Digital bridge from brain to spine enables a paralysed man to stand and walk

On 24 May 2023 a team led by researchers at the NeuroX Institute at EPFL published a high profile article in Nature describing a brain to spinal cord interface that enabled a single participant with chronic tetraplegia to stand and walk in community settings. The work combines fully implanted cortical recording arrays, epidural stimulation of the lumbosacral spinal cord, a real time decoding pipeline and a stimulation controller. European Innovation Council funding featured as a direct support line for projects that contributed to the technology.

The participant and headline outcomes

The participant, Gert-Jan Oskam, is a 40 year old man who sustained a spinal cord injury in 2011 in a cycling accident and lived with chronic tetraplegia. Following neurosurgical implantation and repeated training sessions with the system the team reports that he regained the ability to stand and walk naturally with assistive devices, navigate stairs and variable terrain, and perform community activities. Notably the researchers observed that after roughly a year of use and training Oskam could walk overground with crutches even when the brain–spine interface was switched off. The authors interpret this as evidence that the intervention supported activity dependent neurorehabilitation and neurological recovery.

Key components and implants

Decoding movement intent and closed loop control

The decoding pipeline extracts temporal, spectral and spatial features from ECoG signals linked to attempted leg movements. The core algorithm is described as a recursive exponentially weighted Markov switching multi linear model. That is a hybrid approach that combines a gating classifier for state estimation and several independent regression modules dedicated to predicting direction and amplitude for specific joint movements. The gating classifier identifies which joint or resting state is intended. The expert regressions then estimate intended kinematic outputs. Model coefficients and transition probabilities were updated online every 15 seconds using recursive partial least squares to adapt to ongoing changes in signals. Outputs were constrained to lie within pre established functional ranges before being mapped to stimulation programs for the spinal implant.

Experimental evidence and quantitative measures

The team reported multiple objective measures. Electromyography and a 3D motion capture system quantified muscle activity and joint kinematics during treadmill and overground walking. Principal component analyses compared gait cycles recorded with stimulation alone to gait cycles produced with the brain–spine interface in place. The BSI produced gait features that were closer to those of healthy individuals than stimulation alone. The study used standardized clinical tests such as six minute walk and ten metre walk tests and blinded physiotherapist ratings using validated scoring sheets. The authors also examined decoder stability across sessions, ECoG spectrogram evolution around hip flexion attempts, and long term signal power during resting state showing a modest decline of recorded signal power at about minus 0.03 dB per day over nearly a year.

Crucially the case report highlights that after more than 40 training sessions the participant showed neurological improvement that persisted when stimulation was turned off. In the authors view these results support the hypothesis that the closed loop use of the BSI promoted activity dependent plasticity of spared spinal and corticospinal circuits. The evidence remains a single case so causal claims about general recovery must be cautious.

Safety, complications and trial governance

The study reports a notable adverse event. A Staphylococcus aureus subcutaneous infection at the site of one cortical implant led the principal investigator to explant that device 167 days after implantation. The second cortical implant showed no sign of infection and remained functional. After antibiotics and recovery the participant continued neurorehabilitation and home use. The team later implanted a new cortical device on 9 March 2023.

Funding, partnerships, intellectual property and industry ties

The work assembled an extensive funding and partnership network. European Innovation Council instruments explicitly supported projects connected to the research. Two EIC labelled projects mentioned are the Pathfinder NEMO BMI project and the EIC Transition ReverseParalysis project. The paper lists numerous funders including Swiss foundations, philanthropic awards, industrial collaborators, national science agencies and EU grants.

The authors declare a set of patents related to the technologies and several commercial relationships. Key disclosures include patents with European patent numbers, consulting roles for ONWARD Medical by senior investigators, employment of a coauthor at ONWARD and minority shareholdings in ONWARD by some senior researchers. Medtronic and ONWARD are noted as industrial collaborators. These ties are material when assessing motivations for translation and eventual commercial pathways.

Limitations and open questions

This is a single participant study and the participant had incomplete injury with some spared pathways. That clinical context makes generalisation to people with complete spinal cord transections uncertain. Rehabilitation was intensive and included more than 40 sessions which complicates attributions about what portion of recovery is due to stimulation driven practice versus spontaneous or therapy related improvements. The explantation due to infection highlights the risk profile of invasive implants. Long term device reliability, market readiness, manufacturing scale, cost, reimbursement and the regulatory pathway for broad clinical deployment remain unresolved. The paper documents encouraging early engineering and physiological findings but further controlled and multi participant trials are required to establish efficacy, safety and generalisability.

Why the EIC funding routes are relevant and what they cover

The European Innovation Council funds early stage deep tech through instruments that match different translational stages. The EIC Pathfinder targets high risk research at low technology readiness levels aiming to create novel scientific and technological building blocks. EIC Transition funding aims to mature and validate technology towards higher TRLs and improve market readiness. The projects named in the paper illustrate how EU support can be used to link foundational lab research, translational engineering and early clinical demonstrations. EIC funding can help move technologies from proof of concept towards validation in relevant environments but does not replace the need for formal regulatory approvals, larger clinical trials and reimbursement planning.

Data, code and publication

The Nature article is open access under Creative Commons Attribution 4.0. The authors made datasets and supporting MATLAB scripts available through Zenodo with the DOI referenced in the paper. The study also includes detailed extended data figures, tables, reporting summaries and supplementary videos that document device design, decoding performance, rehabilitation protocols and home use.

Implications and next steps

The study represents a convergence of neuroengineering, surgical practice and adaptive algorithms that together produced a clinically meaningful outcome for one person years after injury. It reinforces the scientific plausibility that closed loop neuromodulation can both restore volitional control and facilitate neurorehabilitation in patients with spared pathways. Yet significant hurdles remain. Robust evaluation across more participants with varied injury profiles will be necessary to quantify effect sizes and risks. Infection control, implant longevity and device miniaturisation will be key engineering challenges. Finally the route to clinical adoption requires commercial scale up, regulatory approvals and health technology assessment. EU funding via Pathfinder and Transition has clearly played a role in advancing the work. Continued public funding and independent clinical validation will be essential to separate promising technology from premature clinical hype.

Selected references and resources

Primary article: Lorach H. et al. Walking naturally after spinal cord injury using a brain and spine interface. Nature 618, 126 133 (2023). DOI as reported in the Nature article. Data and code are available via Zenodo links provided in the publication. ClinicalTrials.gov identifiers: STIMO NCT02936453 and STIMO BSI NCT04632290. For background on EIC instruments consult EIC Pathfinder and EIC Transition programme documentation.