AI brain implant restores movement and touch in quadriplegic patient long-term

What happened?

A team from the Feinstein Institutes for Medical Research (Northwell Health) has enabled Keith Thomas, a man with complete quadriplegia since 2020, to feed himself, drink from a cup with his own hand, scratch his nose, and feel his dog's touch. The results, published on July 16, 2026, in Nature Medicine, show that the benefits persist more than two years after the active phase of the experiment ended. Keith Thomas suffered a spinal cord injury in a diving accident in 2020 that left him without movement or sensation below the neck. After a 15-hour surgery to implant five microelectrode arrays, the system began working. According to Ana Maria Constantin's report in TheNextWeb, the patient can now perform everyday tasks such as drinking from a cup or scratching his nose without continuous assistance.

How does the double neural bypass work?

The system, called double neural bypass, combines three technologies: a brain implant with five microelectrode arrays (brain-computer interface), an AI that decodes brain signals with 84.6% accuracy for five months without retraining, and a dual electrical stimulation system that activates forearm muscles and the somatosensory cortex to restore touch. A 3D-printed splint with sensors detects contact and translates the information into brain stimulation signals. The AI, trained on the patient's movement attempts, maintains its accuracy even after months without adjustments, reducing the need for frequent recalibrations. Electrical stimulation not only activates muscles but also sends tactile signals to the brain, creating a feedback loop that allows the patient to feel what they touch.

Why is this important?

This is the first time a brain implant not only assists movement while active but also appears to rewire the nervous system durably. This suggests that repetitive stimulation can induce neural plasticity, opening the door to therapies that do not require permanent implant use. According to the article from WWWhat's new, researchers observed that even when the system is turned off, the patient retains some movement and sensitivity, indicating that the brain and spinal cord have begun to reconnect. This finding could shift the focus of neuroprosthetics from temporary assistive devices to tools that promote long-term biological recovery. Compared to previous studies, such as BrainGate which allowed moving a cursor with the mind, this advance integrates movement and sensation into a single system, achieving complex real-world actions.

Implications for the future

If these results are replicated in more patients, we could be witnessing a paradigm shift in the treatment of spinal cord injuries. The combination of BCI with AI and electrical stimulation could offer long-term solutions, reducing dependence on external devices. However, the 15-hour surgery and system complexity limit immediate large-scale application. Additionally, the current cost of the procedure is high, though it could decrease with component miniaturization and less invasive implants. In the future, we might see systems that do not require open surgery but are inserted via catheters, as already explored in other fields. It will also be crucial to assess long-term effects, as the patient has had the implant for over two years with no adverse effects beyond typical surgical ones.

“The double neural bypass represents a significant advance in neuroprosthetics, demonstrating that brain plasticity can be induced and maintained even years after injury.” – The Vortiq

What readers should know

• The study involves a single patient, so trials with more participants are needed to confirm the results. Researchers already plan to recruit more volunteers in the coming months.
• The lasting effects are not fully explained; it is speculated that repetitive stimulation promotes neural reconnection, but the exact mechanism has not yet been demonstrated.
• The technology is still invasive and expensive, but could evolve toward less invasive implants in the future, such as electrodes on the brain surface or even external devices.
• No information is available on potential long-term adverse effects beyond those reported in the study, such as infections or device failures. The patient has been continuously monitored with no serious issues detected.
• This advance adds to other recent developments in brain-computer interfaces, such as those from Neuralink or Synchron, but stands out for its focus on restoring sensation and voluntary movement, and for the persistence of benefits.