应用分类树模型构建缺血性脑卒中发病风险的预测模型

目的应用分类树模型构建缺血性脑卒中发病风险的预测模型,并评价其应用价值。方法采用1:1配比病例对照研究设计,选择深圳市2所综合性医院的309名缺血性脑卒中患者为病例组,同时选择按年龄、性别匹配的健康者作为对照;采用卡方自动交互检测(CHAID)法建立缺血性脑卒中发病风险的预测模型,采用错分概率Risk值、索引图及受试者工作特征曲线(ROC)评价模型的应用价值。结果所建立的分类树模型共包括4层,共19个结点,共筛检出6个解释变量;其中最为重要的预测因素为体育锻炼和高血压病史。模型错分概率Risk值为0.207,利用预测概率绘制的ROC曲线下面积为0.789,与0.5比较,差异有统计学意义(P=0.001),模型拟合的效果较好。结论分类树模型不仅能有效地拟合缺血性脑卒中发病风险的预测模型,还可以有效地筛检变量间的交互作用效应。

Study on the Application of Classification Tree Model in Building the Risk Model for Ischemic Stroke

Objective To introduce classification tree in building the risk model for ischemic stroke, and explore the value of this data mining technique. Methods A 1 ： 1 age-gender-matched case-control study was conducted. 309 patients with ischemic stroke were selected from two general hospitals in Shenzhen. The controls were selected from the same hospitals. The classification tree model was constructed using Exhaustive CHAID method and evaluated by the Risk statistics, index map and area under the ROC curve. Results The model had four stratum and nineteen nodes. Six explanatory variables were screened out in the model. The most important risk factors were physical exercise and history of hypertension. The risk value of misclassification probability of the model was 0.207, and the area under the ROC curve was 0.789 which was significantly different from 0.5, suggesting that the classification tree model fitted the actuality very well. Conclusion Classification tree model can not only properly predict the occurrences of ischemic stroke, but also reveal the complex interaction effects among the factors.