AI system
- System
-
Double neural bypass (Bidirectional Neural Bypass System): intracortical brain-computer interface with an LSTM decoder and deep reinforcement-learning agent, patterned spinal cord stimulation and intracortical microstimulation
- Model
-
Not assessed
- Release
-
Not assessed
- Application developer
-
Feinstein Institutes for Medical Research, Northwell Health (Institute of Bioelectronic Medicine, Chad Bouton)
- Role
-
Not assessed
- Configuration
-
Not assessed
- Inputs and outputs
-
Not assessed
- Autonomy
-
Not assessed
- Workflow
-
Chronic complete cervical spinal cord injury (C4 sensory/C5 motor) with movement intentions still readable from motor cortex. Five microelectrode arrays placed in motor and somatosensory cortex during a 15-hour mapped operation with the participant partly awake. An LSTM plus deep reinforcement-learning decoder paired with patterned spinal cord stimulation and intracortical microstimulation. Thirty-five weeks of supervised laboratory sessions inside an approved early-feasibility trial (NCT03680872).
- Human oversight
-
Not assessed
- Relation to AI
-
The decoder turns motor-cortex activity into intended hand movements. A reinforcement-learning agent regulates grasp force: figure 4c of the original paper reports 87% task success with the agent and 27% without it. The separate 86% and 62% arm-strength gains occurred during spinal stimulation and activity-based training before the hand neuroprosthesis. The full system combines AI with surgery, stimulation, an orthosis, and rehabilitation. The paper anonymizes its participant; Northwell identifies him as Keith Thomas.
- AI contribution
-
The decoder turns motor-cortex activity into intended hand movements. A reinforcement-learning agent regulates grasp force: figure 4c of the original paper reports 87% task success with the agent and 27% without it. The separate 86% and 62% arm-strength gains occurred during spinal stimulation and activity-based training before the hand neuroprosthesis. The full system combines AI with surgery, stimulation, an orthosis, and rehabilitation. The paper anonymizes its participant; Northwell identifies him as Keith Thomas.
Benefits
- Outcome identifiers
-
Not assessed
- Benefit
-
87% vs 27% successful grasp-force control with and without reinforcement learning
- Category
-
Improved Physical Health
- Status
-
Not assessed
- Beneficiary groups
-
People.
- Beneficiaries
-
Not assessed
- Reach
-
Not assessed
- Distribution and access
-
Not assessed
- Duration
-
Not assessed
- Expected?
-
Yes
- Costs and harms
-
Not assessed
Measurement
- Studies
-
Not assessed
- Study design
-
Not assessed
- Measurement
-
Not assessed
- Setting and sample
-
Not assessed
- Measures
-
87% vs 27% successful grasp-force control with and without reinforcement learning
- Comparison
-
Grasp-force control with the reinforcement-learning agent switched off, in the same participant
- Results
-
Reference: 27% successful force control without the reinforcement-learning agent. Result: 87% successful force control with the reinforcement-learning agent.
- Additional value from AI
-
Not assessed
- Limitations
-
Not assessed
- Denominators
-
Not assessed
- Materials for replication
-
Not assessed
Money
- Kind of value
-
Not assessed
- Amount
-
Not assessed
- Currency and prices
-
Not assessed
- Period and unit
-
Not assessed
- Realization
-
Not assessed
- Recipients and payers
-
Not assessed
- Gross or net
-
Not assessed
- Included costs
-
Not assessed
- Comparison
-
Not assessed
- Method
-
Not assessed
- Return and payback
-
Not assessed
People and places
- Participants
-
Developer: Feinstein Institutes for Medical Research, Northwell Health (Institute of Bioelectronic Medicine, Chad Bouton). Operator: Northwell Health, North Shore University Hospital.
- Jurisdictions
-
Benefit: Manhasset,
United States. Developer: United States. Operator: United States. - Ownership and control
-
Not assessed
- Dependencies
-
Not assessed
- Cross-border connections
-
Not assessed
- Geopolitical implications
-
Not assessed
Follow-up
- Conditions for replication
-
Decoder accuracy sustained over months (84.6% over five months) so intended movements reliably trigger stimulation. Neurosurgical, engineering and rehabilitation teams present throughout. The gains were measured in the laboratory. Follow-up with the system switched off to confirm that strength and sensation gains persist.
- Replication attempts
-
Not assessed
- Later findings
-
16 July 2026. The Feinstein Institutes' July 2026 release on the Nature Medicine paper reports that the strength and sensation gains were still present at a follow-up more than two years later.
- Availability
-
Not assessed
- Institutional decision
-
Not assessed
- Grounds
-
Not assessed
- Action and result
-
Not assessed
- Success criterion
-
Not assessed
Evidence
- Source numbers
-
AIBD-0118-source-0001–0006
- Citations
-
Numbered references below.
- Language
-
Not assessed
- Source relationships
-
Developer, Independent
- Original evidence
-
The study follows one participant across repeated sessions. Data and code require author access. The text and figure 4c caption disagree on whether the t-test was paired, and the investigators’ report of two-year persistence has no independent evaluation. [1] [2] [3] [4] [5] [6]
- Access and versions
-
Not assessed
- Supported findings
-
Not assessed
- Evidence links
-
Not assessed
- Evidence strength
-
Not assessed
- Missing evidence
-
Not assessed
Grades
- Assessed benefits
-
People.
- Depth
-
People: 3.
- Reasons
-
People: In the laboratory, Thomas used AI-decoded intent to control grasping, including self-feeding, alongside spinal stimulation and sensory feedback. This restored meaningful abilities in one person. The study does not isolate every component’s contribution.
- Observed or intended
-
Not assessed
- Rubric
-
Not assessed
- Assessor and date
-
Not assessed
- Alternative assessments
-
Not assessed
- Fill
-
Independently reported
Entry details
- Kind
-
Not assessed
- Related entries
-
Not assessed
- Field
-
Healthcare (neuroprosthetics and spinal cord injury). Complete cervical spinal cord injury (C4 sensory/C5 motor tetraplegia).
- Purpose
-
ClinicalTrials.gov NCT03680872 describes an early feasibility trial of the Bidirectional Neural Bypass System. It is meant to restore movement and sensation in the hand and wrist of up to seven people with tetraplegia. This benefit is that purpose.
- Stage
-
Not assessed
- Version
-
SHA-256: 0e65c8e90f1a…
References
- Other identifiers
-
oecd_aim: 2023-07-28-1c4b.
- Category mappings
-
Not assessed
- Brief
-
Not assessed
- Revisions
-
6 revisions. Latest: 24 October 2026.
Sources
-
Peer-reviewed study Developer
Chandrasekaran S, Wandelt SK, Jangam A, et al. (2026). A neuroprosthesis for restoring hand movement and sensation in a person with complete tetraplegia. Nature Medicine 32(7):2591–2601. Figure 4c reports the comparison with and without reinforcement learning; figure 2 concerns earlier spinal stimulation and training.
Original maker-authored clinical paper. Figure 4c supports the AI-specific grasp-control comparison; figure 2 concerns spinal stimulation and training. Peer review is not an independent replication.
87% success with RL and 27% success without RL
-
News report Independent author
Barmash J. / Patch Massapequa (2023). Inspirational Quadraplegic Massapequa Man In 'Amazing' Clinical Trial. Patch.
Original 2023 interviews with Thomas and investigator Chandrasekaran add firsthand experience. The patient’s 110% statement is not the paper’s later clinical metric.
Mobility in my right arm has increased by 110% compared to my baseline that I started with
-
Press release Developer
Feinstein Institutes for Medical Research / Northwell Health (August 2024). Double Neural Bypass Update: Bites Transcript. Institutional interview transcript, pp. 1–2.
Original institutional interview transcript for the 2024 follow-up. Read from the media kit linked by Northwell’s 2026 page.
-
Press release Developer
Libassi M. / Feinstein Institutes for Medical Research (2023). Feinstein Institutes researchers first to use double neural bypass to restore feeling, movement in man living with quadriplegia. Northwell Health news release.
Original July 28, 2023 release PDF, recovered from the institutional media kit. Establishes the named participant, accident and surgery dates, and original announcement.
Paralyzed from the chest down, Keith Thomas, 45, of Massapequa, NY, is the first human to use the technology.
-
Press release Developer
Libassi M. / Feinstein Institutes for Medical Research (2026). Feinstein Institutes technology that restores movement and sensation after paralysis featured in Nature Medicine. Northwell Health news release.
Original institutional 2026 report identifies the anonymized paper participant and adds the investigator’s report of persistence beyond two years.
Over 35 weeks of this intervention, Mr. Thomas achieved statistically significant increases of 86% in right arm strength and 62% in left arm strength.
-
Dataset Developer
Bouton C / Northwell Health. Restoring Motor and Sensory Hand Function in Tetraplegia Using a Neural Bypass System. ClinicalTrials.gov NCT03680872, first posted September 21, 2018, updated March 11, 2026. Sponsor-submitted registration; not a regulatory finding of benefit.
Sponsor-investigator registration accessed through the ClinicalTrials.gov API. Establishes protocol and planned enrollment, not regulatory confirmation or independent benefit evidence.
an investigational device called the Bidirectional Neural Bypass System can lead to the restoration of movement and sensation in the hand and wrist of up to seven individuals with tetraplegia