分类和回归树预测重型颅脑损伤的预后分析

Classification and regression tree(CART)for prediction of outcome after severe head injury.

目的建立一个关于重型颅脑损伤预后的简单、便于使用、变量可以在日常工作中快速获得的预测准确率高的预测模型。方法采用分类和回归树(CART)分析方法,选择8个预后因子,对331例重型颅脑损伤患者的预后进行分析。预后指标由外伤后6月Glasgow结果评分(GOS)评价。结果Glasgow昏迷评分(GCS)是最好的预测因子,血糖、头颅CT表现、年龄是强有力的预测因子。入院次日晨白细胞计数也对预后产生有意义的影响。CART中的所有变量都与预后相关,预测准确率达87.9%。结论CART预测模型能较好地预测重型颅脑损伤病人的预后,是一种简单有效、准确率高的预测方法。

Objective To develop a simple and easy predictive model with a high predictive accuracy,which involves variables that can be rapidly and easily achieved in daily routine practice for outcome after severe head injury. Methods C classification and regression tree (CART) technique were employed in the analysis of data from 331 patients with severe brain injury. A total of 8 prognostic indicators were selected, while Glasgow Outcome Scale(GOS) at 6 months after head injury as prognostic criterion. Results Our results indicated that Glasgow Coma Scale(GCS) was the best predictor of outcome. Glucose level,computed tomographic findings and age were proven to be strong predictors, leukocytosis on the 2nd day was found to correlate significantly with prognosis, too. The overall predictive accuracy of CART model for these data was 87.9%. All variables included in this tree have been shown to be related to outcome. Conclusion The CART were proven useful in developing a simple and effective predictive model for outcome after severe head injury,with the variables simply and easily to get.