People with severe paralysis have always faced a drastic choice: undergo risky surgery to implant neuroprosthetics, or give up the ability to communicate effectively. Today, thanks to a scientific breakthrough by Meta, this choice may no longer be necessary.
A new system called Brain2Qwerty he is able to read brain activity from the outside and turn it into written text, paving the way for a new era in assistive communication.
Neuroprosthetics of the present and the future
A team of researchers led by Jarod Levy e Mingfang Zhang has developed an innovative method to decode sentences directly from brain activity. The system, called Brain2Qwerty, uses techniques of deep learning to interpret brain signals recorded via electroencephalography (EEG) or magnetoencephalography (MEG) as participants type previously memorized sentences on a QWERTY keyboard.
The most striking aspect of MEG is its ability to capture neural activity at an impressive 5.000 Hz, much higher than the 0,5 Hz of traditional functional magnetic resonance imaging (fMRI). This high temporal resolution allows for decoding almost in real time even of visual perceptions.
The results are really promising: using MEG, the system achieves an average error rate per character of 32%, significantly exceeding the performance obtained with EEG (67%). In the best cases, the model achieved an error rate of just 19%.
The technology behind Brain2Qwerty
The operation of Brain2Qwerty is based on an architecture of deep learning trained to recognize patterns of brain activity associated with typing. Analysis of errors suggests that decoding depends not only on motor processes but also on higher-level cognitive factors.
I am particularly struck by how this research manages to combine the analysis of typographical errors with the study of higher brain processes. We are not just talking about interpreting simple motor commands, but about understanding the complex intertwining of thought and action that characterizes human communication.
Researchers, including Svetlana Pinet e Jeremy Rapin, they demonstrated that the system can work even with sentences never seen during training, suggesting a real understanding of the brain mechanisms involved in language production.
Real world applications
The potential impact of this technology extends far beyond the medical field. In the field of neurorehabilitation, patients with communication disabilities could interact with their surroundings using only thought. In the field of Gaming and virtual reality, players could control game environments or navigate virtual realities effortlessly, simply by thinking.
Researchers Hubert Jacob Banville e Stephane of Ascoli emphasize how these findings significantly reduce the gap between invasive and non-invasive methods, paving the way for the development of safe brain-computer interfaces for non-communicating patients.
The research, conducted under the supervision of Jean Remi King, is an important step towards the democratization of neuroprosthetics, making this technology potentially accessible to a much larger number of people in need.
Advances in cognitive research
A particularly interesting aspect of this technology is its potential for cognitive research. Scientists could obtain unprecedented insights into how the brain processes visual information in real time. This deeper understanding of cognitive processes could lead to significant advances in our understanding of how the brain works.
The collected data on 35 healthy volunteers show that the technology is reliable and reproducible. MEG, in particular, has demonstrated superior performance compared to EEG, suggesting that it may be the preferred modality for future clinical applications.
Neuroprosthetics, translating the intangible
Our thoughts can become something as concrete as written text. The road to clinical implementation is still long, but the results of this research open up previously unthinkable scenarios. The future of neuroprosthetics could be much more accessible and less invasive than we thought just a few years ago: a true milestone for both healthcare and neuroscience.
As this technology continues to evolve, we can expect to see increasingly sophisticated and intuitive applications that could radically transform the way we interact with the digital world through our thoughts.
