A man paralyzed from the neck down can now quickly write sentences with his mind, thanks to a brain-computer interface (BCI) developed by Stanford University researchers. The BCI allows the man (a research study participant known as “T5”) to create letters on a computer by imagining that he is writing them by hand. During the study, T5 created “mindwriting” on a computer in almost the same amount of time it takes to text on a smartphone using fingers and thumbs.
The research team previously developed a BCI that allowed study participants, including T5, to use their minds to move a cursor on a keyboard for typing. In that study, published in 2017, T5 mentally typed about 40 characters per minute. Building on that research, the team designed algorithms that help interpret the brain’s signals for handwriting, which includes curves, different directions, and speed changes. Interestingly, the BCI can interpret these complex movements more easily and quickly than simpler cursor movements.
In 2016, a neurosurgeon on the team implanted two chips on the left side of T5’s brain. Each chip is about the size of a baby aspirin and contains 100 electrodes. The electrodes sense signals meant to control hand movements as the nerves fire within the motor cortex on the brain’s outer surface.
A computer connected to the chip by wires receives the signals; AI algorithms then analyze them and project the intended motion, which appears on the computer screen. When asked to copy unfamiliar sentences, T5 wrote about 90 characters per minute (or about 18 words per minute). That’s just a little slower than the 23 words per minute observed when texting the same sentences by hand on a smartphone and more than twice the speed of T5s typing in the 2017 study.
When handwriting his thoughts freestyle, T5 averaged about 15 words per minute (the slightly reduced speed reflects brief pauses during the thought process involved in freestyle). With the help of an auto-correct function similar to those in a smartphone, T5 achieved a margin of error much better than some BCIs: below 1% for copying and slightly above 2% for freestyle.
The research team published their findings in Nature. While the Stanford system has years of development ahead, T5’s success illustrates the enormous potential for this type of BCI. It could help patients communicate quickly, easily, and clearly, even after losing the ability to write or speak due to injury, neurological disease, or strokes. That would transform the lives of millions of individuals worldwide.