Automated speech analysis may track MS brain atrophy over time
Smaller vocabulary, frequent repetition could be early signs
Written by |
Researchers analyzed speech patterns in MS patients over two years. (Image from iStock)
- Automated speech analysis provides a noninvasive method for tracking brain tissue loss and neurodegeneration in multiple sclerosis.
- Early linguistic declines, such as narrower vocabulary and phrase repetition, correlate with subsequent gray matter atrophy.
- This cost-effective speech monitoring approach offers an accessible tool for evaluating cognitive disease progression without added patient burden.
Automated analyses of a person’s everyday speech patterns could offer a simple, noninvasive way of tracking brain tissue loss in multiple sclerosis (MS), according to a study.
Researchers found that changes in speech over a period of two years — in particular, a narrower vocabulary range and more frequent repetition of phrases — were associated with the loss of gray matter in the brain over the following five years. Gray matter is tissue rich in nerve cell bodies that mediate information processing, movement, cognition, and other brain processes.
It remains to be seen whether the data, collected entirely from Czech-speaking participants, can be replicated across other patient groups and in different languages, the scientists noted.
The study, “Automated speech analysis reveals progressive linguistic changes associated with subsequent gray matter loss in multiple sclerosis,” was published in the Multiple Sclerosis Journal.
Language issues common in MS
Cognitive MS symptoms can emerge early in the disease course and may progress independently of physical disability. Language problems are also increasingly recognized, with evidence suggesting that as many as 75% of people with MS have language-related difficulties.
These difficulties have traditionally been assessed using clinical tests in which people are asked to perform language-based tasks. Advances in automatic speech recognition and natural language processing now make it possible to objectively analyze spontaneous speech and measure features such as vocabulary diversity, sentence structure, and repetition.
Previous studies have found subtle differences in these measures between people with MS and healthy individuals, but most have examined patients at a single time point, leaving it unclear whether automated speech analysis can capture disease progression over time.
Researchers in the Czech Republic set out to quantify language declines in MS patients over two years and assess whether those early changes were associated with brain tissue loss during the subsequent five years.
The study involved 97 MS patients and 80 healthy people, who served as a control group. MS patients had been living with MS for a mean of 14 years. Most patients (84) had relapsing-remitting MS, nine had secondary progressive MS, and four had primary progressive MS.
Participants completed speech assessments at the start of the study and again two years later, narrating a familiar fairy tale of their choice for about 90 seconds. Recordings were automatically converted to text and analyzed for various language features.
Compared with healthy participants, people with MS showed a significant decline in vocabulary range (the diversity of words they used) and an increase in the use of repeated phrases over two years.
MS patients also exhibited declines in sentence development — a measure of speech complexity that looks at the number of clauses a person uses to connect ideas in a sentence — although this difference did not remain statistically significant in the final analyses.
The researchers said the observed changes may reflect difficulties retrieving words and efficiently accessing the brain networks involved in connecting words and their meanings. Increased repetition may indicate reduced flexibility in generating new words and phrases during spontaneous speech.
The team next examined whether these early language changes were related to later brain atrophy, or the loss of brain tissue. Starting with the two-year speech assessment, patients underwent repeated MRI scans for up to five more years. The analysis ultimately included 608 MRI assessments from 89 people who had at least two scans.
Over that time, significant brain volume loss was detected in MS patients, including loss of both gray and white matter, which largely contains the nerve cell projections that carry signals between brain regions.
Atrophy was also observed in the thalamus, an important relay center for sensory and cognitive information, and the cerebellum, which helps coordinate movement, balance, and aspects of cognition. No significant volume loss was seen in the hippocampus, a region important for memory and learning.
Changes in several speech measures during the first two years were associated with the subsequent rate of gray matter loss. In particular, reduced vocabulary range and greater phrase repetition were associated with greater total gray matter loss.
At the regional level, syntactic complexity — a measure of the variety and complexity of sentence structure — was significantly associated with gray matter loss in the cortex. Changes in vocabulary range, phrase repetition, sentence development, and syntactic complexity were also significantly associated with cerebellar volume loss.
The scientists noted that the study was limited by changes in MRI protocols over time and by the fact that all participants were Czech-speaking, which limits generalizability to other languages.
Despite these limitations, the findings suggest that “expressive language analysis may provide a novel and accessible tool for monitoring neurodegeneration in patients with MS,” the researchers concluded, noting that the approach is cost-effective and imposes no added burden on patients or caregivers.
Leave a comment
Fill in the required fields to post. Your email address will not be published.