Background

Frank R. Willett, PhD, is an Assistant Professor of Neurosurgery at Stanford University and co-director (with Jaimie Henderson) of the Stanford Neural Prosthetics Translational Laboratory (NPTL). Willett is a computational neuroscientist and engineer who specializes in developing machine learning algorithms to decode intended speech and movement from brain signals. His work focuses on making BCIs practical and effective for restoring communication to people who cannot speak due to paralysis from ALS, stroke, and other neurological conditions.

Key Contributions

Willett developed innovative neural decoding algorithms for speech BCIs that have achieved remarkable performance. Using a phoneme-based decoding approach, he and his team enabled a patient with severe ALS to achieve a 9.1% word error rate on a 50-word vocabulary (2.7 times better than previous BCIs) and demonstrated the first successful large-vocabulary speech BCI with a 125,000-word vocabulary at 23.8% word error rate. Most impressively, Willett’s decoding achieved 62 words per minute—3.4 times faster than the previous record and approaching natural conversation speed of 160 words per minute. These achievements represent major advances in practical BCI performance. His machine learning approach breaks attempted speech into constituent phonemes and stitches them together intelligently, enabling natural-sounding decoded speech.

Current Work

Willett continues to advance speech BCI technology at Stanford, focusing on improving decoding accuracy, speed, and robustness. His research explores how to leverage large language models to improve speech decoding, how to adapt decoders to individual patients’ neural patterns, and how to extend BCIs beyond speech to restore multiple forms of communication and control. Through the Stanford Neural Prosthetics Translational Laboratory, Willett works closely with neurosurgeons, engineers, and clinicians to ensure that BCI technologies reach patients who need them.