients who warrant comprehensive motor speech evaluation, but should not be considered a substitute for a full diagnostic assessment.

medRxiv
Mayo Clinic (US)
National Institute on Aging, National Center for Advancing Translational Sciences
Good health and well-being
Voice and Speech Disorders
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon