An NIA-sponsored study indicates that early alterations in speech patterns could signal frontotemporal degeneration (FTD) in genetically predisposed individuals before clinical symptoms manifest. Published in the journal Neurology, the findings propose that digital speech analysis could serve as an effective tool for the premature detection of FTD.
In this research, a team from the University of Pennsylvania and Columbia University examined speech recordings from subjects in longitudinal studies spanning two decades. These subjects were segmented into two cohorts: one consisting of individuals who carried a familial genetic mutation associated with a heightened risk of FTD, and the other comprising individuals without this genetic predisposition.
FTD is characterized by profound personality shifts and the deterioration of language skills. Unlike most dementia types which tend to appear later in life, FTD often strikes at an earlier age, and its symptoms can lead to misdiagnosis. Notably, none of the study participants exhibited symptoms at the time their speech was recorded.
The recordings were made as participants described an image from the Boston Diagnostic Aphasia Examination—a standard assessment tool for aphasia. Aphasia entails the impairment in the ability to comprehend or articulate language, often due to damage to the brain’s language and speech areas. While commonly a result of stroke or head trauma, aphasia can also be an early sign of FTD and other progressive dementias.
Leveraging innovative speech analysis technology, the study team meticulously observed the evolution in the participants’ speech characteristics over numerous years, using recurring tests. They meticulously examined an array of 30 speech elements, encompassing speech pace, the incidence and length of pauses, vocabulary richness, usage of linguistic fillers such as “um” and “uh”, truncated words, redundancies, and word vagueness.
The data revealed a consistent divergence in speech patterns between the group with a familial predisposition for FTD and those without. Over time, the at-risk group’s narrative descriptions became more concise, with reduced word usage compared to those at lesser risk. Typically, the speech of the at-risk group consisted of briefer phrases, a greater inclination towards ambiguous terminology, and a frequent insertion of interjections, like “oh” or “well.” Brain imaging of the at-risk group paralleled these findings, showing a decrement in gray matter within the cerebral language circuitry over time.
The investigative team regards their findings and the speech examination algorithms applied as promising tools that could address the current deficiency in FTD diagnosis, discerning more nuanced and precocial speech alterations than standard cognitive evaluations. They anticipate that this avenue of digital speech and language analytics might evolve into a novel resource for both researchers and healthcare practitioners, enhancing the identification of individuals at potential risk for developing FTD in both medical settings and clinical investigations.
However, the researchers noted the preliminary nature of their findings, referencing the limited scope of participants due to the scarcity of speech data on asymptomatic individuals with familial FTD tendencies. Their aspiration is to broaden their research to encompass more comprehensive brain imaging metrics and to engage a wider pool of participants in upcoming studies.