Scientists have recently created a groundbreaking AI algorithm capable of distinguishing brain patterns linked to specific behaviors. This advancement holds promise for enhancing brain-computer interfaces and facilitating the identification of novel brain patterns.
Maryam Shanechi, who holds the Sawchuk Chair in Electrical and Computer Engineering and is the founding director of the USC Center for Neurotechnology, along with her research team, has devised a new AI algorithm that effectively separates brain patterns associated with certain behaviors. This research, which aims to improve brain-computer interfaces and uncover new brain patterns, has been featured in the journal Nature Neuroscience.
While you’re perusing this article, your brain is managing an array of behaviors.
You might be reaching for a cup of coffee, reading this text out loud for a coworker, and experiencing some hunger all at once. These various actions, like moving your arm, articulating words, and different internal feelings such as hunger, are encoded simultaneously in your brain. This simultaneous encoding leads to intricate and intertwined patterns within the brain’s electrical signals. Consequently, a key challenge is to isolate brain patterns that correspond to a specific behavior, like arm movement, from the myriad of other brain patterns occurring at the same time.
For instance, this isolation is crucial in the advancement of brain-computer interfaces designed to help paralyzed individuals regain movement. Since these patients are unable to convey their intentions to their muscles, brain-computer interfaces translate their intended movements directly from brain activity into control commands for devices like robotic arms or computer cursors.
Shanechi and her former Ph.D. student, Omid Sani, who now works as a research associate in her lab, have developed an innovative AI algorithm to tackle this challenge. The algorithm is called DPAD, which stands for “Dissociative Prioritized Analysis of Dynamics.”
“Our AI algorithm, DPAD, distinguishes the brain patterns related to the behavior of interest, such as arm movement, from all other concurrent brain activities,” Shanechi explained. “This enables us to decode movements from brain activity with greater accuracy than previous techniques, potentially improving brain-computer interfaces. Additionally, our method can identify new brain patterns that might otherwise go unnoticed.”
Sani elaborated, “A vital part of the AI algorithm involves first identifying brain patterns associated with the desired behavior and emphasizing these patterns during the training of a deep neural network. Afterwards, the algorithm can learn to recognize the remaining patterns without them interfering with those related to the behavior of interest. Furthermore, neural networks provide great flexibility in describing various types of brain patterns.”
Beyond just movement, this algorithm might be adaptable for future applications aimed at decoding mental states like pain or depression. This capability could enhance treatment for mental health issues by monitoring patients’ symptom states, allowing for more personalized therapy adjustments.
“We are thrilled to develop and showcase extensions of our method for tracking symptoms in mental health conditions,” Shanechi remarked. “This progress could pave the way for brain-computer interfaces that target not only movement disorders and paralysis but also mental health challenges.”
