New Brain Computer Interfaces for Neurologically Impaired Subjects Augmented with Gaming Technology to Improve Cognitive functions in ADHD Subjects

By: Nithin P.
Year: 2021
School: Northwood High
Grade: 11
Science Teacher: Jane Yoon

Amyotrophic lateral sclerosis (ALS), a progressive neuromuscular degenerative disease, restricts patients’ communication capacity a few years after onset resulting in a severe degradation in their quality of life. ALS patients currently have a means to communicate through non-invasive brain-computer interfaces (BCI).

This research adapts and applies BCI techniques to potentially mitigate ADHD by minimizing “Distraction/mind-wander.” Novel computer games using sophisticated algorithms generating “distraction-feedback” are developed real-time to potentially help subjects refocus when distracted. Two examples with such feedback overlays: a BCI based contact-free “Typewriter” and a “Crossword puzzle” are demonstrated. With extensive simulations, even partial feedback was shown to dramatically reduce user distraction potentially improving their BCI performance by as much as 50%.

Further, as a secondary topic, numerous new algorithms as well as optimization techniques for BCI are introduced to help ALS patients. More specifically, novel BCI schemes including “Huffman” and probabilistic flashboard scanning, auto-word completing suggestions using “Djikstra’s” algorithm and sophisticated multi-level language models fused with smoothing algorithms are developed showing dramatic improvements over current state-of-art techniques. Finally, the entire implementation is migrated to a “Raspberry Pi” creating a sub-100$ interface!

This work has resulted in a peer-reviewed IEEE conference paper [1] setting new directions in BCI applications and techniques opening rich topics for future study.

Results of the BCI optimization are first presented. Optimization results using the 48 subjects show that among the smoothing schemes, “Knesser-Ney” outperforms other techniques. The new diagonal and Huffman scanning techniques introduced along with word auto-completion significantly outperform the current state-of-art with improvements as much as 50% for some subjects in the number of flashes required to decode the text. A big gap between theory and practice is now closed. Next, game simulations show that almost all of the ADHD subject distraction is perfectly detected for medium-to-severe distraction and game completion time is vastly improved with feedback by reducing the total number flashes required. A new direction is set with novel applications of BCI uncovering rich topics for further study.