Succeeded in developing a system that can predict with high accuracy which of prepared questions and answers were selected from brain signals
by
Many scientists and doctors are challenged with the task of 'converting brain signals into words,' for those who can not speak words, write words, or type letters on the keyboard because of illness. You Researchers at the University of California, San Francisco have developed a system that predicts pre-prepared questions and answers from human brain signals and can communicate as quickly as natural conversation.
Real-time decoding of question-and-answer speech dialogue using human cortical activity | Nature Communications
https://www.nature.com/articles/s41467-019-10994-4
Neuroscientists decode brain speech signals into written text | Science | The Guardian
https://www.theguardian.com/science/2019/jul/30/neuroscientists-decode-brain-speech-signals-into-actual-sentences
At the time of writing, people who can not speak words and can not move freely can communicate widely using eye movements and muscle cramps to point out special characters on dials. I am However, with this method it takes time to convey characters, and we can not communicate at the same speed as people talk.
The research team of Edward Chan, who is a neurosurgeon and senior researcher at the University of California at San Francisco, is developing a system that enables people with severe disabilities to communicate more smoothly. 'To date, there is no auxiliary system that allows people with disabilities to interact on the same time scale as conversations,' said Chang.
The research by Chang et al., Funded by Facebook, was conducted in collaboration with three epilepsy patients who visited neurosurgery for surgery. Before surgery, patients were patched with an electrode patch that monitored brain activity for at least a week to determine the site of the epileptic seizure. Using the electrode patch, Chang and others scanned the patient's brain activity.
by
During hospitalization, each patient who could speak normally heard a total of nine questions and responded by choosing from a list of 24 answers. The research team has built and trained a computer model that matches these questions and response patterns with patterns of brain activity.
The model after training is based on only the intracerebral signal pattern without speech, with 76% accuracy for “which question was heard” and 61% accuracy for “how to answer”. It could be identified almost instantly. By identifying the response from the brain signal, it is possible to display in text the content that the patient wants to answer on a display or the like.
In this experiment, it was possible to use the brain signal to answer the genre of music that the patient likes, and answer questions such as whether the room temperature is hot or cold, or whether it is bright or dark. 'It's important to keep in mind that this experiment was accomplished with only a very limited vocabulary, but it is important to keep in mind that we are able to translate what we can translate in the future,' said research team David Moses . I would like to expand the sex and breadth. '
by
Although this research is a big step for creating a system for people with disabilities to communicate smoothly, many issues still remain. One is to improve the software so that brain signals can be converted to speech as well as text.
And the other task is to read 'sentences that are only spoken in the heart.' Although this experiment used the in-brain signal sent to move the lips, tongue, and jaw to read the words that the subject wants to speak, in patients with certain injuries or neurological disorders, they try to give words There is a possibility that brain signals can not be detected sufficiently. Therefore, in order to target people with wider disability, it is necessary to develop a method to detect sentences created in the brain.
However, scanning and reading words that you think in your brain can pose an ethical issue, as it may reveal intimacy that you do not want to let others know. 'Even if it's technically possible, I'm not interested in developing a technology to read words that I think in my brain,' said Chang. On the other hand, if there is a patient who wants to communicate but is suffering because of his disability, doctors and scientists have a responsibility to restore human basic ability to 'communicate with others'. I claimed to be there.
Related Posts:
in Science, Posted by log1h_ik