Paralyzed man types 22 words per minute using a brain implant and imagined finger movements

Before his ALS progressed, he could type as fast as anyone. Now he cannot move his fingers at all. But he can think about moving them - and that, it turns out, is enough.

In a study published in Nature Neuroscience, investigators from Mass General Brigham Neuroscience Institute and Brown University describe a brain-computer interface (BCI) that translates attempted finger movements into typed text on a virtual QWERTY keyboard. Two participants in the BrainGate clinical trial - one with advanced ALS and one with a cervical spinal cord injury - used the system to communicate rapidly and accurately from their own homes.

How the system works

The neuroprosthesis begins with microelectrode arrays implanted in the motor cortex, the brain region that plans and initiates voluntary movement. A standard QWERTY keyboard layout is displayed in front of the participant, with each letter mapped to specific fingers and finger positions - up, down, or curled. When the participant attempts to press a key, the electrodes detect the corresponding neural activity and send it to a computer system that decodes the intended letter.

The raw neural output is then processed through a predictive language model - similar in concept to the autocomplete on a smartphone keyboard - that corrects errors and ensures coherent output. The combination of motor decoding and language prediction is what enables both speed and accuracy.

Calibration required as few as 30 typed sentences, after which participants could use the system for free communication.

Performance numbers

One participant reached a top typing speed of 110 characters per minute, equivalent to 22 words per minute, with a word error rate of just 1.6%. That accuracy is comparable to able-bodied typing. Both participants used the device at home, demonstrating that the system can function outside a controlled laboratory environment - a critical step toward practical, everyday use.

For context, existing communication devices for people with severe paralysis - particularly eye-gaze systems that spell words one letter at a time by tracking eye movements - typically operate at far slower speeds. Many patients find them frustrating enough to abandon altogether. A system that approaches normal typing speed addresses a genuine and deeply felt need.

More than communication

The finger-movement decoding has implications beyond typing. First author Justin Jude, a postdoctoral researcher at Mass General Brigham, noted that the same neural signals that encode individual finger movements could eventually be used to restore complex hand function - reaching, grasping, manipulating objects. The QWERTY keyboard serves as both a practical communication tool and a proof of concept for more ambitious motor restoration.

The researchers also see room to improve the typing interface itself. A stenographic keyboard, which uses key combinations rather than individual letters, could potentially push speeds well beyond 22 words per minute. Personalized keyboard layouts optimized for each user's neural patterns are another possibility.

Where the technology stands

The BrainGate iBCI remains an investigational device, limited by federal law to research use. The implanted microelectrode arrays require neurosurgery, and the external hardware that processes the neural signals is not yet miniaturized for fully portable use. The system also requires periodic recalibration, though the initial calibration burden is modest.

With only two participants, the study demonstrates feasibility and performance in individual cases but cannot establish reliability across the broader population of people with paralysis. Different etiologies - ALS, stroke, spinal cord injury, brainstem stroke - affect the motor cortex differently, and what works for one patient may not transfer directly to another.

The BrainGate consortium, which has been developing and testing implantable BCIs since 2004, positions this work as advancing the frontier that commercial companies - Neuralink, Synchron, Precision Neuro, and others - will eventually turn into consumer medical devices. Senior author Leigh Hochberg emphasized the role of academic research in establishing what is possible so that industry can focus on engineering the final product.

For the two participants in this study, the technology already delivers something that existing alternatives cannot: fast, accurate, intuitive communication using the same mental model - fingers on a keyboard - that they used before paralysis changed everything.