After a car crash in 2003, a 20-year-old’s life changed drastically when a post-surgery blood clot caused him to experience a brainstem stroke. The man, known by his nickname Pancho, woke up from a coma a week later only to realize that he had become paralyzed and could no longer speak. Many of the current technologies that aim to improve patient quality of life through assisted communication consist of slow and effortful typing-based interfaces. To advance these efforts at patient autonomy, Pancho agreed to work with neuroscience researchers at the University of California San Francisco (UCSF) to help them develop a novel approach that decodes words directly from a patient’s brain activity.

The research team implanted a sheet of electrodes in his brain to detect signals from speech-related areas and transmit these signals through a wire to a computer. In 50 sessions across 81 weeks, they asked Pancho to try to say words even though he was unable to physically speak them. Through this incredible feat of science and engineering, the words that Pancho intended to say showed up on the computer screen. The neural implant was placed in an area of the sensorimotor cortex that orchestrates speech by controlling the rapid articulatory movements of the vocal tract. Detected activity patterns in this area of the brain were then translated and classified into words using deep-learning algorithms and computational models. In this way, the electrodes allowed Pancho to communicate about 15 words per minute with near 50% accuracy. They were even able to recognize complete sentences, and his first one was, “My family is outside.”

This groundbreaking achievement could be expanded to help many other patients suffering from speech-impairing conditions and brain injuries. A true pioneer, Pancho was able to show the preservation of the brain’s pathways for speech in a person who had not been able to speak for over 15 years. As more patients participate, the team hopes to engineer more sensitive electrodes that can be implanted wirelessly, and they may even discover individual brain variations. With this novel technology comes questions about its future use for reading people’s thoughts, but right now, it is just about giving a voice to those who wouldn’t otherwise have one.

Dr. Caetano Reis e Sousa is currently an Assistant Research Director and Senior Group Leader of the Immunobiology Laboratory at the Francis Crick Institute. Dr. Enzo Poirier is a post-doctoral fellow at the Francis Crick Institute.

Managing Correspondent: Lauren Davancaze

Press Articles: “Tapping Into the Brain to Help a Paralyzed Man Speak,” The New York Times

Original Journal Article: “Neuroprosthesis for Decoding Speech in a Paralyzed Person with Anarthria,” New England Journal of Medicine

Image Credit: The New York Times

One thought on “Scientists Detect Brain Signals to Help a Paralyzed Man Speak

  1. The great approach of the deep learning algorithm. Scientists are working on this. I am myself learning artificial intelligence. Accuracy improvement is increasing by analyzing models. Python and R are advancing day by day. I think the 99% accuracy could be gain up to 2030 itself.

Leave a Reply

Your email address will not be published. Required fields are marked *