Discipline: Computer Sciences and Information Management
Subcategory: Computer Science & Information Systems
Room: Exhibit Hall A
Cedric Caldwell - Alabama State Univeristy
Co-Author(s): Armand Burks, Alabama State University, Montgomery, Alabama
Heart disease is a serious global issue that is responsible for millions of deaths worldwide each year. There is a large number of factors that doctors can consider in order to determine whether a patient has a heart disease. Therefore, accurate heart disease prediction models can aide as well as improve detection and treatment. We present a genetic programming technique for heart disease detection using the well-known UCI heart disease data set. We also discuss the feasibility of using genetic programming for evolving heart disease prediction models. Although the data set contains a total of 76 attributes, only a subset of 14 attributes have been utilized in many published experiments. However, since genetic programming is known for its capability of performing automatic feature selection and synthesis, genetic programming is also a promising technique for discovering unknown relationships between the other attributes and heart disease. The results of the program we have created yields an average accuracy ranging from 80% to 85%, with the best accuracy return being 88%. The current average and best accuracy are obtained using only conventional genetic programming configuration with minimal parameter tuning. Future work includes exploring advanced genetic programming techniques to further improve the overall accuracy of the program. We are also investigating the feasibility of incorporating the full data set of 76 attributes in effort to discover if there are any unknown relationships between the other 62 data attributes.
Funder Acknowledgement(s): NSF LSAMP
Faculty Advisor: Dr. Armand Burks, email@example.com
Role: I was heavily involved in all aspects of this research. I read through the relevant papers regarding the data set we used in this research to discover the format of the data. I also created the program that reads the data from a file and then uses the data of the different patients in order to determine whether or not the program detects heart disease or not.