A human brain could be said to contain galaxies: each has about as many synapses as there are stars in 1,000 Milky Ways. Because of this vastness, neuroscientists like Yale’s Murat Günel have a saying: “If you’ve seen one brain—you’ve seen one brain.” Which is to say: each brain is unique. That complexity makes understanding and addressing neurological conditions, such as autism spectrum disorder (ASD), a difficult challenge.
But what if you could create a digital version of an individual’s brain and then test the model to see what treatments would be most effective? Using advances in computing and artificial intelligence, and with access to an unprecedented amount of data, that’s the approach that Günel and his colleagues are taking.
Günel, the Sterling Professor of Neurosurgery and professor of genetics and of neuroscience at the Yale School of Medicine, and his collaborators have recently received a $27.7 million grant to develop personalized approaches designed to support autistic individuals through tailored neurotechnologies and treatments for co-occurring conditions. These treatments will include new brain-computer interface devices that can modulate, in real time, brain activity to support communication and reduce symptoms without requiring surgical implantation. The team will also test the use of ultrasound and non-invasive magnetic stimulation to influence brain circuits involved in communication, anxiety, and sleep regulation.
Their goal is both to intercede when patients are experiencing acute symptoms and to use these techniques to achieve functional communication with profoundly autistic, non-speaking individuals in order to improve their quality of life. All studies will follow strict privacy, safety, and ethical oversight through Yale’s institutional review process.
“The past few years have been incredibly exciting, because we are now finding new ways to restore neurological functions,” Günel says. “Advances in several different areas, from molecular and cellular therapies to machine learning to computational modeling of the brain, are enabling us to start making meaningful, concrete progress.”
One of the most exciting advances, says Günel, is the possibility that non-invasive techniques could improve the abilities of people with ASD to communicate. These techniques are also aimed at reducing anxiety and sleep disturbances—issues often found in those with ASD.
Thanks to the grant, which comes from the Aligning Research to Impact Autism (ARIA) initiative, Günel—leading a team of nearly three dozen collaborators across Yale—is poised to leverage Yale’s multidisciplinary strengths to pioneer novel tools and technologies to both better understand brain circuitry in people with ASD and deliver precise, real-time interventions that will improve their lives.
“We are grateful to the ARIA initiative for its vision and generous support of this transformative work. It has the potential to change fundamentally how we treat children with profound neurodevelopmental disorders,” says Nancy J. Brown, the Jean and David Wallace Dean of Yale School of Medicine. “It underscores the strength of partnership in driving innovative solutions and transforming lives.”
Vast Datasets
Yale is especially well suited for this kind of innovative approach, says Günel, in part thanks to the large datasets at its disposal. One dataset in particular, built over many years, comes from people who suffer from drug-resistant epilepsy. Neurosurgeons surgically implanted electrodes in their brains to help them manage their seizures. During these surgeries, intracranial recordings of these patients’ brain patterns were collected.
“Very few other institutions have this kind of data,” Günel says. “We are collaborating with Yale New Haven Health in analyzing and curating that dataset, which stretches back decades.”
Other data sources at Yale include functional MRI brain scans and surface electroencephalograms, or EEGs. Data from patients with Angelman Syndrome, who experience severe speech and language impairments, sleep disturbances, and anxiety due to a gene mutation, will also be included in the study thanks to a collaboration with the Foundation for Angelman Syndrome Therapeutics.
All of this data, Günel says, will be used to train AI and then to test potential interventions virtually, using large-scale computer models of individual brains before applying them in clinical studies. The treatments likely to be the most successful can be used on the patients themselves.
Broad Collaboration, Broad Impact
Günel’s team of more than thirty people includes neurosurgeons, neuroscientists, psychiatrists, and psychologists, as well as engineers, data scientists, mathematicians, and ethicists.
Several Yale centers and institutes will also provide services and expertise. The Yale Child Study Center will play a central role in the clinical care of patients involved in the study. Advanced imaging techniques will be provided by the Yale Biomedical Imaging Institute and the Wu Tsai Institute, ensuring precise mapping of brain structures and functions. The School of Engineering & Applied Science’s expertise in computational modeling and AI will support the brain modeling aspects of the initiative. And the collaboration with Yale New Haven Health ensures that the research findings can be readily integrated into clinical practice.
“By bringing together experts and resources from across the university and collaborating with Yale New Haven Health System to make this work accessible to patients, we harness the power of interdisciplinary research to achieve breakthroughs that would not transpire in isolation,” notes Dean Brown.
The advances from this work will benefit not just those with neurodevelopmental disorders like ASD, adds Günel, but also potentially help people who have lost brain function due to stroke, trauma, or neurodegenerative diseases like Alzheimer’s and Parkinson’s.
Engineering Expertise
For the past several years, Yale’s School of Engineering & Applied Sciences has been expanding its research into AI, including a $150 million, five-year commitment that the university made in 2024 to expand AI research. That commitment includes the hiring of over twenty new faculty members as well as significant upgrades to the computing power available at the university. Günel and his collaborators will have access to that expanded capability to develop better, more accurate diagnostic and computational models for treating each patient.
“One of the goals of Yale Engineering is to use the power and promise of AI as a tool for scientific research but also as a force to improve the world,” says Jeffrey Brock, dean of the School of Engineering & Applied Science and the William S. Massey Professor of Mathematics. “This collaboration will help align with that vision by providing an unprecedented approach to individualized treatment for people with autism.”
Yale Engineering’s advances in quantum computing may also play a role in Günel’s work. The brain signal datasets at Yale are extremely complex; in order to tease out the signal from the noise, Günel says, new AI and applied mathematics methods may be necessary. “Quantum computing may give us what we need to understand these signals,” he says.
The 2024 commitment to expanding AI research builds on an existing foundation of AI expertise at Yale; the team includes researchers who specialize in the use of AI in healthcare.
Ideal Alignment
ARIA is a scientific initiative to accelerate understanding and treatment of autism and related neurodevelopmental conditions through alignment, collaboration, and cutting-edge research. It connects emerging research, insights, and promising technologies from across scientific fields to create more therapeutic opportunities for people with profound autism and people on the spectrum who seek additional support.
“I’m incredibly excited with the unique datasets that we have, coupled with our talented faculty, and these new AI approaches,” Günel says. “Where the state of AI and quantum computing is, alongside our new understanding of ASD, it’s perfect timing for this work and we’re very grateful to have received the funding for it.”
Top image: Kanat Yalcin, Arman Afrasiyabi, Eyiyemisi Damisah, and Murat Günel. Data were provided in part by the Human Connectome Project, WU-Minn Consortium (Principal Investigators: David Van Essen and Kamil Ugurbil; 1U54MH091657) funded by the 16 NIH Institutes and Centers that support the NIH Blueprint for Neuroscience Research; and by the McDonnell Center for Systems Neuroscience at Washington University.