Long-term independent use of an intracortical brain–computer interface for speech and cursor control

Journal: Nature Medicine (2026) · DOI 10.1038/s41591-026-04414-6 Authors: Nicholas S. Card and colleagues (BrainGate2 / UC Davis lineage) Fetch status: summary-derived — nature.com returns a 303 redirect to an IDP auth wall for both the article and the journal index. Figures below are triangulated from multiple independent search summaries of the abstract; the canonical URL is preserved for refetch via PMC/institutional mirror.

What it is

A landmark demonstration that an intracortical BCI can be used independently, at home, over the long term — the capability gap that has separated lab demonstrations from a usable assistive device. The participant is a man with amyotrophic lateral sclerosis (ALS) and severe dysarthria (loss of intelligible speech).

The system is multimodal: a brain-to-text speech decoder plus a computer-cursor decoder, both driven from the same intracortical microelectrode arrays.

Key quantitative findings (quoted figures)

• >3,800 hours of at-home use over nearly 2 years, with no researchers present.
• 183,060 sentences communicated — totalling 1,960,163 words (~1.96 million).
• Average rate of 56 words per minute.
• The participant labelled 92% of sentences as being decoded at least mostly correctly.
• Demonstrated the two capabilities required for practical viability: independent at-home operation (no researcher assistance) and reliable long-term performance supporting accurate speech and cursor decoding.

Why it matters

This is the strongest existing evidence that a high-bandwidth intracortical speech BCI works outside the lab, unsupervised, and durably — not as a one-session benchmark but as a daily communication tool sustained across ~2 years. It moves the speech-BCI frontier from "high accuracy in a controlled session" (e.g. the BIT 10% WER result, the Stanford inner-speech demonstration) to "real-world independent communication restoration" for a locked-in / dysarthric population. It is the clinical-utility complement to the Paradromics and Neuralink speech programmes now entering first-in-human trials.

Provenance note

Body content is summary-derived (Nature paywall/auth redirect). All numeric claims appear consistently across independent secondary summaries of the same Nature Medicine abstract. Recommend a refetch via PubMed Central or an institutional copy to capture the full author list and methods.