The brain interface allows speech decoding and computer control in the patient ALS

University of California, Davis researchers have developed a brain interface (BCI) which allows control of the computer cursor and click, using neural signals from the speech Cortex. A participant with amyotrophic lateral sclerosis (SLA) used the interface for daily life activities, including independent control of a personal desktop and text entrance.

Neurological diseases such as strokes or ALS can interrupt the brain path to muscles, causing loss of movement and communication. The ALS gradually destroys the paths of the upper and lower neurons, leaving intact cognition but causing paralysis in the four members and an alteration of significant speech.

Brain interfaces are intracortical implanted devices that bypass any disturbance by reading neural signals directly from the brain and producing output in the name of the user. Many BCIs have relied on the neural activity of the Dorsal engine cortex, a region of the brain associated with hand and arms movements. When the signals are decoded, users can move a cursor by trying or imagining the movement of the members.

On the other hand, the BCI of speech are based on the ventral precentral gyrus, where neural signals are linked to facial movements and the articulation of speech. The decoding of neural signals of this region allows rapid communication based on speaking, but it has not been demonstrated to support general IT navigation or movement control.

The implantation in dorsal and ventral areas would be ideal, but it is considered to be impractical surgically or unrealizable. Consequently, users and clinicians must choose between control of the cursor and the decoding of speech.

In the study, “the speech Cortex allows control of the BCI cursor and click” Journal of Neural EngineeringThe researchers conducted a case study with a single participation to test whether the neural activity of the speech Cortex could support both the control of the cursor and the decoding of speech with a single implant site.

A participant in Als, a 45 -year -old man suffering from paralysis in the four members and difficulty speaking clearly, participated in research. All sessions were organized at the participant’s home.

Four 64-electrodes of electrodes were surgically implanted in the participant’s pre-central ventral gyrus. The targeting of the electrodes was guided by the preoperative MRI and the cortical alignment with the Human Connectome project.

Neuronal signals were acquired at a sampling rate of 30 kHz and a filtered bandwidth between 250 and 5,000 Hz. Threshold passages and the power of the strip of tips were calculated each millisecond of each electrode. These characteristics were then grouped into bins of 10 milliseconds, producing a flow of vector of characteristics at 512 dimensions which served as a contribution to the decoding systems.

Three task paradigms were used to assess system performance: radial calibration8, grid assessment and simultaneous speech and cursor. A linear speed decoder decoder has checked the movement of the cursor, while a separate linear classifier has decoded click events.

The decoder parameters have been continuously recalibrated using linear regression for speed and logistical regression for click classification with updated weights every few seconds during active control.

The calibration occurred quickly while the participant acquired his first target using neural control within 40 seconds after launching the system.

During subsequent sessions with optimized parameters, the participant used the system to control the cursor with high efficiency, with an average of 2.90 bits per second. Previous sessions have shown lower performance, with an average of 1.67 bits per second. The highest rate recorded in any session was 3.16 bits per second. One bit per second corresponds to the ability to make several specific choices per minute, with higher values ​​indicating faster and more precise control.

Out of 1,263 total trials, 1,175 targets were properly selected, corresponding to a 93%precision. Eighty-eight incorrect selections occurred and no trial ended due to the waiting period. Six clicks were recorded on temporarily disabled targets and 23 clicks occurred outside any target limit.

Click on the classification performance exceeded by chance on the four electrodes. A well -placed painting contributed the most to the decoding of the cursor and corresponded closely to the performance of the complete decoder.

In sessions involving simultaneous control of speech and cursor, the acquisition time of the median target increased to 4.51 seconds. The speech without speech varied from 3.37 to 3.51 seconds, illustrating that the production of speech interfered with the participant’s ability to control the cursor, but has not caused delays in sequential actions. Improvements of the design of the decoder could alleviate interference and improve future conviviality.

A single implant site has supported communication and computer functions in an independent domestic framework, providing concept proof for the feasibility of multimodal BCI systems.

For patients intact cognitively but unable to use their members or to speak, a neural interface that provides both control of computer cursors and speech decoding can restore crucial, independence channels and considerably improve quality of life.

© 2025 Science X Network