Advances and Applications in Statistics
Volume 7, issue 2, Pages 303 - 311
(August 2007)
|
|
CLASSIFICATION TREES IN DETERMINING HEALTH OUTCOMES, USING GENETIC INFORMATION
Demosthenes B. Panagiotakos (Greece) and George Dedoussis (Greece)
|
Abstract: Classification algorithms are used in a variety of medical domains for rule induction, prediction and classification. We sought to evaluate the ability of a classification algorithm that builds decision trees for the task of classifying cardiac patients. We studied a biological database including 848 coronary patients and 1078 matched by age and sex apparently healthy individuals, in order to explore the association between: family history of CHD, various polymorphisms, physical activity and smoking status, as well as presence of hypertension, hypercholesterolemia and diabetes, with the risk of developing non-fatal acute coronary syndromes. The classification algorithms showed several associations between the investigated variables and the outcome and revealed pathways, which may explain the mechanisms leading to an excess risk of coronary heart disease. |
Keywords and phrases: classification, trees, decision-making, genetics, predictive models, epidemiology. |
|
Number of Downloads: 361 | Number of Views: 1087 |
|