The use of classification and regression tree to predict 15-year survival in community-dwelling older people

Melillo, Paolo; Pecchia, Leandro; Bath, Peter A.; Bracale, Marcello (2013)
University of Sheffield
Previous research has identified various risk factors for mortality in older people. The aim of this paper was to use Classification and Regression Tree to predict 15-year survival in community-dwelling older people. Data were obtained from a United Kingdom representative sample of 1042 community-dwelling people aged 65 and over. Outcome was time from 1985 interview to death or censorship on February 29, 2000. Classification and Regression Tree is a non-parametric technique widely used in medical domain classification. We applied CART to the set of risk-factors identified in a previous research. The selected CART model is based on age, dose of drug prescribed and handgrip measures. It predicts survival with a sensitivity rate of 76.3% and a specificity rate of 66.3%. The selection of variables are consistent with previous research. Finally, we observed the range of risk factors and their combination associated with increased and decreased mortality.\ud

Download from

Cite this article