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

Nicholas S. Card, Tyler Singer-Clark, Hamza Peracha, Carrina Iacobacci, Xianda Hou, Maitreyee Wairagkar, Zachery Fogg, Elena C Offenberg, Leigh R. Hochberg, David M. Brandman, Sergey D. StaviskybioRxiv 2025 · DOI

TL;DR

This paper showcases the first long-term, independent use of a multimodal intracortical brain-computer interface (BCI) for both speech and cursor control by an individual with severe paralysis due to ALS. Over nearly two years, the participant used the BCI at home for over 3,800 hours, enabling naturalistic communication, full-time employment, and independent computer use, marking a significant step towards practical assistive technology.

What Problem Does This Paper Address?

For individuals with severe motor impairments, such as those caused by ALS or brainstem strokes, the ability to communicate and interact with digital devices is profoundly impacted. While existing assistive technologies exist, they are often slow, fatiguing, or require constant assistance. Brain-computer interfaces (BCIs) offer a promising alternative, but critical barriers to their practical adoption have remained: the need for independent, at-home operation without researcher intervention, and reliable, long-term performance that maintains accuracy. This paper directly addresses these challenges by demonstrating a BCI system that overcomes these limitations.

Methods

The core of this research is a multimodal BCI system that integrates novel decoding architectures for both speech and cursor control.

  • Speech Decoding: The system employs a transformer-based brain-to-text decoder, an advancement over previous recurrent neural network (RNN) models. This decoder translates neural activity from the speech motor cortex into intended words. Key innovations include continuous background recalibration to adapt to neural signal variability and a large 125,000-word vocabulary.
  • Cursor Control: A separate decoder, also leveraging advanced architectures (initially linear, later RNN-based), translates neural signals into cursor movements and click gestures. This allows for direct manipulation of a computer's interface.
  • Multimodal Integration: The system seamlessly integrates speech and cursor control, allowing the participant to use the speech BCI as a keyboard input and the cursor BCI as a mouse input.
  • Independent Operation Features: To facilitate at-home use, the system includes features like automated startup/shutdown, gaze-based control for interface navigation, and a user-friendly interface for sentence rating and correction.

Datasets

The primary dataset consists of neural activity and speech decoding transcripts recorded from the single participant (T15) over approximately two years of independent use. This dataset is unique in its scale and duration of personal, at-home BCI usage. Formal performance benchmarks were conducted using prompted word copy tasks.

Key Insights and Ideas

  • Independent Home Use is Achievable: The most significant insight is the demonstration that a complex intracortical BCI can be reliably used independently in a home environment for extended periods, without constant researcher oversight.
  • Speech Motor Cortex Versatility: The study reinforces that the speech motor cortex is a rich source of neural signals not only for speech but also for controlling digital cursors, highlighting its potential as a central hub for multimodal BCI control.
  • Transformer Decoder Superiority: The shift to a transformer-based decoder for speech significantly improved accuracy and reduced the need for frequent recalibration, showcasing the power of newer neural network architectures in BCI applications.
  • Long-Term Neural Stability: The neural signals remained remarkably stable over 19 months, a critical factor for sustained BCI performance and a major hurdle in previous BCI research.

Key Results

  • Communication Volume: The participant communicated over 183,000 sentences, totaling 1,960,163 words, at an average rate of 56 words per minute.
  • Accuracy: 92.3% of self-rated sentences were decoded as at least "mostly correct." In formal tests, speech decoding achieved over 99% word accuracy with a 125,000-word vocabulary.
  • Usage Duration: The participant accumulated over 3,800 hours of independent BCI use over nearly two years, using the system on average for over 9.5 hours per day after independent use was enabled.
  • Employment: The participant was able to sustain full-time employment using the BCI.
  • Cursor Control: The cursor BCI provided efficient control, with an RNN-based decoder significantly reducing calibration time.

Limitations and Future Work

The authors acknowledge several limitations:

  • Single Participant: The study involved only one participant, limiting generalizability to other individuals, electrode types, or neurological conditions.
  • Wired Connections & Bulkiness: The system relied on percutaneous wired connections and was bulky, limiting portability and use outside the home.
  • Setup Time: While care partners could manage setup, it still required approximately 20 minutes daily.
  • Conversation vs. Prompted Speech: Speech decoding accuracy during independent conversation was not as high as in structured research sessions.

Future work should focus on developing wireless or fully implantable systems, minimizing setup time, expanding access to diverse user groups, and further improving conversational speech decoding to match the consistency and accuracy of natural speech.

Summary updated

Abstract

Abstract Brain-computer interfaces (BCIs) can provide naturalistic communication and digital access to people with severe paralysis by decoding neural activity associated with attempted speech and movement. Recent work has demonstrated highly accurate intracortical BCIs for speech and cursor control, but two critical capabilities needed for practical viability were unmet: independent at-home operation without researcher assistance, and reliable long-term performance supporting accurate speech and cursor decoding. Here, we demonstrate the independent and near-daily use of a multimodal BCI with novel brain-to-text speech and computer cursor decoders by a man with paralysis and severe dysarthria due to amyotrophic lateral sclerosis (ALS). Over nearly two years, the participant used the BCI for more than 3,800 cumulative hours to maintain rich interpersonal communication with his family and friends, independently control his personal computer, and sustain full-time employment – despite being paralyzed. He communicated 183,060 sentences – totaling 1,960,163 words – at an average rate of 56.1 words per minute. He labeled 92.3% of sentences as being decoded at least mostly correctly. In formal quantifications of performance where he was asked to say words presented on a screen, attempted speech was consistently decoded with over 99% word accuracy (125,000 word vocabulary). The participant also used the speech BCI as keyboard input and the cursor BCI as mouse input to control his personal computer, enabling him to send text messages, emails, and to browse the internet. These results demonstrate that intracortical BCIs have the potential to support independent use in the home, marking a critical step toward practical assistive technology for people with severe motor impairment.
BibTeX
@misc{card2025long,
  author = {Nicholas S. Card and Tyler Singer-Clark and Hamza Peracha and Carrina Iacobacci and Xianda Hou and Maitreyee Wairagkar and Zachery Fogg and Elena C Offenberg and Leigh R. Hochberg and David M. Brandman and Sergey D. Stavisky},
  title = {{Long-term independent use of an intracortical brain-computer interface for speech and cursor control}},
  year = {2025},
  doi = {10.1101/2025.06.26.661591},
  howpublished = {bioRxiv},
}
Figure