[1]李琳,杨日东,王哲,等.基于多类支持向量机递归特征消除方法特征选择的原发性肝癌患者预后预测*[J].生物医学工程研究,2019,01:32-36.
LI Lin,YANG Ridong,WANG Ze,et al.Prognosis prediction of hepatocellular carcinoma patients based on recursive feature elimination based on multiple support vector machine feature selection[J].Journal of Biomedical Engineering Research,2019,01:32-36.
点击复制
基于多类支持向量机递归特征消除方法特征选择的原发性肝癌患者预后预测*(PDF)
《生物医学工程研究》[ISSN:1006-6977/CN:61-1281/TN]
- 期数:
-
2019年01期
- 页码:
-
32-36
- 栏目:
-
- 出版日期:
-
2019-03-25
文章信息/Info
- Title:
-
Prognosis prediction of hepatocellular carcinoma patients based on recursive feature elimination based on multiple support vector machine feature selection
- 文章编号:
-
1672-6278 (2019)01-0032-05
- 作者:
-
李琳1; 杨日东2; 王哲1; 杨红梅2; 华赟鹏3; 周毅2; 张学良1△
-
1.新疆医科大学,乌鲁木齐 830011;2.中山大学中山医学院,广州 510080;3.中山大学第一附属医院,广州 510080
- Author(s):
-
LI Lin1; YANG Ridong2; WANG Ze1; YANG Hongmei2; HUA Yunpeng3; ZHOU Yi2; ZHANG Xueliang1
-
1.Xinjiang Medical University,Urumqi 830011,China;2.Sun Yat-Sen Medical College,Sun Yat-Sen University,Guangzhou 510080,China;3.The First Affiliated Hospital of Sun Yat-Sen University,Guangzhou 510080
-
- 关键词:
-
特征选择; 多类支持向量机递归特征消除方法; 列线图; 预后预测; 原发性肝癌
- Keywords:
-
Feature selection; Recursive feature elimination based on multiple support vector machine; Nomograms; Prognosis; Hepatocellular carcinoma
- 分类号:
-
R318
- DOI:
-
10.19529/j.cnki.1672-6278.2019.01.07
- 文献标识码:
-
A
- 摘要:
-
本研究通过特征选择的方法,分析肝癌患者术前临床信息,提高患者的预后模型的准确性。基于多类支持向量机递归特征消除(recursive feature elimination based on multiple support vector machine,MSVM-RFE)方法对进行过肝切除手术的原发性肝癌患者的临床变量进行重要特征排序,使用5折交叉验证的支持向量机确定最优特征子集,构造原发性肝癌患者术后的1年、3年无瘤生存和总体生存的列线图。通过与临床医生沟通,确认特征排序结果为合理的。患者3年无瘤生存风险和总生存风险的列线图的一致性指数分别为0.701和0.706。使用多类支持向量机递归特征消除方法后的预测模型准确率有所提高,列线图在临床实践中能够提供患者生存风险信息,简单清晰的反映患者的生存风险。
- Abstract:
-
To analyze preoperative clinical information of hepatocellular carcinoma patients by feature selection, improving the accuracy of prognosis prediction model.MSVM-RFE feature selection was used to rank the clinical variables of primary hepatocellular carcinoma patients who had undergone hepatectomy. Then support vector machine(SVM)was applied to confirm the optimal feature subset by using 5 - fold cross validation.The 1-year disease-free survival and 3-year of overall survival models of patients with primary hepatocellular carcinoma were constructed, representing in the Nomograms.The results of feature sequence was proved reasonable by clinicians. The C-index of Nomograms was 0.701 and 0.706 for 3-year disease-free survival risk and overall survival risk, respectively,surpassing the non-feature selection model. The MSVM-RFE feature selection can improve the accuracy of prediction model. Nomograms can provide information about patients′survival risk in clinical practice, and reflect the patients′ survival risk simply and clearly.
备注/Memo
- 备注/Memo:
-
(收稿日期:2018-08-10)国家自然科学基金资助项目(61876194,11661007);国家重点研发计划项目(2018YFC0116902,2018YFC0116904,2016YFC0901602);NSFC-广东大数据科学中心联合基金项目 (U1611261);广州市科技计划项目(201604020016)。△通信作者Email:shuxue2456@126.com
更新日期/Last Update:
2019-05-24