Bionic prostheses & neuroprosthetics

01Overview and value chain

Markers: [EC: MDR 2017/745 | OECD: Biotech-health | Regulator: FDA (USA), EMA (European Union), NMPA (China)]

Bionic prostheses and neuroprosthetics replace a lost limb’s function using AI-driven control, dexterous touch sensing and, increasingly, direct skeletal attachment rather than a traditional socket. Psyonic’s Ability Hand became one of the first commercial dexterous hands officially integrated into NVIDIA Isaac Lab, announced at NVIDIA GTC 2026, and the company partnered with ABB Robotics to study whether its prosthetic touch-sensor data can improve industrial robot grasping. Atom Limbs develops an AI-driven bionic arm combining neural interfaces and advanced robotics to restore natural, intuitive arm and hand function. Coapt secured a new CMS reimbursement code (L6700) for upper-limb pattern recognition prostheses in January 2025 and was acquired by Hanger, Inc. in 2025, and its COMPLETE CONTROL System Gen2 uses AI and machine learning for prosthesis control. Ottobock introduced its next-generation speedhand myoelectric prosthetic arm solution, with modular wrist options and a companion connectgrip app, and presented its prosthetics and orthotics portfolio at OTWorld 2026. Integrum’s OPRA implant system, an osseointegration technology that anchors a prosthesis directly to bone, demonstrated superior treatment outcomes in a two-year follow-up study by Walter Reed National Military Medical Center, and the company is expanding implant system access in Ukraine. In China, BrainCo’s BrainRobotics division expects bionic hand sales to grow significantly, per Chinese trade press coverage.

The key directions of bionic prostheses & neuroprosthetics are:

1. Dexterous touch-sensing hands (Dexterous Touch-Sensing Hands): high-degree-of-freedom prosthetic hands with integrated touch sensors, dexterous enough for robotics research applications beyond prosthetic use — Psyonic’s Ability Hand.
2. AI neural prosthetic control (AI Neural Prosthetic Control): neural-interface-driven bionic limbs using AI to restore natural, intuitive movement — Atom Limbs.
3. Myoelectric pattern recognition (Myoelectric Pattern Recognition): AI/machine-learning control systems that decode muscle-signal patterns for intuitive multi-grip prosthetic control, now with dedicated reimbursement coding — Coapt’s COMPLETE CONTROL System, and Ottobock’s speedhand myoelectric arm.
4. Osseointegrated prostheses (Osseointegrated Prostheses): implants that anchor a prosthetic limb directly to bone rather than a socket, validated in military rehabilitation populations — Integrum’s OPRA system.

Sectoral value chain

[control/attachment technology design] ──> [preclinical/clinical validation] ──> [regulatory clearance (FDA/MDR/NMPA)]
                                                    │
                                       (fitting/reimbursement coding)
                                                    │
                                                    ▼
[capability/indication expansion] <─── [user outcome data generation] <─────┘

Value chain levels

Cross-cutting technologies of the sector:

• Dexterous touch sensing (Dexterous Touch-Sensing Hands): integrated tactile sensors giving a prosthetic hand touch feedback precise enough to inform robotic manipulation research, not just prosthetic use.
• AI/neural control decoding (AI Neural Prosthetic Control, Myoelectric Pattern Recognition): machine-learning algorithms that decode neural or muscle-signal patterns into multi-grip, intuitive prosthetic movement.
• Osseointegration (Osseointegrated Prostheses): direct bone anchoring of a prosthesis, eliminating socket-related discomfort and improving proprioceptive feedback.

02US

The US hosts the leading concentration of bionic prosthetics innovation, spanning dexterous touch-sensing hands, AI-driven neural interfaces and myoelectric pattern-recognition control with new reimbursement pathways.

Psyonic, Atom Limbs, Coapt, FDA

• Psyonic: its Ability Hand became one of the first commercial dexterous hands officially integrated into NVIDIA Isaac Lab, announced at NVIDIA GTC 2026, and the company partnered with ABB Robotics to study whether its prosthetic touch-sensor data can improve industrial robot grasping.
• Atom Limbs: develops an AI-driven bionic arm combining neural interfaces and advanced robotics to restore natural, intuitive arm and hand function.
• Coapt: secured a new CMS reimbursement code (L6700) for upper-limb pattern recognition prostheses in January 2025 and was acquired by Hanger, Inc. in 2025; its COMPLETE CONTROL System Gen2 uses AI and machine learning for prosthesis control.
• FDA framework: bionic prostheses clear FDA 510(k)/PMA per device class, while reimbursement coverage runs through separate CMS HCPCS coding, as with Coapt’s new L6700 code specifically for pattern-recognition control.

03CN

China’s confirmed bionic prosthetics activity centers on a Hangzhou-based AI bionic hand maker reporting strong sales growth expectations.

BrainCo (BrainRobotics), NMPA

• BrainCo (Zhejiang Qiangnao Technology / 强脑科技; BrainRobotics division): its intelligent bionic hand is expected to see bionic hand sales grow significantly this year, per Chinese trade press including Lianhe Zaobao coverage, and the product is exhibited at China’s assistive-technology (福祉) expo circuit.
• NMPA pathway: bionic prosthetic devices register with NMPA as medical devices, with control-system software components subject to the same registration as the physical device.

04EU

Germany and Sweden anchor Europe’s bionic prosthetics capacity, spanning the world’s leading myoelectric prosthetics maker and a specialized osseointegration implant system with military rehabilitation validation.

Ottobock, Integrum, MDR

• Ottobock (Germany): introduced its next-generation speedhand myoelectric prosthetic arm solution, with modular wrist options and a companion connectgrip app, and presented its prosthetics and orthotics portfolio at OTWorld 2026.
• Integrum (Sweden; Nasdaq Stockholm: INTEG B): its OPRA implant system, an osseointegration technology that anchors a prosthesis directly to bone, demonstrated superior treatment outcomes in a two-year follow-up study by Walter Reed National Military Medical Center, and the company is expanding implant system access in Ukraine.
• MDR framework: bionic prostheses and osseointegration implants sold in the EU require conformity assessment under the Medical Device Regulation (MDR 2017/745), with Integrum’s implant system a Class III device under this framework.

05Leading companies and research institutes

06Tech stack and innovations

The stack applies AI-driven control, dexterous touch sensing and direct skeletal attachment to restore lost limb function beyond what traditional socket-and-cable prosthetics can achieve.

1. Dexterous touch-sensing hand technology (Dexterous Touch-Sensing Hands):
   • Psyonic’s Ability Hand integrates touch sensors precise enough that its data now informs ABB’s industrial robot grasping research, alongside its NVIDIA Isaac Lab integration.
2. AI neural prosthetic control (AI Neural Prosthetic Control):
   • Atom Limbs combines neural interfaces with AI and advanced robotics to restore natural, intuitive bionic arm movement.
3. Myoelectric pattern recognition control (Myoelectric Pattern Recognition):
   • Coapt’s COMPLETE CONTROL System Gen2 and Ottobock’s speedhand both use AI/ML-based muscle-signal decoding for multi-grip control, with Coapt’s approach now backed by a dedicated CMS reimbursement code.
4. Osseointegration and bone-anchored attachment (Osseointegrated Prostheses):
   • Integrum’s OPRA system anchors a prosthesis directly to bone, validated in a Walter Reed National Military Medical Center two-year follow-up study, eliminating socket-related discomfort for eligible amputees.

07Value chains and production pipelines

Industrial pipeline of a bionic prosthesis (FDA 510(k)/PMA / EU MDR 2017/745 / NMPA)

┌───────────────────────────┐      ┌───────────────────────────┐
│ 1. Control/attachment     │ ───> │ 2. Preclinical/clinical   │
│    technology design         │      │    validation                │
└───────────────────────────┘      └───────────────────────────┘
                                                   │
                                                   ▼
┌───────────────────────────┐      ┌───────────────────────────┐
│ 4. Fitting/reimbursement  │ <─── │ 3. Regulatory clearance  │
│    coding                    │      │                           │
└───────────────────────────┘      └───────────────────────────┘
              │
              ▼
┌───────────────────────────┐      ┌───────────────────────────┐
│ 5. User outcome data      │ ───> │ 6. Capability/indication  │
│                              │      │    expansion                │
└───────────────────────────┘      └───────────────────────────┘

Stage 1: Control and attachment technology design

The prosthetic’s control interface (myoelectric pattern recognition, neural interface) or bone-anchoring mechanism (osseointegration) is designed, as with Coapt’s pattern-recognition algorithms versus Integrum’s OPRA osseointegration hardware.

Stage 2: Preclinical and clinical validation

Dexterity, control accuracy and, for osseointegration, bone-implant stability are tested in trial and rehabilitation-center populations, the evidence base behind Integrum’s Walter Reed two-year follow-up study.

Stage 3: Regulatory clearance

The device clears FDA 510(k)/PMA, EU MDR conformity assessment, or NMPA registration, with Coapt’s control system and Integrum’s implant both progressing through their respective device-class pathways.

Stage 4: Fitting and reimbursement coding

A prosthetist fits the device to the individual user, supported by a specific insurance/CMS reimbursement code, as with Coapt’s new L6700 code specifically enabling pattern-recognition-control reimbursement.

Stage 5: User outcome data generation

Data on grip success, daily-use adoption and, for osseointegration, long-term implant stability are collected, exemplified by Integrum’s Walter Reed outcomes data and Psyonic’s touch-sensor data now feeding into ABB’s robotics research.

Stage 6: Capability and indication expansion

Outcome evidence supports new control capabilities, market expansion or adjacent applications, the pattern behind Psyonic’s NVIDIA Isaac Lab/ABB robotics expansion beyond prosthetic use, Integrum’s Ukraine market expansion, and BrainCo’s reported bionic-hand sales growth in China.