Neuralink and Brain–Computer Interfaces: Engineering the Link Between the Human Brain and Machines
Department of Artificial Intelligence and Data Science, Ajeenkya DY Patil School of Engineering, Pune, India · India
Abstract
An implanted brain–computer interface (BCI) allows a person with a disability to Paralysis use a computer to read the activity of motor-cortex neurons. Its performance is largely dependent on two factors:It depends primarily on the number of how many and how long recording channels it can record from. keep working. This paper takes an engineering perspective on the Neuralink N1 implant. point of view. It is made of flexible polymer threads that are able to bend. robots installed in the brain connected to electronic devices embedded in the robot. implant itself. We take the specifications that the company has. published or reported, use them to work out data-rate and insertion-time budgets, and build a synthetic motor cortex to see how The more the channels and to the loss, the better the decoding quality. of whole threads. For our simulation, we achieved the best result with a decoder using a Kalman filter. R2 = 0.200 ± 0.055 with 16 channels, 0.866±0.004 with 1024 channels, and 0.917±0.006 with 3072 channels. After 85% of the Because the decoder that didn’t change, it fell to, a negative R2 (−0.096±0.146), even though its correlation was still 0.777. Reconfiguring decoder to the ones that still existed brought R2 back up to 0.661±0.020. It would take 1024 channels of raw samples to be streamed off the device. The data rate is approximately 205 Mb/s, which is approximately 1250 times the data rate required for binned spike. counts. The reason for compression is that gap. implant. All the simulated results are based on the assumed data and are stated assumptions, therefore they are only pointing in the direction – not necessarily. measurements of the real device.
Keywords
How to Cite This Article
Rights
BODHIVRUKSHA JOURNAL OF DIVERSE DISCIPLINE