New technology enables the collection and analysis of speech behaviors, but this requires an extensive database, which the ADDF seeks to help develop.
Nicole Bjorklund, PHD
The Alzheimer’s Drug Discovery Foundation (ADDF) has called for the formation of a consortium of clinicians, researchers, and data scientists to help develop and monitor biomarkers such as changes in speech and language that can be indicative of early decline in Alzheimer disease (AD).1
Alterations in speech and language have shown promise as early biomarkers in AD. Technologies and data analytics for analyzing speech and language are ever improving, and the growing prevalence of personal smart devices enable easy remote data collection.
"Rapidly expanding use of smart devices, such as smart phones and digital wearables, is making it easier than ever to collect large amounts of speech and language data," said lead author Nicole Bjorklund, PhD, assistant director, Scientific Affairs, ADDF, in a statement.2 "What we need now is a unified approach for collecting, analyzing and sharing this information to create algorithms that can predict who will go on to develop Alzheimer's."
These technologies can analyze biomarkers associated with early AD and its progression. These include acoustic features such as pitch and amplitude and lexical and syntactic aspects of speech and features of written language such as text contextual or semantic information.
The ADDF noted that while necessary, a comprehensive speech-sample database would be costly and hard to achieve for individual research groups and startups. Therefore, the foundation is calling for a comprehensive, harmonized, open-access speech-sample repository that is a joint effort of physicians, data scientist, researchers, and organizations.
"We are optimistic that generating this type of high-quality data would be incredibly enabling for the field. In particular, approaches to data privacy are continually evolving and best practices should be implemented, revisited and refined when appropriate as the repository takes shape," co-author Kristina Malzbender, MPH, associate director, Health and Life Sciences, Gates Ventures, added to the statement.
The ADDF envisions the creation of a speech sample repository that pulls from a large, diverse, cohort of patients at different disease stages that represent different accents, languages, speech and language components. All samples must be characterized with harmonized protocols and standards. The costs associated with the formation of such a database, such as those associated with positron emission tomography, magnetic resonance imaging, and blood-based biomarkers used in participant characterization, put such an endeavor out of reach for most individual organizations.
"By combining the strength of experts in dementia research, linguistics, data analytics, and clinical trials, we can generate a gold standard data set," co-author Lampros Kourtis, PhD, adjunct assistant professor, Tufts University and managing director, Circadic, said in the statement. "We can then comb through this data to find the patterns consistent with early signs of disease."
Howard Fillit, MD, founding executive director and chief science officer, ADDF concluded the statement by saying "alone, researchers have not been able to take full advantage of the opportunities digital technology afford, but together we can facilitate truly seismic shifts in neurodegeneration research."