UBC researchers get Best Paper Award for brain disorder identification techniques 

Imagine a future where medical doctors can detect brain disorders such as ADHD and autism early and accurately, aided by advanced AI tools. 

Former UBC Master of Computer Science student Sadaf Sadeghian and her collaborators, have been working toward helping to make that notion a reality. Their recent research initiative proposed a new Machine Learning (ML) model, called HyperBrain, that could help scientists understand and treat those who are neurodivergent.  

The research gained recognition when Sadeghian presented the work at the Machine Learning for Cognitive Neuroscience Workshop in Marrakesh, Morocco on October 10, 2024. This workshop was held as part of the prestigious MICCAI (Medical Image Computing and Computer-Assisted Intervention) conference and is one of the leading events in medical AI and neuroscience. Sadeghian’s poster and oral presentationintroduced the work to the international scientific community, and the judges gave it Best Paper Award.  

The paper: HyperBrain: Anomaly Detection for Temporal Hypergraph Brain Networks. 

Sadeghian’s studies and research in ML and her interest in neuroscience came together while she was pursuing her Master of Science in the University of British Columbia’s Computer Science (CS) Department under the mentorship of Dr. Margo Seltzer. Seltzer is a professor, renowned for her work in computer systems, is also co-Head of the Computer Science department and senior member of the Systems laboratory, Systopia. 
 
The research is a collaboration with Dr. Seltzer and Dr. Xiaoxiao Li, an assistant professor for UBC’s Electrical and Computer Engineering (ECE) Department and Associate Member of the UBC Computer Science Department. Dr. Li’s research includes work in medical imaging and neuroscience. Their combined expertise and enthusiasm for the topic led to the exciting inter-disciplinary opportunity.  

Getting hyper-focused with HyperBrain 

In her research, Sadeghian used temporal hypergraphs to model the brain in a way that captures complex relationships between multiple brain regions. Conventional brain models tend to simplify brain connections, linking only two regions at a time without considering the group interactions that often occur in real life. Sadaf’s approach incorporates these “higher-order” interactions that involve several brain regions activating together, which provides a more realistic representation of how the brain functions. Her work also explores the role of time (temporal factors) in understanding brain function. Rather than treating brain activity as static, the model captures how brain region interactions change over time.  
 
Sadeghian said, “HyperBrain helps to answer questions such as ‘What patterns show us something is wrong?’ and ‘How can we use these patterns to help doctors make a diagnosis sooner?’” 
 
HyperBrain could help researchers and doctors move beyond basic classifications of brain health, allowing them to pinpoint specific patterns of abnormal activity associated with neurodivergency. This knowledge may also help in understanding individual symptoms and   lead to more personalized treatments. 

“It’s about making real-world impact,” Sadeghian said, “and I believe this research can help us get closer to that goal. I’ve always found the brain fascinating and it’s amazing to think our work could contribute to help doctors in their work by using ML and AI.” 

Looking to the future 

While her model is still in the research phase, Sadeghian envisions a future where AI tools like hers can greatly assist doctors. Her goal is not for AI to replace doctors but to provide powerful tools to support their work. “My hope is that it will aid doctors in detecting issues early, improve the diagnoses, and aid in tailoring medications,” she said. 

Reflecting on the journey that brought her here, from an early interest in listening to neuroscience podcasts to working alongside experts such as Dr. Seltzer, Sadeghian said, “I’m really grateful to have such supportive mentors. They’ve helped me grow as a researcher and find my voice.” She explained that Dr. Seltzer played a key role, guiding Sadeghian through the complexities of modeling brain activity and helping her explore how ML could capture intricate brain patterns. “And Dr. Li provided such valuable insight and feedback, also helping with the submission process for the paper.”  

She added, “Contributing to the neuroscience community in this way has been very satisfying. It’s been so meaningful to me and an interesting opportunity to learn more about neuroscience. Having this paper published makes me very happy!”  

Dr. Li added, “Human brains are incredibly complex systems. We’ve seen how AI, particularly neural networks inspired by brain function, can advance our understanding of the brain, especially in addressing neurodisorders. Collaborating with Sadaf and Dr. Seltzer has been invaluable, and I’m thrilled to see the growing cross-discipline interest among computer science researchers and students in this field.” 

More about Sadaf Sadeghian:  She graduated from UBC this past May with her MSc and is now a data scientist and AI engineer for CHUBB, an insurance company based out of Toronto. In her role, she utilizes advanced machine learning methods such as Generative AI, graph analytics, and reinforcement learning on insurance data to solve business challenges in forecasting, pricing, underwriting, and risk engineering. Prior to attending UBC, she studied Computer Engineering at the University of Tehran. 

More about Dr. Margo Seltzer: Dr. Seltzer is currently a co-Head for UBC Computer Science and a Canada 150 Research Chair, and has been elected to the Royal Society of Canada. Her research interests are in systems, construed broadly: systems for capturing and accessing provenance, file systems, databases, transaction processing systems, storage and analysis of graph-structured data, new architectures for parallelizing execution, interpretable machine learning, graph learning for intrusion detection, and systems that apply technology to problems in healthcare. 

More about Dr. Xiaoxiao Li:  Dr. Li is as an Assistant Professor in the Electrical and Computer Engineering Department at UBC, is an Associate Member of the CS Department at UBC, and is also a Faculty Member at the Vector Institute. Additionally, Dr. Li is recognized as a CIFAR AI Chair. Dr. Li’s research interests focus on the intersection of AI and healthcare, aiming to develop next-generation responsible and trustworthy AI algorithms and systems. Before joining UBC, Dr. Li was a Postdoctoral Fellow at Princeton University and earned a PhD from Yale University.