Deep learning refines how bionic eyes communicate with the brain

Researchers from the University of California, Santa Barbara (UCSB) and other institutions have demonstrated that artificial intelligence (AI) could enhance the precision, predictability, and responsiveness of future visual prostheses, such as bionic eyes. UCSB associate professor of computer science Michael Beyeler and his colleagues utilized a deep-learning model to design electrical stimulation patterns for electrodes implanted in the visual cortex of a blind participant.

This model improved the researchers' control over neuron responses and aided in predicting the participant's perceptions. This proof-of-concept study, published in the journal Neuron, marks a step towards developing visual cortical prostheses that better communicate with the brain.