Digital Speech Biomarkers Show Promise in Recording Speech Function, Dysarthria in ALS


Those who reported higher ALSFRS-R speech scores at baseline had significantly higher diadochokinetic and word rate than those who reported speech disturbances, or lower scores.

Andrew Geronimo, PhD, assistant professor, department of neurology, Penn State University

Andrew Geronimo, PhD

Through a tablet and microphone, investigators recently found that digital speech biomarkers may serve as a potentially beneficial objective measure of speech function for patients with amyotrophic lateral sclerosis (ALS). With data collection ongoing, these findings may contribute towards the goal of developing an objective measure of dysarthria for use in clinical practice and research.1

Presented at the 2022 Annual Northeast Amyotrophic Lateral Sclerosis (NEALS) Meeting, held November 1-3, in Clearwater Beach, Florida, the small-scale study included 8 individuals who performed twice-monthly home recordings of 3 speaking tasks. Using a tablet and microphone (BioSensics), patients were assessed on speech rate through the "rainbow” passage, diadochokinetic (DDK) rate using monosyllable repetition of “pa”, and fundamental frequency (f0) by the holding of a vowel.

Senior investigator Andrew Geronimo, PhD, assistant professor, department of neurology, Penn State University, and colleagues calculated speaking rate, syllable rate, and f0 during detected periods of speech and compared them with ALS Functional Rating Scale-Revised (ALSFRS-R) speech scores of 2, 3, or 4, using an ANOVA test. Mean f0 outcomes were plotted for each subject and trends were calculated using linear regression.

The cohort included 7 patients with ALS and 1 with primary lateral sclerosis (PLS). ALSFRS-R speech scores of 2, or intelligible with repeating, was documented in 3 individuals, while scores of 3—detectable speech disturbance—was found in 1 individual and normal scores of 4 were found in 4 individuals. For those with average scores, the average syllables per second were 3.8 and the average words per min were 225.8 whereas for those with ALSFRS-R speech scores of 2, the average syllables per second were 1.7 and average words per min were 129.1

On average, individuals completed 91.9% (range, 53.8% to 100.0%) of assigned rainbow passage recordings and 89.7% (range, 50% to 100%) of assigned monosyllable repetition recordings. All told, those at enrollment with normal speech processes, or ALSFRS-R scores of 4, trended towards higher DDK rate than those who reported speech disturbances (P = .051). Additionally, those with normal speech had a significantly higher speech rate than those who reported speech disturbances (P <.05).

The Significance of AMX0035 and a Fueled Hope for ALS

Once thought nearly impossible to treat, ALS has seen another step of progress with the approval of AMX0035 and a future of potential therapeutics on the way.

In the f0 analysis, one patient demonstrated a significant trend with time (r = 0.81; P <.001) while the remaining subjects showed no trends with time. Geronimo et al wrote that the trend may be due to "the progression of MND [motor neuron disease] or the repeated administration of botulinum toxin for neck dystonia during this period."

This was not the first time that speech biomarkers have shown promise in quantifying dysarthria in ALS. A 2021 study used a multimodal dialog platform in which patients were instructed to interact with the platform and complete a battery of speaking tasks designed to probe the acoustic, articulatory, phonatory, and respiratory aspects of their speech. All told, the findings showed that multiple acoustic and visual speech metrics demonstrated statistically significant differences between controls, bulbar symptomatic, and bulbar pre-symptomatic patients.2

Click here for more coverage of NEALS 2022.

1. Annagiri S, Simmons Z, Geronimo A. Tracking digital biomarkers of speech in patients with motor neuron disease. Presented at: 2022 Annual NEALS Meeting; November 1-3; Clearwater Beach, FL. Abstract 36
2. Neumann M, Roesler O, Liscombe J, et al. Investigating the utility of multimodal conversational technology and audiovisual analytic measures for the assessment and monitoring of amyotrophic lateral sclerosis at scale. Interspeech. Published August 30, 2021. doi:10.21437/Interspeech.2021-1801.
Related Videos
Michael Levy, MD, PhD
Michael Kaplitt, MD, PhD
Michael Kaplitt, MD, PhD
video 4 - "Amyloid Cascade Hypothesis of Alzheimer’s Disease"
Video 3 - "Amyloid Precursor Protein and Amyloid Beta Species in Alzheimer’s Disease"
Svetlana Blitshteyn, MD, FAAN, director and founder of Dysautonomia Clinic
© 2024 MJH Life Sciences

All rights reserved.