People with severe motor disabilities can often feel socially isolated. This project prototyped several MEMS-based solutions for translating movement into keyboard inputs, with the goal of helping people with mobilit issues play computer games. A tongue actuated joystick as a mouse and head mounted gyroscope was found to be the most accurate and usable solution.
A Multimodal Natural User Interface for People with Severe Motor Disabilities
This project built off the findings of the previous year. While MEMS based solutions were highly accurate and resulted in fair gameplay, the bulky hardware made the devices cumbersome to use. This project developed computer vision solutions to allow people with severely reduced mobility to control a keyboard and mouse system using facial muscles and eye movements. The results showed 100% accuracy accross all trials for facial feature tracking and comparable results for eye-tracking.
A Novel Graph Theory Based Approach for Identifying Drug Discovery Targets in Computationally Derived Gene-Regulatory Networks (cGRNs)
Gene-targeted drugs operate by regulating gene-gene interactions in cellular Gene Regulatory Networks. However, identifying which genes in the networks are the most optimal genes to target can be a long and expensive process. This project used graph theory methos to analyze gene regulatory networks and predict which genes exhibit the most control over the network which could make them possible drug candidates.
Identifying Pathways Enriched with Differential Dependencies using LASSO
June 2016-July 2016
This research was conducted as part of the Helios Scholar Internship Program at the Translational Genomics Research Institute in Phoenix. The research focused on using an L1-penalized linear regression technique called LASSO to infer Gene Regulatory Networks from expression data and then using statistical metrics to compare context-specific GRNs across cell classes to identify pathways that showed significant changes across cell classes. To do this, we calculate the Coefficient of Determinance and the Hamming Distances to assess a given differentiality. We combine this with a modified permutaiton test to asymptotically assess significance. Validation on simulated data yielded positive results. Feel free to download and read my paper.
I also had the pleasure of meeting President Obama as one of the finalists for the 2013 Broadcom MASTERS science competition for my work on modeling negotiations using Rock, Paper, Scissors.