作者: Eduardo Rivo , Javier de la Fuente , Ángel Rivo , Eva García-Fontán , Miguel-Ángel Cañizares
DOI: 10.1007/S12094-012-0764-8
关键词: Knowledge extraction 、 Receiver operating characteristic 、 Odds ratio 、 Lung cancer surgery 、 Cross Industry Standard Process for Data Mining 、 Logistic regression 、 Lung cancer 、 Pneumonectomy 、 Emergency medicine 、 Medicine
摘要: The aim of this study was to assess the applicability knowledge discovery in database methodology, based upon data mining techniques, investigation lung cancer surgery. According CRISP 1.0 a (DM) project developed on warehouse containing records for 501 patients operated with curative intention. modelling technique logistic regression. finally selected model presented following values: sensitivity 9.68%, specificity 100%, global precision 94.02%, positive predictive value 100% and negative 93.98% cut-off point set at 0.5. A receiver operating characteristic (ROC) curve constructed. area under (CI 95%) 0.817 (0.740–0.893) (p<0.05). Statistical association perioperative mortality found variables [odds ratio 95%)]: age over 70 [2.3822 (1.0338–5.4891)], heart disease [2.4875 (1.0089–6.1334)], peripheral arterial [5.7705 (1.9296–17.2570)], pneumonectomy [3.6199 (1.4939–8.7715)] length surgery (min) [1.0067 (1.0008–1.0126)]. CRISP-DM process is very suitable analysis, improving decision making as well quality management.