The competition included 18 teams from across the country and was held at the University of South Carolina on February 7–9, before the National Big Data Health Science Conference.
“Winning this event demonstrates that our students are trained well in not only statistical methodologies, but also are trained to communicate well,” said Sudipta Dasmohapatra, director of the MSS program. “The students presented their findings to a panel of industry and academic professionals—this is a showcase of how students are able to apply theories that they have learned in a classroom setting to a real-world business scenario. These students work under pressure to build models, as well as showed their critical thinking, team work, and leadership skills.”
During the competition, the teams worked to develop a solution to a case inspired by real-life events that occurred in January 2005. Two trains collided outside in Graniteville, South Carolina, exposing nearby residents to dangerous chlorine gas. The residents had to be quarantined and quickly triaged en masse.
Coronado, Han, and Voisin explained that their solution prioritized usability and balanced accuracy, interpretability, and scalability. They used machine learning methods, including K-means and Random Forests, as well as metrics like Hamming distance and time complexity, all while using as few inputs as possible via mobile platforms that could work alongside existing systems.
“This was a great opportunity to put the skills we've learned in class to the test in something that can have a real impact,” the team said. “It forces you to think outside the box and consider all aspects other than just the model/algorithm which is common a classroom or project setting. Winning the competition means that our method provided a viable new technology that could be used in real-life to improve patient care and save lives.”
The three team members will split a $3,000 first-place prize.