Нейроинтерфейс с ИИ вернул парализованному пациенту с БАС речь и работу

Researchers at the University of California, Davis created a system based on a neurointerface and AI that restored speech and the ability to work fully to a paralyzed patient with amyotrophic lateral sclerosis. This is the first documented case where such technology has moved beyond clinical settings into real professional life.

What is ALS and what's the problem

Amyotrophic lateral sclerosis is a neurodegenerative disease in which motor neurons fail. Patients gradually lose control of their muscles: first mobility, then speech. Yet their intellect and consciousness remain intact. This is precisely what makes the disease especially devastating: a person understands everything but cannot speak, write, or gesture in response. Existing assistive technologies—eye trackers and special keyboards—are extremely slow. An average eye tracker user types around 10–15 words per minute. At this pace, neither normal conversation nor professional work is possible.

How the neurointerface works

The team of scientists implanted an electrode array in the patient's motor cortex, which records neural activity at the moment when the person mentally "pronounces" a word. Signals are transmitted in real time to an external computer, where a deep learning algorithm decodes the intention and synthesizes speech—in the patient's own voice, recorded before the illness.

Key characteristics of the system:

• Up to 62 words per minute—roughly half the pace of normal conversation
• Intent recognition accuracy—over 97%
• Speech synthesis latency—less than one second
• Neural array of 256 recording channels

Transition from lab to real life

The principal distinction of this work from previous research is not results in a controlled environment, but everyday use. The patient has returned to work full-time: attending meetings, communicating with colleagues, and performing professional tasks. The system operates for hours daily, not brief clinical sessions. This is critically important. Many neurointerfaces lose accuracy as the brain adapts to the implant. Long-term stable operation in real-world conditions is more compelling evidence of technology maturity than any laboratory metrics.

"This is not just a medical device—it's a return to professional identity and independence," noted researchers from UC

Davis.

What's next

The team is working on a wireless version of the system: the wires currently connecting the implant to the computer limit freedom of movement. A wireless variant will allow the device to be used outside a home environment. In parallel, negotiations are underway for expanded clinical trials—currently the system is available only within narrow research protocols.

What this means

ALS is diagnosed in thousands of people annually. Strokes, spinal cord injuries, and other neurological disorders deprive millions of people of the ability to speak or move. The success of this project is a signal that AI-neurointerfaces have matured for the transition from experimental medicine to clinical practice. The question is no longer whether this works technically, but how soon such systems will become available to a wider range of patients.