Engaging new cognitive and motor signals to improve communication prostheses

Stanford University, Stanford CA

Investigators

Linked publications, trials & patents

Abstract

PROJECT SUMMARY Communication technology for people with severe speech and motor impairments (SSMI) continues to improve, with recent advances being made in the neural control of communication devices. In prior NIDCD- supported research, our research team developed high-performance intracortical brain-computer interfaces (BCIs) to enable point-and-click control of computer cursors and virtual hands via brain activity alone. Despite these advancements, typing interfaces remain limited in communication speed (8 words per minute) due to the inherent slowness of selecting keys one by one with a pointing device. Recently, we set a new communication record with a speech BCI that decoded neural activity evoked by attempted speech into text (62 words per minute). Speech BCIs currently show the most promise for achieving the long-standing goal of restoring fluent communication at conversational speeds to people with SSMI. However, speech BCIs still require further improvements in performance to allow for more widespread adoption. The goals of this project are to improve speech BCI performance using new methods and brain signals in 3 ways: (1) reduce the calibration burden on the user by translating self-recalibrating algorithms we have demonstrated in handwriting to the decoding of speech, (2) improve ease-of-use and speed by leveraging signals related to inner speech (as opposed to motor signals related to attempted speech, which can be slow and difficult for people with SSMI), and (3) improve accuracy by leveraging complementary signals from Broca's area that represent abstract linguistic features of the user's intended message, as opposed to the articulator movements that make it up. Upon completion, this project will advance both the capabilities of speech BCIs for communication and our understanding of the function of speech-related areas of the brain.