Miriam Shenchi, Dean’s Professor of Electrical and Computer Engineering and founding director of the USC Center for Neurotechnology, and her team have developed a new machine learning method that separates these patterns from the influence of visual inputs across different subjects. It shows surprisingly consistent brain patterns. .published in work Proceedings of the National Academy of Sciences (PNAS).

When performing various everyday movements, such as reaching for a book, our brains have to take in information, often in the form of visual input — for example, seeing where the book is. Our brain then has to process this information internally to coordinate our muscle activity and execute the movement. But how do millions of neurons in our brain perform such tasks? Answering this question requires studying the collective activity patterns of neurons, but doing so by eliminating the influence of input from the neuron’s intrinsic (aka intrinsic) processes, whether or not they are movement-related. .

That’s what Shenchi, his PhD student Parsa Wahidi, and a research colleague in his lab, Umid Sani, did by developing a new machine learning method that considers both movement behavior and sensory input from neural activity. Makes a model.

“Previous approaches to analyzing brain data have either considered neural activity and input but not behavior, or considered neural activity and behavior but not input,” Shenchi said. said “We developed a method that can consider all three signals — neural activity, behavior and input — when extracting hidden patterns of the brain. Which movements were related to behavior and which were not.”

Shenchi and his team used this method to study three publicly available datasets, during which three different subjects performed one of two different movements, moving a cursor over a grid on a computer screen. Involves moving or moving it to random locations sequentially.

“When using methods that didn’t consider the three signals, the patterns in the neural activity of the three subjects looked different.” Wahidi said. But when the team used the new method to consider all three signals, the neural activity of all three subjects revealed a remarkably consistent hidden pattern that was related to movement. This similarity was despite the fact that the tasks of the three subjects were also different.

“In addition to revealing this new consistent pattern, the method also improved the prediction of neural activity and behavior when machine learning did not consider all three signals, as in previous work.” Sani said. “The new method allows researchers to more accurately model neural and behavioral data by accounting for different measured inputs to the brain, such as sensory inputs in this task, electrical or optogenetic stimulation, or input from different brain regions. Enables.”

This method and the pattern discovered may help to understand how our brains move, guided by the information we receive from the outside world. Further, by modeling the effect of input and isolating internal patterns related to behavior, this method may help develop future brain-computer interfaces that control abnormal brain patterns with external inputs such as those in major depression. Deep brain stimulation therapy by improving .

“We are excited about how this algorithm can facilitate both scientific discoveries and the development of future neurotechnologies for millions of patients with neurological or neuropsychiatric disorders,” Shenchi said.