New device allows decoding the thoughts of patients with paralysis

An international team of researchers has achieved a significant breakthrough in communication for people with paralysis. Thanks to an electrode interface connected to a computer, it is now possible to identify words that a person only imagines saying , without having to actually say them. This development opens the door to a new form of dialogue for those who have lost the ability to speak due to diseases such as ALS or stroke.

The experiment was carried out as part of the BrainGate2 clinical trial , which seeks to connect the human brain to machines using electrical signals. One of the participants, Casey Harrell, implanted with microneedles in his motor cortex, allowed the computer to record his brain signals. Thanks to artificial intelligence , the system predicted nearly 6,000 words with 97.5% accuracy , recreating Harrell's own voice from previous recordings.

The team led by neuroscientist Erin Kunz compared the signals of intended words and imagined words , managing to identify thought words with up to 70% accuracy , a figure that improves with additional training and examples of inner speech.

The breakthrough raises ethical challenges related to the privacy of thoughts . During testing, the system was able to detect words that participants did not wish to communicate, such as numbers thought of while counting objects on the screen. This highlights the need for controls that protect patient autonomy.

The researchers developed two main mechanisms:

1. Differentiation of intended and internally spoken speech , blocking the decoding of unwanted thoughts.
2. Mental password , where patients imagine a unique phrase to activate decoding. This method achieved up to 98.75% accuracy , ensuring that the computer only translates thoughts when intended.

Experts such as Marcus Lionel Brown emphasize that these measures protect patient autonomy , while others, such as Evelina Fedorenko of MIT, caution that the technology cannot yet decode the spontaneous content of the human mind .

Although brain-machine interfaces do not yet allow for the translation of the free flow of thoughts , the progress of BrainGate2 represents an encouraging precedent . Soon, patients who have lost their voice could once again communicate through simple thoughts, provided adequate safeguards for their privacy and control over the information are maintained.