|本期目录/Table of Contents|

[1]毛钤镶,承垠林,赖聪,等.基于机器学习的经皮肾镜碎石取石术结石残留研究*[J].生物医学工程研究,2021,02:114-120.
 MAO Qianxiang,CHENG Yinlin,LAI Cong,et al.Research on percutaneous nephrolithotomy kidney stone residue based on machine learning[J].Journal of Biomedical Engineering Research,2021,02:114-120.
点击复制

基于机器学习的经皮肾镜碎石取石术结石残留研究*(PDF)

《生物医学工程研究》[ISSN:1006-6977/CN:61-1281/TN]

期数:
2021年02期
页码:
114-120
栏目:
出版日期:
2021-06-25

文章信息/Info

Title:
Research on percutaneous nephrolithotomy kidney stone residue based on machine learning
文章编号:
1672-6278 (2021)02-0114-07
作者:
毛钤镶1 承垠林1 赖聪2 汤壮2 许可慰2 周毅3△
1.中山大学生物医学工程学院,广州 510006;2.中山大学孙逸仙纪念医院,广州 510030;3. 中山大学中山医学院医学信息教研室,广州 510080
Author(s):
MAO Qianxiang1CHENG Yinlin1LAI Cong2TANG Zhuang2XU Kewei2ZHOU Yi3
1.School of Biomedical Engineering, Sun Yat-sen University, Guangzhou 510080, China; 2.Sun Yat-sen Memorial Hospital, Sun Yat-sen University, Guangzhou 510030; 3.Department of Biomedical Engineering, Medical College, Sun Yat-sen University, Guangzhou 510080
关键词:
随机森林XGBoost 支持向量机经皮肾镜碎石取石术肾结石
Keywords:
Random forestsXGBoostSupport vector machinePercutaneous nephrolithotomy Kidney stone
分类号:
R318
DOI:
10.19529/j.cnki.1672-6278.2021.02.02
文献标识码:
A
摘要:
本研究基于随机森林(random forests,RF)、XGBoost 和支持向量机(support vector machine, SVM)等多种机器学习方法建立预测模型,探讨其对肾结石患者行经皮肾镜碎石取石术(percutaneous nephrolithotomy,PCNL)术后结石残留情况的预测价值。通过准确率、特异度及灵敏度等指标评价模型性能,绘制受试者工作曲线(receiver operating characteristic curve,ROC),计算曲线下面积(area under curve,AUC),RF为0.838,XGBoost为0.818,SVM为0.839,均高于传统结石评分系统。由RF和XGBoost建立的预测模型得到不同变量预测重要性占比,筛选出结石负荷、结石数量、结石CT值等是影响肾结石PCNL术后残留的重要预测因素。本研究对改善患者术后预后结果,提升临床治疗水平具有积极意义。
Abstract:
We established a prediction model based on various machine learning methods, such as random forests (RF), XGBoost and support vector machine (SVM), to explore the predictive value of these methods on stone retention after percutaneous nephrolithotomy (PCNL) in patients with renal calculi. The model performance was evaluated by accuracy, specificity and sensitivity. Receiver operating characteristic curve (ROC) was drawn and the Area Under curve (AUC) was calculated. The random forest was 0.838, XGBoost was 0.818, and SVM was 0.839, which were higher than the traditional stone scoring system. The importance ratio of different variables were obtained through random forest and XGBoost,the stone load, stone number and stone CT value were selected as the important predictors of residual renal calculi after PCNL operation. It is of positive significance to improve postoperative prognosis and enhance clinical treatment level.

参考文献/References

备注/Memo

备注/Memo:
(收稿日期:2020-02-25)国家重点研发计划项目(2018YFC0116900);国家自然科学基金资助项目(61876194);广东省科技创新战略专项项目(202011020004);广东省自然科学基金项目(2021A1515011897)。△通信作者Email:zhouyi@mail.sysu.edu.cn
更新日期/Last Update: 2021-07-21