Device helps people with paralysis communicate

October 7, 2026

Device helps people with paralysis communicate

Paralysis from strokes or neurodegenerative diseases can cause people to lose their ability to speak and move. This loss affects social participation and quality of life.

To help these individuals communicate, scientists have developed brain-computer interfaces, or BCIs. These systems help translate brain activity into words or movements. However, what's been missing is a BCI that can translate both speech and gestures at the same time, as often occurs during natural communication.

An NIH-funded research team led by Dr. Edward Chang at the University of California, San Francisco sought to develop a BCI that could translate both verbal and non-verbal communication. To do so, they used a tiny, surgically implanted array of electrodes to record study participants' neural signals. The signals were translated into commands that controlled a virtual model of the user, called an avatar. The results were published in Nature Neuroscience on September 14, 2026.

Chang and his team had previously tested the device's ability to translate speech in one person with severe paralysis. They showed the system performed well in translating speech into text, audio, and an avatar's facial expressions.

In the new study, the team tested whether the BCI could translate speech and body movements at the same time. They homed in on signals from the motor cortex, the brain area that controls speech and movement.

The researchers tested the device in three participants with severe vocal tract paralysis. Two of the participants also had severe limb paralysis and the third had moderate limb paralysis. The participants were connected to a virtual full-body avatar to capture upper body movement. Common gestures included a hand wave or thumbs-up signal.

The results showed that a single array of electrodes could capture and decode a variety of speech and gestures at the same time. Also, the device reliably decoded isolated movements from the upper body, mouth, and face of the three participants.

The team tried training the BCI using data from participants attempting speech and gestures at the same time. They also tried training using data from speech or gestures in isolation. Training using both types of data improved the BCI's accuracy at decoding both speech and gestures.

Two of the participants used their avatar to copy specific actions and respond to conversational prompts. The BCI achieved up to 82% accuracy for copying gestures alone and 65%-70% for copying simultaneous gestures and speech. Conversational accuracy was at least 85% for gestures and 75% for speech.

"Conversation is about much more than the words being spoken," Chang says. "It's a multilayered, dynamic process involving the whole motor cortex. This proof-of-concept shows us it's possible for a BCI to restore some of this freedom and flexibility."

In the future, his team plans to transition from a wired system to a fully implantable, wireless version.

Related Links

• Brain-computer device helps man speak
• Brain-computer interface restores natural speech after paralysis
• Decoding inner speech from brain signals
• Scientists translate brain activity into music
• Device allows paralyzed man to communicate with words
• System turns imagined handwriting into text
• Stroke (link is external)
• Amyotrophic Lateral Sclerosis (ALS) (link is external)
• Voice, Speech, and Language (link is external)

References

Simultaneous speech and gesture decoding for multimodal communication in paralysis.(link is external) Brosler SC, Liu JR, Silva AB, Hallinan IP, Kurtz-Miott CM, Wilker JFPD, Tu-Chan A, Ganguly K, Chang EF. Nat Neurosci. 2026 Sep 14. doi: 10.1038/s41593-026-02446-2. PMID: 42736415.

Funding

NIH National Institute on Deafness and Other Communication Disorders (NIDCD).