It's been just over 100 days since Noland Arbaughthe first participant in the PRIME experimental study Neuralinkreceived his implant Links at the Barrow Neurological Institute in Phoenix, Arizona. After managing to get Arbaugh to play chess with his mind, Neuralink recently published some updates on the progress made, describing not only the experience of its first patient and the results achieved, but also revealing some problems encountered with its BCI (Brain Computer Interface), or his brain-computer interface.
The Neuralink Brain Implant Problem: What Happened
In the press release issued by the company Elon Muskwe read that «in the weeks following the operation, some wires have retracted from the brain (by Arbaugh, Ed), causing a clear decrease in the number of effective electrodes.” Neuralink also noticed the problem thanks to the fact that Noland is constantly monitored in using the BCI, which he uses for approximately 8-10 hours a day between structured sessions for the purposes of the PRIME study and personal activities. The causes of the detachment are not known at the moment.
When evaluating Link's performance, variations were found BPS (Bits-Per-Second), it standard by which the speed and accuracy of cursor control is measured. This value is calculated using a grid system, as can be seen in the video below.
The higher the BPS values, the better the patient's control of the mouse pointer using the brain-computer interface. During his first research session, Noland set the world record for cursor control with a BCI system, reaching the value of 4.6 BPSwhich he then improved to 8.0 BPS. For comparison, Neuralink engineers using a mouse get values around 10 BPSwhich represent the company's objective.
The results on the first patient: what solution did the Neuralink team find
Due to the problems encountered in the weeks following the surgery, a decrease in BPS values was recorded, as seen in the graph below. To solve this the Neuralink team has changed the recording algorithm, so as to make it more sensitive to signals coming from neurons. The techniques for translating these signals into cursor movements and the user interface have also been improved. This not only led to Noland reaching his initial performance, but even exceeding it!
The development of brain-computer interfaces: where we are
The results achieved by Neuralink in the development of an effective brain-computer interface give rise to hope, especially if we compare them with the work carried out in recent years by other researchers in the field of BCI.
I study BrainGate2 published in 2017 by researchers at Stanford University (in collaboration with several other important bodies), for example, demonstrated how effective BCIs can be for some patients to write texts through the use of a brain-computer interface.
One of the study participants with ALS, in a series of “free typing” sessions, was able to answer questions posed to her by moving the mouse pointer over a keyboard with an optimized layoutin which the characters were arranged to minimize the distance the cursor travels while typing text.
Given the large number of people with disabilities that impede movement, such advances in brain-computer interfaces bode well for the future.
