UC Davis Health said on 15 June that Casey Harrell, a 47-year-old participant in the BrainGate2 clinical trial, used an implanted speech brain-computer interface at home for more than 3,800 hours. The release said he communicated more than 183,000 sentences and close to 2mn words, with an average speed of 56 words per minute.
EurekAlert described the work as a peer-reviewed Nature Medicine publication showing that a person with severe ALS paralysis could use the system at home to communicate, work and operate digital platforms without researcher support. The Washington Post reported the same long-use figures, including 1,960,163 words over nearly two years.
That changes the evidentiary focus. Many brain-computer interface stories are demonstrations: a participant speaks a phrase, controls a cursor or reaches a high decoding accuracy in a laboratory setting. Demonstrations can be scientifically important, but clinical value depends on durability, reliability and whether the system fits into ordinary life. A device that works only when a research team is in the room is not the same as a communication tool.
The UC Davis system remains investigational. According to the university, Harrell received four microelectrode arrays implanted in the left precentral gyrus in 2023, recording from 256 cortical electrodes. In plain terms, the device records brain activity from a speech-related motor area while the user attempts to speak, then software decodes those signals into text or computer control.
The earlier New England Journal of Medicine paper on the same participant reported a rapidly calibrating speech neuroprosthesis that enabled stable decoding for an individual with ALS. UC Davis said at the time that the system translated brain signals into speech with up to 97% accuracy. The new report asks a different question: what happens after the initial accuracy result, when the participant lives with the system?