作者: Shafi Habibi , Maryam Ahmadi , Somayeh Alizadeh
关键词: C4.5 algorithm 、 Tree (data structure) 、 Medicine 、 Data pre-processing 、 Mass screening 、 Decision tree 、 Data mining 、 Body mass index 、 Type 2 Diabetes Mellitus 、 Diabetes mellitus
摘要: OBJECTIVES: The aim of this study was to examine a predictive model using features related the diabetes type 2 risk factors. METHODS: data were obtained from database in control system Tabriz, Iran. included all people referred for screening between 2009 and 2011. considered as “Inputs” were: age, sex, systolic diastolic blood pressure, family history diabetes, body mass index (BMI). Moreover, we used diagnosis “Class”. We applied “Decision Tree” technique “J48” algorithm WEKA (3.6.10 version) software develop model. RESULTS: After preprocessing preparation, 22,398 records mining. precision identify patients 0.717. age factor placed root node tree result higher information gain. ROC curve indicates function identification those individuals who are healthy. high capability model, especially healthy persons. CONCLUSIONS: developed decision T2DM which did not require laboratory tests diagnosis.