Abstract
Abstract
INTRODUCTION
Brain computer interfaces (BCI) could pave the way for the 5.4 million paralysis patients to regain at-home autonomy.
METHODS
Two four-contact electrodes were chronically implanted over the dominant hand/arm region of primary motor cortex and connected to a subclavicular implantable generator that streams 4 channels of ECoG data in real-time (Resume II leads/Activa PC+S, Medtronic, Minneapolis, MN). Motor intent data were collected by instructing the subject to move/rest their dominant hand with random duration of each state. Decoding methods and adaptive parameter refitting methods were determined using cross-validation to select decoder architecture and hyper-parameters. A smart phone application was developed to allow the subject simplified control of the BCI. The BCI application was configured to run on a small battery powered PC mounted on the subject's wheelchair.
RESULTS
Using the mobile application, the subject can set decoder sensitivity and initiate decoder recalibration independently. The BCI was designed to extend the control output beyond hand grasp for additional functionality (e.g. ability to trigger on/off external smart devices). All components of the BCI were housed in a custom-designed case, mounted on the subject's wheelchair to avoid impairing patient mobility while using the BCI.
CONCLUSION
We demonstrate a successful implementation of a BCI for at-home use, driven by user feedback and ease-of use. Our next steps will focus on long-term, at-home performance and user-guided improvements to expand the control output. Control could be expanded to smart-home functions or prosthetic devices. Future work will focus on the decoding of additional degrees of freedom for the already implanted device that could be used to allow further independence.