Next-generation neural interface: ALS patient transmits nearly 2 million words through the power of thought
The world of neurotechnology has achieved another breakthrough. Researchers from the University of California, Davis, presented the results of an unprecedented experiment: a patient with amyotrophic lateral sclerosis (ALS) named Casey Harrell successfully used a speech neurointerface at home for 19 months. During this time, he transmitted 183,060 sentences — a total of 1,960,163 words. The average typing speed by thought reached 56 words per minute. This is not a laboratory demonstration, but full-fledged everyday functioning.
Over 3,800 hours of communication without researcher assistance
The key difference of this study from previous ones is complete autonomy. The system operated without the presence of scientists. Harrell used the neurointerface for over 3,800 hours to communicate with family, friends, colleagues, and doctors. He sent messages, participated in video calls, worked on the internet, and maintained full employment despite complete paralysis.
According to the patient's own estimates, 92% of sentences were decoded as at least "mostly correct." In formal tests, accuracy exceeded 99% with a vocabulary of 125,000 English words, and the peak rate reached 99.2%. When Harrell's assistants were allowed to independently connect the equipment, the average daily usage time increased from 3.7 to 9.5 hours — the system was active on 444 out of 653 days after implantation.
How does this neurointerface work?
In 2023, Harrell had four microelectrode arrays implanted in the left precentral gyrus — the brain region responsible for speech coordination. The system reads signals from 256 cortical electrodes when the patient attempts to speak. Every 80 milliseconds, the algorithm converts neural activity into phoneme probabilities, and a language model selects the most likely sequence of words from a vocabulary of 125,000 units. The text is displayed on the screen in real time, and after the phrase is completed, the system voices it with a synthesized voice mimicking Harrell's voice before the illness.
Interestingly, the patient used the same cursor decoder to control the computer. Previously, it was believed that speech and movements required different brain areas, but the team proved the opposite: both modes can be implemented through signals from the speech motor cortex.
From laboratory to real life
Earlier in 2024, the team demonstrated high accuracy in controlled sessions: within 30 minutes, the system achieved 99.6% accuracy with a vocabulary of 50 words. The new work shifts the focus to long-term home use. As study co-author Sergey Stavisky noted, 3,800 hours of brain activity recording became the largest individual dataset with resolution at the level of individual neurons.
However, the authors emphasize that this is just one clinical case. It is not yet known how applicable the results are to other patients, implantation sites, or types of electrodes. The system remains experimental, uses wired connections, and requires daily setup by trained assistants.
My expert opinion: This experiment is not just a step, but a leap into the era of practical neurointerfaces. While previously we saw impressive but isolated laboratory successes (like Neuralink with playing World of Warcraft), here we observe full integration of the technology into everyday life. A speed of 56 words per minute is already comparable to normal speech. The main challenge now is miniaturization and wireless data transmission to make such systems accessible for widespread use.