Digital tests based on speech and smell are being developed to screen for cognitive decline, researchers told the Alzheimer’s Association International Conference in Toronto.

Two speech apps are already some way down the road, testing several markers including speech speed, vocabulary and rhythm in different languages to establish a baseline for testing, a session convened by the Davos Alzheimer’s Collaboration (DAC) heard.

A third initiative using smell is also in the mix, primarily testing people’s ability to smell certain scents via inhalers.

DAC supports an innovation ecosystem to accelerate healthcare solutions to end Alzheimer’s disease globally, and DAC-supported projects in Kenya, India, Egypt and Chile have afforded the companies access to multicultural groups to refine their innovations.

Better screening tests are essential as an estimated three-quarters of people with Alzheimer’s are never diagnosed and, as the burden grows in the global South, tests for low-resource settings as crucial.

TELL’s Adolfo Garcia described his company’s product as “a digital speech biomarker app” that can run on multiple platforms, based on a collection of speech tasks ranging from spontaneous to non-spontaneous tasks.

Several speech features “are very revealing about your mental health status”, said Garcia. 

Using speech timing metrics, for example, the app can measure “the rhythm with which people speak; the number of pauses that they make, how long those pauses are, how variable they are, how long the syllables that they produced are”, said Garcia.

TELL has been tested in over 20 countries with over 40,000 hours of data from around 9000 participants. But Garcia describes the in-depth research with the DAC-supported researchers headed by Dr Karen Blackmon at Agha Khan University in Kenya, as  “phenomenal, instructive and fulfilling.”

Blackmon’s team has been testing Swahili-speaking Kenyans for cognitive decline based on simple speech timing metrics, while TELL has trained a machine learning regressor with various speech metrics. 

There has been a “moderate to strong correlation” between the results from TELL’s machine model and the real-time scores from people that Blackmon’s team has tested on “simple speech timing metrics, which are quite scalable across different languages”, Garcia noted. 

Nicklas Linz of Ki Elements said his speech app aims to find “something that works across languages, across cultural contexts, so that we have something that is culturally fair, neutral and usable in all of these contexts”.

His group has worked with DAC teams in Egypt, India, Kenya and Chile who speak Arabic, Hindi, English, Swahili and Spanish.

Loss of smell and neurodegeneration

Subhanjan Mondal of Sensify said that the idea to use olfaction (smell) as a measure for neurodegeneration “came from COVID”, where many people who contracted the virus lost their sense of smell.

People with the ApoE e4 allele, the gene variant that increases the risk of developing late-onset Alzheimer’s disease, also have an increased risk of olfactory decline.

“There is an anatomical connection between olfaction and neurodegeneration for Alzheimer’s and Parkinson’s and many other neurodegenerative diseases,” said Mondal. “And there is also a genetic component, as ApoE carriers have a strong disposition to olfactory decline.”

Sensify has developed a digital smell test, ScentAware, with smells contained in different inhalers that are QR-coded. Using an app connected to a mobile phone camera, people can conduct the smell test at home or in a clinical setting.

“People found it easy to use, fun, and it can be done in a short time,” said Mondal.

But there is some way to go. The field is so new that there are no common smell elements across cultures. 

“Can this be incorporated somewhere upstream in a screening mechanism in normal individuals with higher risk factors?” Mondal asks. 

Multicultural challenges

“There are a lot of challenges in adapting speech tasks from multilingual contexts like Kenya,” said Aga Khan’s Blackmon.

“But these are challenges that we’re going to have to face across as, in the Global South, multilingualism is a norm in most post-colonial settings where… major languages like English are spoken in schools but not necessarily in homes.”

“It’s been excellent to work with TELL, Ki Elements and Sensify Aware because, in each of these partnerships, we are identifying problems and we’re proactively solving them,” said Blackmon.

For example, an app using automatic speech recognition did not do so well when people were switching languages, she explained.

“Although our samples may seem small, the way that we’re approaching this is to solve problems [before the app is] scaled to larger populations.”

The end goal is to integrate the apps into health systems to enable affordable and accurate early detection.

“These tools are friendlier. Their interfaces have been really well designed. They’re user-friendly. Tools like the olfactory Sensify Aware are fun for patients,” said Blackmon. 

“It’s very different from a standard neuropsychological test setting that I’m used to, where people are sweating. 

“It’s an opportunity to do it well and do it differently with input from stakeholders across the global South. We may even discover new speech features that turn out to be diagnostically relevant, like the number of times someone switches language could tell us something about early signs of dementia.”

Societal and genetic influences

Professor Amy Kind of the University of Wisconsin (UW) addressed how cognitive decline is influenced by both genetic and societal elements, known as the “exposome”.

“The term was first coined in 2005, and it means the integrated compilation of physical, chemical, biological and social influences across an entire life course that influence biology,” said Kind.

“The environmental exposures are modifiable. These are things that we can intervene upon, over and above individual-level factors, to improve health. And this construct is thought of as precision health, not just precision medicine,” she said.

“Individuals who live in adverse exposomes experience poor brain health, and hundreds of studies have shown this,” added Kind.

She heads the largest study of the social exposome in the United States, The Neighborhoods Study, which works with large community-based surveys as well as with tissue from people who have donated their brains to the study to better understand the environmental risks influencing Alzheimer’s.

“We work with brain tissue, and we link it back to the life-course social exposome,” said Kind.

“Across 25 academic institutions, there are over 9,000 descendants’ brain donors in the cohort, and it allows us to link, with some certainty, the association between certain types of exposure – be that occupational, social, toxic, metabolic – to findings within the tissue.

Kind and colleagues try to identify and mitigate the risks for people living in adverse exposomes.

“Are there critical windows of life course across these pathways?

“Perhaps all of our future for our brain health is written in our childhood, [so] gestational and early childhood effects could be profound, as we think about late life brain health,” said Kind.

“Some of our newest work is focused on lead and heavy metal poisoning, because these metals accumulate in the tissue across the life course, and yet lead exposure is so common in our water supply, in the air that we breathe and in other places.”

Kind and colleagues have worked closely with the Inner City Milwaukee Water Works Department in order to decrease the lead line pipe infrastructure to decrease lead in the water supply. 

Lead exposure has been linked to Alzheimer’s and cognitive decline, and a large study was released at the conference this week identifying the impact of leaded gasoline on the memory of older Americans.

Reaching the Global South 

DAC founder George Vradenburg said one of the motives for collaboration is to link the global north and global South.

“The majority of cases, by far, are already in the Global South, and by mid-century, it’s going to be 80% of cases of dementia are in the global South. So we have not fulfilled a patient mission if all we deal with is the top 20% of white people in the United States and Europe,” said Vradenburg.

“DAC brings together researchers, healthcare systems, governments and funders to accelerate progress where it’s most needed,” DAC COO Drew Holzapfel told the meeting.

It is based on three programmes – global cohort development, global clinical trials, and healthcare system preparedness – to address gaps in Alzheimer’s research and treatment.

The cohort development programme aims to “increase the amount of research in diverse populations so that we can find better targets for drug development and associated biomarkers,” said Holzapfel.

DAC’s global clinical trials programme aims to do clinical trials “better, faster, cheaper” around pharmacological and non-pharmacological interventions for brain health in parts of the globe that have never had those types of trials, he explained.

The third component, healthcare system preparedness, aims to prime health systems to implement the innovations. 

“Our implementation scientists like to talk about how the time it takes for an innovation to go from availability to full clinical utilisation is about 17 years. We think that’s too long, so we’re trying to take the high-speed train and put it on high-speed tracks so that we can help patients,” said Holzapfel.

By the end of this year, DAC will have worked in about 70 healthcare systems, implementing new tools for detection and diagnosis for about 60,000 patients.

Image Credits: Cristina Gottardi/ Unsplash.