|
Close Details
|
|
2012 - 11th Annual NC State Summer Undergraduate Research Symposium
|
|
Session Time :
8/1/12 1:30 PM - 8/1/12 2:45 PM
|
|
Content Area : Research Experience for Undergraduate Mathematics: Modeling and Industrial Applied Mathematics
|
Lead Student Presenters : Glenn Daniel Sidle
|
Co-Presenters :
Jay Xu
|
Co-Presenters :
Kayla Coleman
|
Co-Presenters :
Lauren Grana
|
Abstract Title : HIV Patient Health Prediction Using Data Mining Techniques
|
Abstract :
Many models of HIV infection require some sort of underlying assumption about the biological mechanisms of infection, and simple statistical techniques often struggle to capture the nonlinear dynamics of this system. In order to capture these nonlinear dynamics without impressing biological assumptions on the system, we will examine the use of data mining techniques. The techniques we use are artificial neural networks, a type of generalized nonlinear regression modeled after biological neural networks, and regression trees, a type of data mining procedure modeled after a decision flow chart. We will use these techniques based on past patient CD-4 count, CD-8 count, viral load, and drug adherence to predict future patient health. We will describe techniques for creating both patient-specific models and a model created for all patients. After analyzing the model’s ability to predict both short and long term patient health, we will describe how we can use these techniques to create a “virtual” clinical cohort on which we can evaluate the impact of different clinical trials. Finally we will show how this model can be used by physicians to understand how to optimally treat each patient.
|
Mentor and/or Co-Author : John A. David
|