摘要
目的:利用决策树模型挖掘常见的临床检验资料信息,进一步提高慢性乙型肝炎及相关疾病的确诊率.方法:将临床收集的102例慢性乙肝患者和80例肝癌及肝硬化患者常见的17种信息和临床检测结果综合分析,利用决策树卡方自动交互探测(CHAID)和分类与回归树(CRT)两种算法构建预测模型,并采用正确预测率和交互印证对其进行风险评估.结果:进入CHAID和CRT两种算法模型的主要变量是年龄和胆红素指标及职业等,两模型预测慢性乙型和肝炎肝硬化及肝癌的总体准确率分别为71.4%和74.2%.结论:决策树模型在数据挖掘,资料再利用方面效果良好.
AIM: To collect and analyze the potential information from the clinical data by the decision tree model, and improve the accuracy of chronic viral hepatitis B (CVHB) , cirrhosis of liver(CL) and hepatocellular carcinoma (HCC) diagnosis. METHODS: Tree structure models were built with chisquared automatic interaction detection (CHAID) and classification and regression tree (CRT) algorithms using normal clinical data from 102 CVHB and 80 CL or HCC patients, and same time were evaluated by cross-Validation and Correct Rate of Predicative. RESULTS: In the two types of model, both age and bilirubin were risk factors, the accuracy rate in predicted classification was 71.4% and 74.2% , respectively. CONCLUSION: Decision tree model is a promising method in data mining and re-utilizing.
出处
《第四军医大学学报》
北大核心
2008年第12期1135-1137,共3页
Journal of the Fourth Military Medical University
基金
河北省科技厅资助(05276101D-86)
关键词
肝炎
乙型
决策树
诊断
hepatitis B
decision trees
diagnosis