Sparsh Bahadur
Pronouns: He/Him
Research Mentor(s): Cynthia Chestek
Research Mentor School/College/Department: Biomedical Engineering / Engineering
Program:
Authors:
Session: Session 1: 9:00 am – 9:50 am
Poster: 68
Abstract
Brain-Computer Interfaces (BCIs) have emerged as transformative tools for individuals with motor disabilities, offering innovative solutions like robotic limbs, speech detection devices, and ventures such as Neuralink. Despite their potential, a common challenge across these advancements is excessive power consumption due to the presence of numerous unused channels. In this study, we used an ML model that finds the most important features in the prediction and recreates the movement by utilizing just the significant channels to create an optimized model for predicting movement through neural signals. We tested this by leaving a various number of features in the model to find the optimal number of channels needed to make an accurate prediction through correlation numbers. Using the remaining features, we recreated the video and watched the astonishing similarities to the original. We found that around 90% of the channels are unused and waste power. This research represents a pivotal step in the field of BCIs, contributing a technique crucial for the development of implementable BCI devices. By mitigating power consumption concerns and ensuring prolonged usage without frequent recharging, our optimized model holds promise for the widespread adoption of BCIs, fostering independence and improved quality of life for individuals with motor disabilities.
