Paralyzed Man Feeds Himself and Pets His Dog After Double Neural Bypass

After years of complete tetraplegia, Keith Thomas regained usable arm strength, touch, and everyday hand tasks through a hybrid brain–body interface that also left lasting gains when the system was off.

Today’s Progress

Keith Thomas of Massapequa, New York, was left paralyzed from the chest down after a diving accident in 2020 that damaged his spinal cord. In March 2023 he became the first person to receive a “double neural bypass” at Northwell Health’s Feinstein Institutes for Medical Research on Long Island—a hybrid system that pairs an intracortical brain–computer interface with patterned stimulation of the spinal cord and cortex.

Surgeons, including neurosurgeon Dr. Ashesh Mehta, implanted five electrode arrays in regions of Thomas’s motor and sensory cortex that control the right hand and fingers. When he intends to move, the implants record neural activity. Machine-learning models decode those signals and drive electrical stimulation of his arm muscles and spinal cord so his own hand and arm carry out the action. Pressure sensors in a custom orthotic measure grasp and trigger stimulation in sensory cortex, restoring a form of touch.

Results from more than three years of the first-in-human trial appear in Nature Medicine. After about 35 weeks of training with the system, strength rose 86 percent in his right arm and 62 percent in his left. Range of motion improved enough that he can scratch his nose, wipe his mouth, lift a cup, and feed himself—tasks that were impossible after the injury. He has also reported petting his dog, Bow, and feeling fur on a previously insensate wrist. Researchers say decoding accuracy for intended movements reached about 85 percent.

Importantly, some gains endured after the electronics were turned off for months. Study co-author Chad Bouton, who leads the work at Feinstein, told New Scientist that maintained recovery after long shutdowns is “unheard of” in this setting. Bouton has described the approach as not only bypassing the injury but helping rewire remaining pathways.

Good News Network, ABC7 New York, Popular Science, Smithsonian Magazine, and STAT have covered the peer-reviewed update and Thomas’s functional milestones. Outside full lab sessions he has self-reported independent face wiping and scratching and improved wrist sensitivity on sensory testing.

Why This Matters

For people with high-level complete spinal cord injury, loss of hand and arm use removes feeding, hygiene, and simple touch with family and pets. Restoring even partial voluntary movement and sensation can cut dependence on caregivers and restore moments of ordinary connection. Thomas’s case shows both immediate assistive control—thoughts translated into grasp and lift—and a therapeutic signal: strength and sensation that persist when the bypass is offline, consistent with activity-dependent plasticity below the injury.

The benefit is concrete and patient-level, not a corporate claim or a distant target. It does not cure paralysis or restore full independence for a population. It does demonstrate that a bidirectional brain–body interface plus spinal neuromodulation can deliver usable daily function and durable physiological change in at least one person with complete tetraplegia.

Evidence and Context

Evidence stage is demonstrated for this participant: multi-year clinical data, quantitative strength and sensory measures, and a Nature Medicine report of the double neural bypass design—intracortical recording and stimulation, reinforcement-learning grasp control, and transcutaneous spinal cord stimulation aimed at lasting recovery.

Limits are clear. Thomas is the first enrollee; results may not generalize. Full bidirectional control still depends on lab hardware and training sessions. Touch is restored as patterned cortical stimulation, not identical to pre-injury sensation. Surgical implantation carries risk. Long-term safety of chronic arrays, scalability, cost, and access remain open. No large controlled trial has yet confirmed the durability finding across many patients.

Trade-offs include invasive brain surgery versus noninvasive alternatives that typically offer less precise control. The team’s claim of persistent gains after months offline is striking and needs independent replication.

What Made This Possible

The work builds on years of brain–computer interface research, spinal neuromodulation, and AI decoding at Feinstein’s Institute of Bioelectronic Medicine under Bouton, with clinical implantation by Mehta and colleagues at Northwell. Thomas enrolled soon after injury into a multi-year trial designed for both assistive and therapeutic ends. Public reporting and the Nature Medicine paper made the methods and outcomes checkable.

Progress Toward Global Goals

The outcome aligns with SDG 3: Good Health and Well-being, specifically improved function and reduced disability burden for a person with severe neurological injury through a measured clinical intervention. It does not by itself expand population-level access to care. No UDHR article is claimed beyond the general dignity interest in health research; forced rights framing would overreach.

Building on This Success

The following possibilities were generated with the assistance of AI to explore how this progress might be improved, expanded, or adapted. They are ideas for further investigation, not established findings or recommendations from the people featured in the original reporting.

If durable gains after system-off periods hold up, clinics might test whether shorter “booster” stimulation schedules could maintain function with less hardware time. Home-use wearable stimulators paired with less invasive recording could be studied if safety and decoding quality allow. Cross-site trials with standardized strength, sensory, and activities-of-daily-living endpoints would show whether Thomas’s trajectory is exceptional or teachable.

How did a double neural bypass restore lasting arm movement and touch for Keith Thomas after complete tetraplegia?

The double neural bypass links intention, muscle activation, and sensation across the break in Thomas’s spinal cord. Motor-cortex arrays capture movement intent; decoders and spinal/muscle stimulation execute grasp and lift with his own limb. Sensory-cortex stimulation driven by fingertip pressure sensors supplies touch feedback, closing the loop so he can modulate force. Repeated closed-loop practice appears to have driven neuroplastic changes—measurable biceps strength and wrist sensitivity—that remained after months without the device. That combination of real-time assistive control and offline therapeutic carryover is what turned lab sessions into feeding himself and feeling his dog again.

Through a clinical-trial lens, the measurable test is already partly in place: bilateral arm dynamometry, monofilament sensory scores, and documented ADLs with the system on and off. Through a translation lens, the next obstacle is moving from a single highly supported participant to multi-center protocols with predefined durability criteria. Through a daily-life lens, the intervention only fully succeeds if simplified hardware lets users practice outside the lab without losing the plasticity signal.

Three Promising Next Steps

1. Multi-participant durability trial — Feinstein and partner SCI centers enroll additional complete tetraplegia volunteers on the same DNB protocol; primary endpoint is arm strength and wrist sensation retained ≥3 months after device-off periods, compared with baseline.
2. Simplified home stimulation arm — Engineers and clinicians test whether noninvasive or minimally supervised spinal/muscle patches, guided by clinic-trained intent models, preserve a fraction of lab gains between visits; success metric is weekly self-fed meals or equivalent ADLs logged for 12 weeks.
3. Open methods package — The team publishes decoding, stimulation timing, and rehab schedules in enough detail for independent labs to attempt replication; measurable test is a second site reproducing ≥50 percent of the strength gains under a pre-registered protocol.

What Readers Can Watch

• Follow-on Nature Medicine or clinical-trial registry updates listing additional DNB participants and device-off durability data.
• Northwell Health / Feinstein Institutes releases on home-use or less invasive versions of the bypass.
• Independent replication attempts at other BCI or spinal-stimulation centers using comparable strength and sensory endpoints.
• Regulatory or ethics board milestones if the system moves beyond single-patient experimental status.

What Readers Can Do

Readers seeking established SCI rehabilitation or research opportunities can consult clinical trial registries (for example, ClinicalTrials.gov searches for brain-computer interface or spinal cord stimulation and tetraplegia) and national spinal cord injury associations for verified trial listings and support services—not unvetted crowdfunding or unproven clinics.