Brain-computer interface or BCI measures and transforms brain activity into an artificial output which improves or complements the central nervous system output. The brain-computer interfaces are used to artificially control the movements of limbs and prosthetic devices. While BCIs have made groundbreaking progress over the years, these interfaces still have some drawbacks to date.
One of the biggest hurdles is the persistent need to recalibrate the BCI. Recalibration is required because the BCI lacks the ability to adapt to the learning process of the brain. Users are forced to stop in the middle of what they are doing and have to reset the connection continuously, making it a cumbersome process. This is like us using a smartphone and recalibrating it again and again in the middle of an activity.
Talking about the problem, Dr. Karunesh Ganguly, associate professor, UCSF Department of Neurology, stated:
To counter the problem, a research team from the University of California, San Francisco, has come up with the first-ever “plug and play” brain prosthetic machine that enabled an a paralysed individual to control the computer curosr using his brain. Some modifications to the traditional algorithm allowed the BCI to function without having to reset again and again.
In this experiment, the team used ECoG electrode arrays for the BCI, that is, a collection of electrodes about the size of a post-it note. This pad of electrodes is usually surgically implanted on the brain’s surface. For this experiment, the scientists received a special approval to implant the BCI for a longer duration of time to study the stability and efficacy of long term prosthetic devices. Why this was important is so that the BCI got atuned to the natural learning process of the brain without any external recalibration.
With a lot of tweaks, the researchers had the algorithm update consistently without requiring any external resetting. Over the days, an evident improvement was noticeable in terms of the interplay between the brain and the algorithm. The person’s brain established a solid control over the BCI interface through a consistent mental ‘model’. The interesting part is that the daily recalibration of the BCI was eliminated at this stage. Dr. Ganguly, also the senior author of the study, stated:
After weeks of continuous learning, when the BCI was finally reset (as opposed to the daily resetting), the person restored the same neural activity to operate the system. This means that the machine learning algorithm was able to retrain to the former state. Ultimately, the researchers could do away with the need to update and retrain the algorithm altogether. The participant could use the device for over 44 days without any retraining. No deterioration of performance was observed during this period. In fact, event after 44 days, a very little decline in performance was noticed.
Perhaps, with further research, these plug-and-play brain decoding devices can become the new normal as it is already observed to have outperformed the traditional pincushion-style BCI devices. Albeit ECoG arrays are less receptive as compared to the traditional counterparts, the long-term stability seems to make up for it. Perhaps BCI can be used to operate commercial assistive technology and to improve the functionalities. The next phase of the research involves experimenting with the long-term stability of this interface to control complex prosthetic devices. Long-term stability of BCI will be extremely handy for those with congenital anomalies and neurological disorders as this will bring them one step closer to living life with minimal physical restrictions. Ganguly stated:
A detailed report of the study has been published in the journal Nature Biotechnology.