支持向量机模型在脑出血早期预后判断中的应用  

Evaluation of early prognosis in intracerebral hemorrhage based on support vector machine

作　　者：伍刚[1] 刘广韬[1] 周青[1] 刘策[1] 常鹏飞[1] 

机构地区：[1]中国人民解放军第三O九医院神经外科,北京100091

出　　处：《山西医科大学学报》2016年第6期536-538,共3页Journal of Shanxi Medical University

基　　金：解放军第三○九医院院内科研课题资助项目(2014MS-009)

摘　　要：目的通过应用支持向量机模型预测脑出血患者早期预后转归。方法收集自发性脑出血患者310例,随访观察1个月的临床转归情况。将入组患者按时间顺序以3∶1的比例分为数据训练组和验证组,训练组作为训练样本,用于筛选变量和建立预测模型,计232例;验证组作为验证样本,用于评价模型预测效果,计78例。结果通过支持向量机模型对78例脑出血患者的预测判别验证,支持向量机的预测准确度76.9%,敏感度77.3%,95%可信区间54.6%-92.2%,特异度76.8%,95%可信区间63.6%-87.0%。结论采用支持向量机模型能较好地判断自发脑出血患者早期预后。Objective To predict the early prognosis of patients with intracerebral hemorrhage（ICH）using support vector machine model. Methods Totally 310 patients with spontaneous intracerebral hemorrhage were collected and clinical outcome were followed up for 1 month. All the patients were divided into two groups with 3 ∶ 1 ratio by time sequence：training group（n = 232）and validation group（n = 78）. The patients in training group were chosen as the training samples for establishing the prediction model. The patients in validation group were chosen as the validation samples to evaluate the forecasting results. Results The prediction accuracy for ICH early prognosis was 76. 9%,the sensitivity was 77. 3%（95% CI 54. 6%- 92. 2%）,the specificity was 76. 8%（95% CI 63. 6%-87. 0%）in validation group with SVM model. Conclusion SVM model can better predict the prognosis in the patients with spontaneous intracerebral hemorrhage.

关 键 词：脑出血 支持向量机 预后 

分 类 号：O29[理学—应用数学]