A patient whose body below the neck was completely paralyzed in a diving accident has been able to move a hand again and even feel the touch when holding a family member's hand and stroking a pet's fur, thanks to electrodes implanted in the brain. A computer using artificial intelligence (AI) decoded neural signals and acted as an intermediary between the brain and the body. In particular, the rehabilitation gains held even when the computer was turned off, indicating partial recovery of the patient's neural network. The severed route caused by the accident was linked by a detour.
Northwell Health, a nonprofit medical organization in New York state, said on the 17th (local time) that Professor Chas Bouton's team at the Feinstein Institutes for Medical Research succeeded with brain-computer interface (BCI) technology in enabling a patient who was quadriplegic after a spinal cord injury to move a hand again and feel sensation. The study was published the same day as the cover paper in the international journal Nature Medicine.
BCI is a technology that detects neural signals originating in the brain and enables the exchange of information with external devices such as computers. Thanks to it, the motor signals generated when a paralyzed patient tries to move can be delivered directly to limb muscles, and thoughts can be converted into computer-generated speech or text, or into the movements of a robot.
◇ Building a neural bypass to restore movement and touch
Keith Thomas, a 48-year-old man, injured his cervical vertebrae while diving in a swimming pool in 2020, damaging the motor and sensory nerves of the spinal cord passing through them. The patient was completely paralyzed below the neck. He could not lift his hand and could not feel the touch of objects on his hand. The research team restored motor function and touch with a "double neural bypass" using BCI technology.
First, five chips with arrays of microelectrodes were implanted on the surface of the patient's brain. The electrodes captured the neural signals that arose in the brain when the patient tried to move a hand and transmitted them to a computer. Artificial intelligence (AI) decoded these signals and delivered corresponding electrical stimulation through patches attached to the hand and arm. In effect, brain motor signals skipped the cervical spinal cord and were delivered directly to the hand and arm.
AI read the intent to move from brain signals with 84.6% accuracy. Thanks to the BCI system, the patient could again raise an arm overhead, grip a cup, and eat independently. The research team attached an assistive device with force sensors to the patient's hand to help bend the fingers according to the computer's signals. A kind of artificial tendon assisted the hand movements. Over three years of clinical trials, muscle strength increased by 86% in the right arm and 62% in the left.
The restoration of touch began with mapping the brain's sensory signals. The team decoded where and what signals appeared in the sensory cortex when the patient imagined a situation where an object touched the hand. If signals identical to those appearing in the brain were delivered to the sensory cortex, the corresponding tactile sensation could be produced.
When an object touches the patient's hand, sensors in a device mounted on the wrist detect pressure. AI decodes the sensor signals, converts them into electrical signals, and sends them to the implanted brain chips. The brain then perceives the feel of the object touching the fingers. Because it can compare the motor signals sent from the brain with the touch sensed by the hand, more precise movements became possible when moving objects. In an egg-transfer test, the patient achieved an 87% success rate. AI micro-adjusted the gripping force dozens of times per second.
◇ Beyond assistance to therapy, even neural network reconfiguration
The research team expected that some rehabilitation gains would remain even if the BCI device were turned off. To find out, they planned to stop electrical stimulation for a month, but a fire in the laboratory building forced a three-month suspension of stimulation.
Surprisingly, during that time the patient maintained hand strength, sensation, and function. Even though the computer consolidation was cut off, when pressure was applied, the patient said there was a tingling sensation in the wrist. The team said recent assessments showed the improvements in movement and touch have still been maintained more than two years later.
Sergey Stavisky, a professor at the University of California, Davis, said, "If such improvements persisted even when the system was off, it can be seen as having done more than temporarily restoring function," and noted, "Neural networks may have reconfigured themselves through neural plasticity."
Neural plasticity is the body's ability to form new neural networks and reconfigure existing connections in response to learning, experience, or brain injury. This means the BCI system not only assisted movement and touch but also served as therapy that promoted long-term recovery of the nervous system. The research team also inferred that continuously delivering electrical stimulation not only to the hand and arm muscles but also to the damaged spinal cord helped reconfigure the neural network.
◇ Hopes for treating stroke patients as well
Professor Bouton said the results offered hope to about 15 million people with spinal cord injuries worldwide. More than half of spinal cord injury patients are quadriplegic. Contrary to expectations, paralyzed patients said they wanted to use their hands first rather than walk or regain bowel and bladder function.
In the past, BCI technology has been used to move limbs or operate robotic arms, but even grasping a single object required tremendous concentration from the patient. By contrast, the team said the current system greatly reduced that burden thanks to AI. It enabled control natural enough for the patient to perform delicate movements while conversing with others.
The same approach is expected to help rehabilitate stroke patients. Stroke affects one in four adults aged 25 or older worldwide. Among patients, 75% lose motor abilities in the arms and hands, making daily life difficult.
However, these results come from a single patient, and large-scale clinical trials are needed for commercialization. Work is also needed to automate the system so patients can use it easily without experts.
For reference, Professor Bouton's double neural bypass was selected by Time in 2024 as one of the Best Inventions. Last year, Time also inducted it into the Hall of Fame recognizing inventions that have had the greatest impact worldwide over the past 25 years.
References
Nature Medicine (2026), DOI: https://doi.org/10.1038/s41591-026-04498-0
Time (2024), https://time.com/collections/best-inventions-2024/7094710/northwell-health-double-neural-bypass/