3 Current Technology

Prosthetics are designed and manufactured using sophisticated CAD-CAM technology able to mill to thousandths of an inch for extreme accuracy ⁽⁶⁾ and precise patient fit. Artificial intelligence (AI) is seen as the key control mechanism, and is now built into the latest prosthetics ⁽⁵⁾. Existing technology is computer driven but invasive. This requires electrodes to be inserted into the body to receive signals from existing nerves. EMG removes these requirements and reduces detection to external electrodes and sophisticated pattern matching algorithms.

3.1 DARPA (Defense Advanced Research Projects Agency)

In 2006 DARPA initiated a program called ‘Revolutionizing Prosthetics’ ⁽⁷⁾, originally for upper limbs, they pioneered investigation into neural control of prosthetics. In May 2012 researchers at Providence VA Medical Center, Brown University and Massachusetts General Hospital demonstrated the ability to control an advanced prosthetic arm using a direct neural interface system and AI in humans⁽⁸⁾. This technology is now moving, along with AI and pattern recognition, into the latest computer controlled prosthetics.

AI (Artificial Intelligence)

In the article ‘What is Artificial Intelligence?’ by Istvan S. N. Berkeley Ph. D⁽⁸⁾, AI is defined as ‘Artificial Intelligence is the study of man-made computational devices and systems which can be made to act in a manner which we would be inclined to call intelligent’. This is achieved through the latest computer technology and miniaturization of components now possible. These new limbs appear as natural as the original through the use of adaptive AI.

4 Computer Technologies

After EMG (Electromyography) ⁽⁹⁾, a diagnostic test to record electrical activity of muscles is completed, this information is passed to computers for pattern recognition. This recognition software is based on a sophisticated Bayesian network ⁽¹⁰⁾. Patients use a real-time virtual representation of their limbs and prosthetics to practice control. This has been found to be extremely effective in speed of limb control and error rate reduction.

In the Time article ‘The First Bionic Leg Controlled Only by Brain Power’ we see the results of this research ⁽¹¹⁾. Microprocessors in the joint receive signals from the pattern recognition software embedded in the Microchips. This information is used, real-time, to instructs the limb how to react. In addition to this pattern recognition of your own neural impulses, microchips in the limb process telemetry to understand how the limb is moving. These signals are combined with the neural signals to adjust the limb.

Data collected via telemetry and EMG allowed computational error rates in the software to be reduced from 12.9% to 1.8%. Serious computational errors are considered to be anything that might cause the patient to fall ⁽¹⁰⁾. The computational requirement to model a limb in a 3D environment is complex, but over the last 5 years has been fully documented. With the advent of smaller microprocessors this technology is readily available for artificial limbs ^(12)(13).

5 Future

The future of EMG is adaptive. Advanced pattern matching within these microprocessors creates a system which can learn. Johns Hopkins is pushing the technology to allow a learning style of advanced matching algorithm. This will allow the computers within the limb to understand when a movement was incorrectly completed, and create small adjustments whilst monitoring its own error rate ⁽¹⁴⁾. The computer will make real-time adaptions to its own algorithms as necessary. Over time the limb will start to learn and understand the user’s intentions. The limb will adapt in real-time, adjust to errors, and interface to a PC (remote or local) ⁽¹⁴⁾. These advancements will remove the burden on the user for supervised device training and allow the use of prosthetics in any environment from land to sea to air.

6 References