TELL Is Building the Leading Voice Biomarkers for Latin America
Fernando Johann, PhD, and his team are bringing groundbreaking voice analysis research to patients through their AI-driven platform. With a strong focus on native Spanish and non-English speakers, TELL is focusing on making early Alzheimer’s detection more accessible.
Investors, learn how you can become a Health Moonshot Champion and support Health Transformers.
Challenge
You can tell a lot about a person from the way they speak. For centuries doctors have been trained to listen closely to their patients' voices for signs of illness, as subtle changes like slurring and tremors can be a sign that something is going on beneath the surface.
Over the years science has tripled down on this idea, going deeper and deeper into the connection between speech and our brains, muscles, and respiratory system.
What once might have been casually called “listening closely” or “performing a good history and physical” has blossomed into a robust academic field for voice biomarkers. Technology has allowed researchers to go far beyond the abilities of a human ear and detect changes in the voice that are so subtle they may predict disease many years in the future.
One of the great applications for digital voice biomarkers is in the early detection of Parkinson’s and Alzheimer’s disease. Well published research suggests that AI-assisted voice analysis can be incredibly useful in these diagnoses, while also having the benefit of being one of, if not the most, accessible and non-invasive detection methods possible.
So far so good. Here’s the problem. For many years, the research that forms the foundation of voice digital biomarkers for Alzheimer’s has been stuck in academia. Researchers have used it, but it has not scaled and has not achieved its potential for opening up access to care.
That’s where TELL comes in.
Origin Story
About 15 years ago Adolfo García, PhD, started researching neurodegeneration and language. Over time he became a leading academic voice, particularly in Latin America, on the connection between speech and the brain. He published prolifically and had a vision for how this work could transform patients’ lives through early disease detection.
But that didn’t happen.
Year after year, even as he made new discoveries, his studies never seemed to make it beyond the realm of research. That began to change when he met Facundo Carrillo, who had completed a PhD in computer science on the topic of natural language processing. By putting their heads together, they were able to build the MVP of a product that brought Garcia’s research into a usable digital format. They distributed it to twelve of their colleagues, who started using the app in research and clinical scenarios.
The initial response was very positive, and the duo was approached by a life science fund in Argentina that was interested in scaling the technology. They needed a seasoned CEO to bridge the worlds of technology and business and tasked engineer-turned-VC Fernando Johann, PhD, with the task of finding this right-fit executive. The more Johann worked on the project – and reflected on the loss of his own two grandmothers to Alzheimer’s – the more he realized that he might be the man for the job. Everyone agreed, and TELL kicked into high gear.
Under the Hood
TELL can take different forms, depending on the application, so it’s good to start with what’s happening beneath the surface. The foundation is a body of voice data that’s constantly growing and improving through dozens of academic collaborations. These collaborations are training TELL’s AI models, making them smarter and smarter.
These AI models have the ability to be used in the analysis of a single voice, at a single moment in time. They also have the ability to analyze changes in one voice over time, and provide longitudinal trend data, grading a person’s voice against their own personal benchmarks. This personal benchmarking – understanding how a person’s voice has altered from their norm, not the global norm – begins to look a lot more like personalized medicine, says Johann.
There are three types of tasks that are typical to a TELL assessment. The first is called “spontaneous elicitation” and it’s a spontaneous prompt that requires you to think on the spot. For example, “tell me about your morning routine.” Then they have “guided elicitation” where the individual responds to a fixed prompt, like describing an image. Finally they have “forced elicitation” which is focused on the person’s mechanical abilities, like, “say the letter R until you run out of air.”
Within these categories, the actual questions are customizable to each patient population.
“Voice biomarkers are a match made in heaven for everyone,” says Johann. “Voice is the most robust way of accessing the body non-invasively. Voice allows for increased engagement in treatment. Voice allows for early signs of behavior change. For example, in Parkinson's, the fluctuation is part of the actual condition. So we could actually report that you are at this part of the cycle. You are at this part of the cycle, and you would go and do other testing. It's a great tool to justify the trip to the doctor's office.”
Unlike some Alzheimer’s tests, TELL’s voice analysis is a measure of subtle disease symptoms that already exist, not a prediction of what may come later.
TELL has a unique potential for adding to the global world of knowledge around speech, because of their access to native Spanish speakers. For Johann this “human data acquisition” piece of the project gives the project even greater importance.
“We have a competitive advantage in Latin America because we are the only model trained in Spanish by Spanish-speaking participants.”
Their focus on Latin America also means that Johann and his team are addressing a region of extreme need, with an eye towards radical accessibility. There are very few geriatricians and neurologists available in Latin America, and TELL has the potential to break open brain health diagnostics for millions of people. TELL can be deployed over any VOIP (voice over IP) phone line, like a WhatsApp or Zoom call.
“In the future this benchmarking could even be done at the DMV, just like you test your eyes and ears,” says Johann.
Final Word
Fernando Johann, PhD, and his team at TELL bring an exciting and fascinating new dimension to StartUp Health’s Alzheimer’s Moonshot. They open the door to an entire field of study – voice biomarkers – and add that knowledge to the collective progress being made in blood biomarkers, eye tracking, and even brain stimulation.
We appreciate TELL’s vigorous academic approach. They are continually making new studies and leading their field in research.
“We are publishing quarterly,” says Johann. “And we are creating new biomarkers almost every quarter.”
We’re also bullish on TELL because they’ve already successfully landed major research collaborations, currently deploying their technology with 22 institutions. That gives them a firm foundation, helping them grow their data set without spending a lot of money.
The use cases for TELL are broad, and their early non-academic partners have ranged from the Davos Alzheimer’s Collective (with a project in Kenya) to a pharmaceutical company in Brazil.
In the field of voice biomarkers, there is talk, says Johann, that in five to 10 years we will have passive analysis of speech happening through technology like Alexa and Google Home and your smartphone. There are obviously privacy concerns attached to that dream, and we’re not there yet, but that’s where Johann sees TELL heading.
Perhaps the most exciting thing of all is the potential for TELL to aid clinical trials and increase the speed of development of brand new Alzheimer’s treatments.
“We are developing the best microscope out there [for voice analysis]. We want others to use it so they can be the heroes that cure this disease.”
Join us in welcoming Fernando Johann, PhD, Adolfo García, PhD, and the team at TELL to StartUp Health’s Alzheimer’s Moonshot.
Connect with TELL via email
Call for Alzheimer’s Innovation
With support from the Alzheimer’s Drug Discovery Foundation (ADDF) and Gates Ventures, we’ve launched a new global initiative created to develop a collaborative innovation community alongside leading companies, research teams, and stakeholders with a mission to accelerate progress in prevention, diagnosis, and management of Alzheimer’s disease and related dementias. Learn more and apply for an Alzheimer’s Moonshot Fellowship.
Become a Health Moonshot Champion
Health moonshots are fueled by passionate families, foundations, and industry organizations committed to achieving health moonshots. Learn how you can join other champions of the T1D Moonshot or one of our other global health moonshots.
Follow us on social media for daily updates on Health Transformers: X (Twitter), LinkedIn, Facebook, and Instagram.
Published: Jul 26, 2024