TUT’s doctoral student Oguzhan Gencoglu selected as finalist in machine learning competitionThe finalists of the ICHI2016 Healthcare Data Analytics Challenge have been announced. Doctoral student Oguzhan Gencoglu from the Personal Health Informatics research group at TUT is one of the top five finalists. His one-man-team will represent TUT in Chicago in October 2016.
In the challenge, the participant teams were given healthcare forum posts and they needed to develop machine learning solutions for predicting the topic of each of these posts using seven different categories, such as disease, treatment and family support. For the task, thousands of healthcare discussion forum posts and their corresponding topic categories were provided in text format, in addition to 3,000 posts without the ground truth for the teams to predict.
“The informal and untidy nature of the text data poses the main challenge,” Gencoglu says.
The competition has lasted for two months and altogether five teams were selected as finalists. They will present their solutions at the IEEE International Conference on Healthcare Informatics in Chicago in October.
Gencoglu's natural language processing and machine learning solution will be competing against those of IBM Watson Research Center, New York University Langone Medical Center, University of Waterloo and Humana Inc.