Researchers have developed a tool based on artificial intelligence (AI) which makes it possible to translate thoughts into written text and which, in the long term, could help people with aphasia, a language disorder, to communicate.
- Aphasia is a language disorder that can lead to a partial or total loss of speech, with difficulties in communicating orally and writing.
- Researchers have developed a new tool, based on artificial intelligence (AI), which makes it possible to translate text into text.
- Ultimately, scientists hope that this tool can be used by people with aphasia to help them communicate.
In France, aphasia affects near 300,000 people each year. People who have this language disorder can undergo a partial or total loss of speech, with difficulties in communicating orally and writing. Aphasia generally follows a stroke (stroke), head trauma, brain tumor, degenerative disease or infection in the brain.
An artificial intelligence tool to translate writing thoughts
Researchers of the University of Texas, in Austin, the United States, have just developed a new tool, based on artificial intelligence (AI), which could ultimately make it possible to translate the thoughts of people with text aphasia. Thus, these patients would communicate more easily with those around them. Their work was published in the journal Current Biology.
In previous research, scientists had already developed this AI -based tool. This first model was already able to translate the brain activity of a person in text. But the algorithm needed a 16 -hour learning. Meanwhile, the patient had to remain motionless in a scanner while listening to podcasts … a duration far too long for patients with aphasia.
In this new version, researchers have improved learning time. The algorithm no longer needs only one hour with the patient. This speed comes from the fact that, from now on, the AI trains the brain activity of a new person on the brain of a patient whose activity was previously used to cause the cerebral decoder. With this base, the tool reaches the same result but much faster.
Next step, the test with patients with aphasia
“”[Cet outil] shows a deep overlap between what’s going on in your brain when you listen to someone tell you a story and what’s going on in the brain when watching a video that tells a story, indicates Alex Huth, principal author, in a press release. Our brain treats both types of stories in the same way. This also tells us that what we decode is not really language. These are representations of something above the level of language which are not linked to the modality of the entry“Thoughts therefore transcend language because the tool uses data from the brain, which have not been translated into speech.
One of the limits of this study is that the participants were not affected by aphasia. But, during their work, the researchers imitated the brain damage patterns of aphasic people and showed that the decoder was able to translate thought into text. Even if this remains to be proven by conducting studies on people with aphasia, researchers are convinced that their tool will work on this target audience.
“”It is fun and rewarding to think about how to create the most useful interface and make [l’apprentissage de l’outil] as simple as possible for participants, Indicates Jerry Tang, one of the authors. I am really delighted to continue working on how our decoder can be used to help people. “
