Integration of Neural Decoding in Prosthesis

            The clinical applications of neural innervation in prosthetics are limited by two main factors: invasiveness of neural surgery and the pattern recognition limitations of the current software to decode neural communication patterns. The article, “Neural Prosthesis Using Brain Activity to Decode Speech,” addresses recent breakthroughs in overcoming such limitations. They introduce a new machine learning model that utilizes intracranial EEG electrodes to decode speech patterns. This breakthrough surpasses prior methods that require outside devices such as keyboards or microphones to decode speech patterns and utilizes BCI processing which is controlled directly by the brain. 

Prior to the most recent breakthrough, the challenge in utilizing neural prosthesis is that the surgery required to implant electrodes into brain tissue is both highly invasive and dangerous. This required a surgery that used a vast amount of electrodes to cover a large area of the cortical surface. In the long term, this is not feasible and has a great amount of risk associated with it. The article discusses a new way in achieving the same effect as neural prosthesis by utilizing a smaller surface and less electrodes which minimizes the invasiveness of the procedure. Using sEEG’s, a two stage architecture for internal speech representations can be extracted from records of brain activity to produce log-mel spectral coefficients with a 55% accuracy rate from 5 channels of electrodes and 70% with 8 electrode channels. 

“Robotic Leg Control with EMG Decoding in an Amputee with nerve transfers” discusses a similar approach to neural innervation from a prosthetic standpoint, but focuses on re-innervating nerves and then using EMG signaling to derive meaning from the re-innervated nerves. This then utilizes a pattern recognition algorithm to decode the neural communication patterns from the 8 natively innervated residual muscle tissues which must be trained on the spot. This technique relies heavily on high quality EMG signal recordings through electrodes which must remain in contact with the residual limb the entire time during walking. This poses a challenge in itself as this can be uncomfortable for the amputee as well as hard to consistently ensure. 

Both articles lean into the neural innervation or decoding aspect of prosthetics to linearize the connection between the amputee and their prosthetic. By utilizing the neurological pathways, communication between the amputee and their prosthetic can mimic that of the average human being. It also helps the amputee to establish more efficient routines in their daily life. 

Hargrove, L. J.; Simon, A. M.; Young, A.; Lipschutz, R. D.; Finucane, S. B.; Smith, D. G.; Kuiken, T. A. Robotic Leg Control with EMG Decoding in an Amputee with Nerve Transfers. The New England Journal of Medicine 2013, 369 (13), 1237–1242. https://doi.org/10.1056/nejmoa1300126. 

News, N. Neural Prosthesis Uses Brain Activity to Decode Speech - Neuroscience News. Neuroscience News. https://neurosciencenews.com/speech-decoding-neuroprosthetic-22304/ (accessed 2023-12-13).