Neurodegenerative disorders: Can digital speech testing help reduce global health inequity?

The global rise in life expectancy stands as a major societal achievement. Yet, this progress brings significant challenges – chiefly the sharp and unprecedented rise in neurodegenerative disorders, especially Alzheimer’s and Parkinson’s disease.

Such ageing-related conditions currently affect over 60 million people, with more people losing their memories, mobility and independence, increasingly every day

In fact, by 2050, their prevalence will double in high-income countries, with dementia cases growing threefold in low- and middle-income ones.

Marked by memory and motor deficits, respectively, Alzheimer’s and Parkinson’s are major causes of disability, caregiver burden and even death. They also entail a huge financial burden: for example, current global direct and indirect costs of Alzheimer's, estimated at nearly $1.5 trillion, are projected to approximate 10 trillion by 2050.

Worryingly, these diseases remain incurable; while some treatments ease symptoms, effective disease-modifying therapies are still in development. Hence, these disorders pose serious challenges for patients, families and health systems at large.

Early detection proves vital in this context, as it can accelerate the adoption of neuroprotective lifestyle changes, facilitate timely financial planning and reduce costs by fostering routine over emergency care.

Nevertheless, standard diagnostic tools (clinical tests, brain scans, biofluid markers) are expensive, stressful, subject to scheduling delays and unavailable in many cities. A pressing quest has thus begun for affordable, patient-friendly, immediate, scalable markers for disease detection and monitoring.

Such requisites are met by digital speech biomarkers. Here, participants are asked to speak and digital tools extract acoustic (e.g. pitch changes) and linguistic (e.g. word selection) features from their recordings and transcripts, respectively.

Such features are highly informative because different aspects of speech and language hinge on brain regions and networks that are distinctly affected by these disorders. Subtle anomalies in natural speech can then be framed as indicators of specific neurocognitive dysfunctions, representing a powerful asset for clinical testing.

Digital speech biomarkers have been used to detect specific neurodegenerative diseases, capture symptom severity, predict underlying brain anomalies and yield new endpoints for clinical trials.

For example, recent works have reached success of almost 90% in detecting persons with Alzheimer’s using automated analysis of the words they use and persons with Parkinson’s using speech timing and articulatory measures, all based on two-minute tasks.

The approach is gaining momentum as it is cost-effective (reducing the need for expert staff), non-invasive (avoiding discomfort and risks), time-efficient (offering on-the-fly results) and applicable for home-based testing (overcoming geographical barriers).

As such, digital speech biomarkers have been noted as a promising tool to counter worldwide inequities in neurodegeneration assessments. In particular, low- and middle-income countries account for 60% of Alzheimer's cases but less than 25% of global investment in research, prevention, diagnosis and treatment.

Technologies such as these could enable massive testing, better phenotyping and more efficient monitoring in such contexts, reducing the gap with well-served populations.

Paradoxically, digital speech biomarkers run the risk of becoming a new source of inequity, as most companies and research initiatives exclusively target populations from high-income countries.

Fortunately, groundbreaking initiatives aim to subvert this tendency. Created in 2022, the Include Network brings together over 150 researchers from 90 centres in nearly 30 countries to boost the discovery of speech biomarkers that generalize across world regions.

Also noteworthy are the projects led by the Davos Alzheimer’s Collaborative, bringing digital speech testing and other clinical innovations to underserved populations, including in Africa and India.

This pursuit is supported by science-first companies, such as Toolkit to Examine Lifelike Language, which provides speech and language testing in over 15 countries across Latin America, Africa, Europe and the United States.

Yet, additional efforts are needed. Increased attention to digital biomarkers from funding agencies (such as the National Institutes of Health in the United States or the European Research Council in Europe) should be strategically focused on these scalable technologies.

Also, goal-oriented alliances should be made between policymakers and speech biomarker experts for the massive deployment of relevant platforms in clinics worldwide.

Moreover, direct use of these systems in household settings could allow the detection of persons at risk for such disorders even before key diagnostic symptoms become evident. All these actions should prioritize underserved regions, where low-cost, expert-independent tools such as these can make the strongest impact.

Overall, this use of digital language tools demonstrates how they can be used as a force for good. Automated speech testing constitutes a powerful asset in reducing global disparities around Alzheimer’s and Parkinson’s.

But as we await definitive breakthroughs in neurodegeneration diagnostics, we must act now – by leveraging the words people speak, we can help bridge the gap in care and equity.