Skip to main content

Picture - SykesThis semester as a quantitative Gil Intern, I was placed at RTI International. In all honesty, I was rather nervous with this placement as I knew RTI was a massive company that did research in virtually every possible domain. I wasn’t sure what to expect from such a placement, and I didn’t know whether I would be able to contribute meaningfully to a project or if I would simply run programs and crunch numbers as instructed.

Contrary to all my fears, my experience at RTI has been phenomenal. I was placed under the senior research statistician, Dr. Georgiy Bobashev, to assist with various modeling projects. His research focuses on personalized treatments for substance use with a systems approach to addiction, and the use of predictive modeling to forecast health outcomes. My projects all deal with modeling addiction, either direct computational predictive modeling or visual modeling of an addiction behavioral model.

The first project I worked on during the first few weeks of my internship was a 3D Brain Model of the regions implicated in Koob’s Addiction Model. Due to my undergraduate background in neuroscience, I was able to provide an ‘expert’ consulting role to refine and review the model. For this project I also worked with the modeling visualization expert and was able to learn and observe how programs used to develop video games (for example, Unity) can be used to generate scientific visualizations. The extensive modeling technology developed for video games provided an easier platform from which to display our 3D Brain that can be rotated and regions selectively highlighted. This model is more intuitively informative that two dimensional depictions, as relative location and size are easier to display and understand.

The second project I have focused on for the majority of my internship is a computation model predicting the behavior of cocaine addicted rats. We have real lever-pressing rat data of rats exposed to and subsequently addiction to cocaine, and we are subsequently trying to generate a computational model that will predict their behavior given different cocaine exposure times. This project has been fascinating and infuriating in equal parts, for as I dug deeper into the model and script (written in code and run in the statistical program R) what was meant to be a simple project became more complex and engaging. I have found several bugs in the script, for example one bug hindered the expansion of our model from 3 to 6 hours of cocaine exposure. I have also worked in refining various equations and optimizing the many parameters in the model, all in trying to get a more accurate model that follows the behavior of our actual rat data.

While I have thoroughly enjoyed working entirely with computers (I never saw a single human patient or rat), I realize this type of work is not for everyone. Research of this kind requires a certain kind of dedication to keep searching for the bug you didn’t find the first 15 times, and celebratory moments aren’t often accompanied by fanfare. Still, my internship at RTI was a perfect placement for my interests in statistics, code, and modeling of data, and I hope to continue working in a similar vein of research for years to come.

Print Friendly, PDF & Email
Comments are closed.