In an initial study, Canary analyzed speech recordings with AI and identified early stage Alzheimer’s with 96 percent accuracy.
Dave Arnold retired in his 60s and began spending time volunteering in local schools. But then he started misplacing items, forgetting appointments and losing his sense of direction. Eventually he was diagnosed with early stage Alzheimer’s.
“Hearing the diagnosis made me very emotional and tearful,” he said. “I immediately thought of all my mom had experienced.” His mother suffered with the condition for years before passing away. Over the last year, Arnold has worked for the Alzheimer’s Association as one of its early stage advisors, sharing his insights to help others in the initial stages of the disease.
Arnold was diagnosed sooner than many others. It's important to find out early, when interventions can make the most difference. One promising avenue is looking at how people talk. Research has shown that Alzheimer’s affects a part of the brain that controls speech, resulting in small changes before people show other signs of the disease.
Now, Canary Speech, a company based in Utah, is using AI to examine elements like the pitch of a person’s voice and their pauses. In an initial study, Canary analyzed speech recordings with AI and identified early stage Alzheimer’s with 96 percent accuracy.
Developing the AI model
Canary Speech’s CEO, Henry O’Connell, met cofounder Jeff Adams about 40 years before they started the company. Back when they first crossed paths, they were both living in Bethesda, Maryland; O’Connell was a research fellow at the National Institutes of Health studying rare neurological diseases, while Adams was working to decode spy messages. Later on, Adams would specialize in building mathematical models to analyze speech and sound as a team leader in developing Amazon's Alexa.
It wasn't until 2015 that they decided to make use of the fit between their backgrounds. ““We established Canary Speech in 2017 to build a product that could be used in multiple languages in clinical environments,” O'Connell says.
The need is growing. About 55 million people worldwide currently live with Alzheimer’s, a number that is expected to double by 2050. Some scientists think the disease results from a buildup of plaque in the brain. It causes mild memory loss at first and, over time, this issue get worse while other symptoms, such as disorientation and hallucinations, can develop. Treatment to manage the disease is more effective in the earlier stages, but detection is difficult since mild symptoms are often attributed to the normal aging process.
O’Connell and Adams specialize in the complex ways that Alzheimer’s effects how people speak. Using AI, their mathematical model analyzes 15 million data points every minute, focusing on certain features of speech such as pitch, pauses and elongation of words. It also pays attention to how the vibrations of vocal cords change in different stages of the disease.
To create their model, the team used a type of machine learning called deep neural nets, which looks at multiple layers of data - in this case, the multiple features of a person’s speech patterns.
“Deep neural nets allow us to look at much, much larger data sets built out of millions of elements,” O’Connell explained. “Through machine learning and AI, we’ve identified features that are very sensitive to an Alzheimer’s patient versus [people without the disease] and also very sensitive to mild cognitive impairment, early stage and moderate Alzheimer's.” Based on their learnings, Canary is able to classify the disease stage very quickly, O’Connell said.
“When we’re listening to sublanguage elements, we’re really analyzing the direct result of changes in the brain in the physical body,” O’Connell said. “The brain controls your vocal cords: how fast they vibrate, the expansion of them, the contraction.” These factors, along with where people put their tongues when talking, function subconsciously and result in subtle changes in the sounds of speech.
Further testing is needed
In an initial trial, Canary analyzed speech recordings from phone calls to a large U.S. health insurer. They looked at the audio recordings of 651 policyholders who had early stage Alzheimer’s and 1018 who did not have the condition, aiming for a representative sample of age, gender and race. They used this data to create their first diagnostic model and found that it was 96 percent accurate in identifying Alzheimer’s.
Christian Herff, an assistant professor of neuroscience at Maastricht University in the Netherlands, praised this approach while adding that further testing is needed to assess its effectiveness.
“I think the general idea of identifying increased risk for cognitive impairment based on speech characteristics is very feasible, particularly when change in a user’s voice is monitored, for example, by recording speech every year,” Herff said. He noted that this can only be a first indication, not a full diagnosis. The accuracy still needs to be validated in studies that follows individuals over a period of time, he said.
Toby Walsh, a professor of artificial intelligence at the University of New South Wales, also thinks Canary’s tool has potential but highlights that Canary could diagnose some people who don’t really have the disease. “This is an interesting and promising application of AI,” he said, “but these tools need to be used carefully. Imagine the anxiety of being misdiagnosed with Alzheimer’s.”
As with many other AI tools, privacy and bias are additional issues to monitor closely, Walsh said.
Other languages
A related issue is that not everyone is fluent in English. Mahnaz Arvaneh, a senior lecturer in automatic control and systems engineering at the University of Sheffield, said this could be a blind spot.
“The system may not be very accurate for those who have English as their second language as their speaking patterns would be different, and any issue might be because of language deficiency rather than cognitive issues,” Arvaneh said.
The team is expanding to multiple languages starting with Japanese and Spanish. The elements of the model that make up the algorithm are very similar, but they need to be validated and retrained in a different language, which will require access to more data.
Recently, Canary analyzed the phone calls of 233 Japanese patients who had mild cognitive impairment and 704 healthy people. Using an English model they were able to identify the Japanese patients who had mild cognitive impairment with 78 percent accuracy. They also developed a model in Japanese that was 45 percent accurate, and they’re continuing to train it with more data.
The future
Canary is using their model to look at other diseases like Huntington’s and Parkinson’s. They’re also collaborating with pharmaceuticals to validate potential therapies for Alzheimer’s. By looking at speech patterns over time, Canary can get an indication of how well these drugs are working.
Dave Arnold and his wife dance at his nephew’s wedding in Rochester, New York, ten years ago, before his Alzheimer's diagnosis.
Dave Arnold
Ultimately, they want to integrate their tool into everyday life. “We want it to be used in a smartphone, or a teleconference call so that individuals could be examined in their home,” O’Connell said. “We could follow them over time and work with clinical teams and hospitals to improve the evaluation of patients and contribute towards an accurate diagnosis.”
Arnold, the patient with early stage Alzheimer’s, sees great promise. “The process of getting a diagnosis is already filled with so much anxiety,” he said. “Anything that can be done to make it easier and less stressful would be a good thing, as long as it’s proven accurate.”
In 2013, the Human Brain Project set out to build a realistic computer model of the brain over ten years. Now, experts are reflecting on HBP's achievements with an eye toward the future.
Scholars have found that the project did help advance neuroscience more than some detractors initially expected, specifically in the area of brain simulations and virtual models. Using an interdisciplinary approach of combining technology, such as AI and digital simulations, with neuroscience, the HBP worked to gain a deeper understanding of the human brain’s complicated structure and functions, which in some cases led to novel treatments for brain disorders. Lastly, through online platforms, the HBP spearheaded a previously unmatched level of global neuroscience collaborations.
Simulating a human brain stirs up controversy
Right from the start, the project was plagued with controversy and condemnation. One of its prominent critics was Yves Fregnac, a professor in cognitive science at the Polytechnic Institute of Paris and research director at the French National Centre for Scientific Research. Fregnac argued in numerous articles that the HBP was overfunded based on proposals with unrealistic goals. “This new way of over-selling scientific targets, deeply aligned with what modern society expects from mega-sciences in the broad sense (big investment, big return), has been observed on several occasions in different scientific sub-fields,” he wrote in one of his articles, “before invading the field of brain sciences and neuromarketing.”
"A human brain model can simulate an experiment a million times for many different conditions, but the actual human experiment can be performed only once or a few times," said Viktor Jirsa, a professor at Aix-Marseille University.
Responding to such critiques, the HBP worked to restructure the effort in its early days with new leadership, organization, and goals that were more flexible and attainable. “The HBP got a more versatile, pluralistic approach,” said Viktor Jirsa, a professor at Aix-Marseille University and one of the HBP lead scientists. He believes that these changes fixed at least some of HBP’s issues. “The project has been on a very productive and scientifically fruitful course since then.”
After restructuring, the HBP became a European hub on brain research, with hundreds of scientists joining its growing network. The HBP created projects focused on various brain topics, from consciousness to neurodegenerative diseases. HBP scientists worked on complex subjects, such as mapping out the brain, combining neuroscience and robotics, and experimenting with neuromorphic computing, a computational technique inspired by the human brain structure and function—to name just a few.
Simulations advance knowledge and treatment options
In 2013, it seemed that bringing neuroscience into a digital age would be farfetched, but research within the HBP has made this achievable. The virtual maps and simulations various HBP teams create through brain imaging data make it easier for neuroscientists to understand brain developments and functions. The teams publish these models on the HBP’s EBRAINS online platform—one of the first to offer access to such data to neuroscientists worldwide via an open-source online site. “This digital infrastructure is backed by high-performance computers, with large datasets and various computational tools,” said Lucy Xiaolu Wang, an assistant professor in the Resource Economics Department at the University of Massachusetts Amherst, who studies the economics of the HBP. That means it can be used in place of many different types of human experimentation.
Jirsa’s team is one of many within the project that works on virtual brain models and brain simulations. Compiling patient data, Jirsa and his team can create digital simulations of different brain activities—and repeat these experiments many times, which isn’t often possible in surgeries on real brains. “A human brain model can simulate an experiment a million times for many different conditions,” Jirsa explained, “but the actual human experiment can be performed only once or a few times.” Using simulations also saves scientists and doctors time and money when looking at ways to diagnose and treat patients with brain disorders.
Compiling patient data, scientists can create digital simulations of different brain activities—and repeat these experiments many times.
The Human Brain Project
Simulations can help scientists get a full picture that otherwise is unattainable. “Another benefit is data completion,” added Jirsa, “in which incomplete data can be complemented by the model. In clinical settings, we can often measure only certain brain areas, but when linked to the brain model, we can enlarge the range of accessible brain regions and make better diagnostic predictions.”
With time, Jirsa’s team was able to move into patient-specific simulations. “We advanced from generic brain models to the ability to use a specific patient’s brain data, from measurements like MRI and others, to create individualized predictive models and simulations,” Jirsa explained. He and his team are working on this personalization technique to treat patients with epilepsy. According to the World Health Organization, about 50 million people worldwide suffer from epilepsy, a disorder that causes recurring seizures. While some epilepsy causes are known others remain an enigma, and many are hard to treat. For some patients whose epilepsy doesn’t respond to medications, removing part of the brain where seizures occur may be the only option. Understanding where in the patients’ brains seizures arise can give scientists a better idea of how to treat them and whether to use surgery versus medications.
“We apply such personalized models…to precisely identify where in a patient’s brain seizures emerge,” Jirsa explained. “This guides individual surgery decisions for patients for which surgery is the only treatment option.” He credits the HBP for the opportunity to develop this novel approach. “The personalization of our epilepsy models was only made possible by the Human Brain Project, in which all the necessary tools have been developed. Without the HBP, the technology would not be in clinical trials today.”
Personalized simulations can significantly advance treatments, predict the outcome of specific medical procedures and optimize them before actually treating patients. Jirsa is watching this happen firsthand in his ongoing research. “Our technology for creating personalized brain models is now used in a large clinical trial for epilepsy, funded by the French state, where we collaborate with clinicians in hospitals,” he explained. “We have also founded a spinoff company called VB Tech (Virtual Brain Technologies) to commercialize our personalized brain model technology and make it available to all patients.”
The Human Brain Project created a level of interconnectedness within the neuroscience research community that never existed before—a network not unlike the brain’s own.
Other experts believe it’s too soon to tell whether brain simulations could change epilepsy treatments. “The life cycle of developing treatments applicable to patients often runs over a decade,” Wang stated. “It is still too early to draw a clear link between HBP’s various project areas with patient care.” However, she admits that some studies built on the HBP-collected knowledge are already showing promise. “Researchers have used neuroscientific atlases and computational tools to develop activity-specific stimulation programs that enabled paraplegic patients to move again in a small-size clinical trial,” Wang said. Another intriguing study looked at simulations of Alzheimer’s in the brain to understand how it evolves over time.
Some challenges remain hard to overcome even with computer simulations. “The major challenge has always been the parameter explosion, which means that many different model parameters can lead to the same result,” Jirsa explained. An example of this parameter explosion could be two different types of neurodegenerative conditions, such as Parkinson’s and Huntington’s diseases. Both afflict the same area of the brain, the basal ganglia, which can affect movement, but are caused by two different underlying mechanisms. “We face the same situation in the living brain, in which a large range of diverse mechanisms can produce the same behavior,” Jirsa said. The simulations still have to overcome the same challenge.
Understanding where in the patients’ brains seizures arise can give scientists a better idea of how to treat them and whether to use surgery versus medications.
The Human Brain Project
A network not unlike the brain’s own
Though the HBP will be closing this year, its legacy continues in various studies, spin-off companies, and its online platform, EBRAINS. “The HBP is one of the earliest brain initiatives in the world, and the 10-year long-term goal has united many researchers to collaborate on brain sciences with advanced computational tools,” Wang said. “Beyond the many research articles and projects collaborated on during the HBP, the online neuroscience research infrastructure EBRAINS will be left as a legacy even after the project ends.”
Those who worked within the HBP see the end of this project as the next step in neuroscience research. “Neuroscience has come closer to very meaningful applications through the systematic link with new digital technologies and collaborative work,” Jirsa stated. “In that way, the project really had a pioneering role.” It also created a level of interconnectedness within the neuroscience research community that never existed before—a network not unlike the brain’s own. “Interconnectedness is an important advance and prerequisite for progress,” Jirsa said. “The neuroscience community has in the past been rather fragmented and this has dramatically changed in recent years thanks to the Human Brain Project.”
According to its website, by 2023 HBP’s network counted over 500 scientists from over 123 institutions and 16 different countries, creating one of the largest multi-national research groups in the world. Even though the project hasn’t produced the in-silico brain as Markram envisioned it, the HBP created a communal mind with immense potential. “It has challenged us to think beyond the boundaries of our own laboratories,” Jirsa said, “and enabled us to go much further together than we could have ever conceived going by ourselves.”
