ALS patient sets communication record via brain implant: 62 words per minute

In the new research, the Stanford team wanted to know if neurons in the motor cortex also contain useful information about speech movements. That is, could they detect how “subject T12” was trying to move its mouth, tongue, and vocal cords as it attempted to speak?

These are small, subtle movements, and according to Sabes, a big finding was that only a few neurons held enough information to allow a computer program to predict, with good accuracy, the words the patient was trying to say. This information was transmitted by Shenoy’s team to a computer screen, where the patient’s words appeared as they were spoken by the computer.

The new result builds on previous work by Edward Chang of the University of California, San Francisco, who wrote that speech involves the most complicated movements people make. We expel the air, add vibrations that make it audible, and turn it into words with our mouth, lips, and tongue. To make the “f” sound, you place your upper teeth over your lower lip and push the air out – just one of dozens of mouth movements needed to speak.

A way forward

Chang previously used electrodes placed above the brain to allow a volunteer to speak through a computer, but in their preprint, the Stanford researchers say their system is more accurate and three to four times faster.

“Our results show a possible route to restoring communication with paralyzed people at conversational speeds,” wrote the researchers, who included Shenoy and neurosurgeon Jaimie Henderson.

David Moses, who works with Chang’s team at UCSF, says the current work is reaching “impressive new performance benchmarks.” Yet even as records continue to be broken, he says, “it will become increasingly important to demonstrate stable and reliable performance over multi-year timescales.” Any commercial brain implant could struggle to pass regulators, especially if it degrades over time or if registration accuracy decreases.

WILLETT, KUNZ ET AL

The way forward will likely include both more sophisticated implants and closer integration with artificial intelligence.

The current system already uses a few types of machine learning programs. To improve its accuracy, the Stanford team used software that predicts which word usually comes next in a sentence. “I” is more often followed by “am” than “ham,” even though these words sound similar and might produce similar patterns in someone’s brain.