Researchers at EPFL have developed a new brain-machine interface chip called MiBMI that is much smaller than Neuralink’s chip but achieves a high accuracy for converting thought to text.
The MiBMI consists of two thin chips measuring just 8mm2 total, much smaller than Neuralink’s 23x8mm chip. It also uses less power.
In tests, the MiBMI achieved 91% accuracy in converting neural activity read from previous experiments into text in real-time.
It works by identifying distinctive neural codes or DNCs that represent individual letters, acting as a shorthand to efficiently process brain signals.
The small size and low power consumption brings practical implantable solutions for communication closer for people with motor impairments.
While not yet implanted in humans, the researchers believe the MiBMI shows promise for uses like speech decoding and movement control beyond text.
The system currently decodes 31 characters but they aim to expand that to 100. It represents an advancement in ultra-compact, high accuracy brain-computer interfaces.
Source: New Atlas
