{"title":"Decision Trees for Predicting Risk of Mortality using Routinely Collected Data","authors":"Tessy Badriyah, Jim S. Briggs, Dave R. Prytherch","country":null,"institution":"","volume":62,"journal":"International Journal of Computer and Information Engineering","pagesStart":227,"pagesEnd":231,"ISSN":"1307-6892","URL":"https:\/\/publications.waset.org\/pdf\/248","abstract":"It is well known that Logistic Regression is the gold\r\nstandard method for predicting clinical outcome, especially\r\npredicting risk of mortality. In this paper, the Decision Tree method\r\nhas been proposed to solve specific problems that commonly use\r\nLogistic Regression as a solution. The Biochemistry and\r\nHaematology Outcome Model (BHOM) dataset obtained from\r\nPortsmouth NHS Hospital from 1 January to 31 December 2001 was\r\ndivided into four subsets. One subset of training data was used to\r\ngenerate a model, and the model obtained was then applied to three\r\ntesting datasets. The performance of each model from both methods\r\nwas then compared using calibration (the \u03c72 test or chi-test) and\r\ndiscrimination (area under ROC curve or c-index). The experiment\r\npresented that both methods have reasonable results in the case of the\r\nc-index. However, in some cases the calibration value (\u03c72) obtained\r\nquite a high result. After conducting experiments and investigating\r\nthe advantages and disadvantages of each method, we can conclude\r\nthat Decision Trees can be seen as a worthy alternative to Logistic\r\nRegression in the area of Data Mining.","references":null,"publisher":"World Academy of Science, Engineering and Technology","index":"Open Science Index 62, 2012"}