A groundbreaking study from Chalmers University of Technology in Sweden has unveiled a revolutionary nerve interface that decodes phantom limb movements directly from peripheral nerves, paving the way for prosthetic legs that respond with natural, intuitive control and sensation. Published in Nature Communications on February 8, 2026, the research demonstrates how signals from the sciatic nerve in above-knee amputees can be interpreted with unprecedented accuracy, marking a significant leap in bionic limb technology.

This innovation addresses a critical gap in prosthetic design: the lack of direct neural communication. Traditional prostheses rely on surface electromyography (EMG) from residual muscles, which often fails to capture fine-grained intentions for complex movements like toe wiggling or ankle dorsiflexion. By implanting ultrathin electrodes directly into the nerve, researchers captured multi-unit activity tied to volitional phantom limb actions, enabling biomimetic control that feels more like the original limb.

🧠 The Science Behind Phantom Limb Decoding

Phantom limb pain and movement sensations occur because nerves severed during amputation continue firing signals as if the limb is still there. The study targeted the tibial branch of the sciatic nerve—the primary conduit for lower leg motor commands—in two transfemoral (above-knee) amputees. Four transversal intrafascicular multichannel electrodes (TIMEs), each with 14 active sites (56 channels total), were surgically implanted. These flexible, hair-thin probes, developed at the University of Freiburg in Germany, penetrate nerve fascicles to record high-resolution signals without damaging tissue.

Participants performed repeated phantom movements: knee flexion/extension, ankle dorsiflexion/plantarflexion, and toe flexion/extension. Neural signals were filtered (350–7500 Hz for electroneurography, ENG), spike-detected via adaptive thresholding, and binned into firing rates. Notably, 91% of channels in one participant modulated with specific joint and direction preferences—knee signals peaked at 24.81 Hz, ankle at 11.07 Hz, and toes at 7.19 Hz—revealing spatially organized nerve activity mirroring pre-amputation anatomy.

Spiking Neural Networks: Mimicking the Brain for Prosthetic Control

Conventional decoders like support vector machines (SVM) or multilayer perceptrons (MLP) struggle with sparse, spike-based nerve data. The team employed spiking neural networks (SNNs), biologically inspired models using leaky integrate-and-fire (LIF) neurons that process discrete electrical impulses, just like the nervous system. Input spikes from ENG were encoded via thresholding or LIF simulation, feeding into a shallow SNN (56 inputs to 6 outputs for full leg movements).

Results were striking: SNNs achieved 58% accuracy (6 classes) in one patient and 68% in the other, outperforming SVM (44–54%) and MLP (47–51%). Hybrid models combining intraneural ENG with surface intermuscular EMG boosted performance to 64–73%. Confusion matrices showed reliable joint distinction (e.g., <17% knee-ankle errors) but some flexion/extension overlap, highlighting refinement opportunities.

• Key Advantage of SNNs: Low-power, event-driven processing suits implantable devices, unlike rate-based AI requiring constant computation.
• Single Implant Bidirectionality: Same electrodes record motor outflows and stimulate for sensory feedback, minimizing surgery.

Restoring Natural Sensation: Beyond Motor Control

The true revolution lies in sensory restoration. Intraneural stimulation evoked localized percepts—knee/calf tingles, ankle/heel pressure, toe/sole touch—matching motor maps with minimal overlap (7–16%). This early segregation in the sciatic nerve allows targeted feedback: pressure sensors in a prosthesis could stimulate specific fibers, conveying terrain texture or balance without paresthesia (unnatural buzzing).

Unlike sensory substitution (vibrations on skin), direct neural input creates "phantom sensations" that feel embodied, reducing cognitive load and improving gait stability. Lead researcher Giacomo Valle notes, "A single neurotechnology can provide both natural neural control and sensory feedback."

Photo by Leonhard Niederwimmer on Unsplash

European Collaboration: Powering Biomedical Innovation

This work exemplifies pan-European synergy. Chalmers (Sweden) led decoding and SNNs; ETH Zurich/University of Zurich (Switzerland) handled neuroinformatics; University of Freiburg (Germany) supplied TIME electrodes; University of Belgrade (Serbia) recruited patients and performed surgery; Medical University of Vienna (Austria) contributed expertise. Funded partly by EU grants, it underscores Europe's leadership in neuroprosthetics.

Chalmers' Department of Electrical Engineering focuses on neuromorphic computing for low-power implants. ETH Zurich's Institute of Neuroinformatics advances brain-machine interfaces. Freiburg's BrainLinks-BrainTools Center pioneers flexible neural probes. Such networks amplify impact, training PhD students and postdocs in multidisciplinary skills.

Patient Outcomes and Real-World Promise

The two Serbian patients, experienced prosthesis users, tolerated implants well (recordings at 13–90 days post-op). No adverse events; signals stable. Decoding toe intentions—a first—hints at dexterous control for uneven terrain. Elisa Donati emphasizes, "Aligning computational models with biology extracts movement intent efficiently."

Short-term, this enhances myoelectric prostheses. Long-term: fully implanted systems for wireless, chronic use, rivaling native legs in agility.

Challenges and Next Steps in Neural Prosthetics

Challenges persist: signal drift, electrode longevity (TIMEs last months), computational demands for real-time decoding. SNNs address power via neuromorphic chips (e.g., Intel Loihi). Clinical trials (e.g., NCT03350061) test integrated legs.

Ethical considerations: informed consent, reversibility, equitable access. EU regulations like MDR ensure safety.

Broader Impacts on Amputee Rehabilitation

1.3 million EU amputees (mostly lower limb) face mobility limits; this could halve falls, boost employment. Economic: €20B+ annual prosthetic market grows with neural tech.

Training: European MSc/PhD programs in neuroengineering (e.g., Chalmers' Bionics) prepare talent. Links to research jobs in this field abound.

Photo by Leonhard Niederwimmer on Unsplash

Related European Advances

ETH Zurich's FeelAgain (ERC-funded) restored hand sensation. Montpellier's STIMEP powers stimulation. Italy's SSSA advances legged robots. Convergence promises bionic legs rivaling biology.

Read the full study: Nature Communications paper.

Future Outlook: Towards Fully Embodied Bionic Limbs

By 2030, wireless neural prostheses could restore 80% natural function. Valle envisions, "Prostheses controlled directly, returning natural sensation." Europe's universities drive this, from bench to bedside.

For academics eyeing this field, opportunities in faculty positions at pioneering institutions.