Research Department: Bioinformatics Graduation Date: May 2019
Abstract: Traumatic Brain Injury (TBI) is an insult to the brain from an external mechanical force, which can impair cognitive, physical, and psychosocial functions. It can also cause diffuse axonal injury and the shifting and compression of structures within the brain. Each year 2.8 million people sustain a TBI in the U.S. TBI is the leading cause of death and disability from ages 1 to 44. Despite high prevalence, the standard of care largely uses subjective measures, classifying patients into rough injury categories: mild, moderate, severe, and does not predict individual quantitative outcomes. New tools that objectively measure injury severity and predict outcomes could improve patient management and guide clinical decisions regarding the use of experimental interventions. This research builds predictive models that use image based features from semi-acute MRI, 2 weeks post injury, to predict 6 month outcomes. Using a database of 29 TBI patients with 2 week and 6 month T1 MPRAGE MRI and GOS-E outcome scores, we construct multiple, deep learning, dense neural network models that are well suited to predict 6 month GOS-E outcomes and additional models to infer the difference in GOS-E from 2wks to 6 months post injury. The models achieve accuracy far above chance, including 62.5% accuracy predicting the GOS-E 6mo outcome for individuals with good statistically reliability (p-value = 0.01).
What does research mean to you? I’ve always wanted to do work that brings communities closer and makes other people’s lives easier. For me, research is the opportunity to make a difference in society because research is the hope of finding solutions to our society’s most current issues. Biomedical research motivates me the most because the work that I do in the lab has the potential of benefitting those in need. The countless hours that I spend developing methods and algorithms could mean that critical patients will have access to the resources and tools when they need it the most. By researching more effective technological approaches, individuals could have a safe, efficient recovery that allows them to quickly return to their lives and communities. Tell us about your journey. I was first introduced to research when my friend asked me to join them in the SpaceX Hyperloop Competition. I didn’t know much about anything, but I was willing to learn. Eventually, our team was able to push our way up to the semifinalist round where we competed alongside schools like MIT, Princeton, and Cornell. Being among top tier schools made me feel like our work had significance and encouraged me to pursue more research. Afterwards, I joined Dr. Izen’s lab where I researched wire bonds for the Large Hadron Collider throughout my sophomore year. By junior year, I had wanted to explore research in the biomedical sciences, so I applied to the Green Fellowship.
I applied to be part of Dr. Albert Montillo’s deep learning lab because I was interested in how mathematical models and artificial intelligence could change the way we approach medicine. I had minimal experience with coding, but Dr. Montillo was very patient and helped me get up to speed with some basic coding and A.I. concepts. Soon, I was put on my own project where I researched image-based deep learning as a possible solution for predicting prognosis of TBI patients. During my research, I was most surprised by how much of my work was trial and error. Because there isn’t much conclusive work on TBIs, we didn’t have a clear-cut direction for our project. Often, we would start implementing a method or algorithm, only to realize that, in the end, it wouldn’t work. However, before the end of the semester, Dr. Montillo and I were able to pull together some preliminary results which we could use to apply for more data nationally to build better predictive models. Working at UT Southwestern has inspired me to pursue research as a career. Although I’m still unsure which type of degree I will pursue, I know that in the end I want to be making scientific advances in the biomedical field.
Advice for Future Green Fellows
For those looking to apply: If you have any inclination towards research, don’t hesitate. Research at UT Southwestern is much different from research at UTD. For one, all your time will be dedicated towards research rather than split between classes and assignments. You’ll also get to interact with brilliant personalities in the scientific community, which could be valuable to shaping your perspective and ambitions. For those already in the program: Make the most of your time and resources. Talk to everyone and make some meaningful connections. The opportunities and people you meet during your Fellowship could help you along the way as you develop your career.