基于机器学习的膀胱癌患者生存预测模型研究  

Research on Survival Prediction Model of Bladder Cancer Patients Based on Machine Learning

作　　者：方昱衡 李泽伟 许迎盈 李功利 李科[1] FANG Yuheng;LI Zewei;XU Yingying;LI Gongli;LI Ke(University of Electronic Science and Technology of China,Chengdu 610054,China;The People's Hospital of Pingshan,Yibin 644000,China)

机构地区：[1]电子科技大学,四川成都610054 [2]屏山县人民医院,四川宜宾644000

出　　处：《现代信息科技》2024年第16期83-87,共5页Modern Information Technology

基　　金：2022年四川省重点研发项目(22ZDYF0376);2022年宜宾市科技计划项目(2022ZYD06)。

摘　　要：该研究旨在构建基于机器学习的生存预测模型,预测膀胱癌(BC)1、3和5年生存率,帮助医生准确识别预后较差的患者,并辅助临床预后方案制定。从监测、流行病学和最终结果(SEER)数据库中获取患者数据,基于逻辑回归(LR)、随机森林(RF)和梯度提升决策树(GBDT)和Cox比例风险模型(Cox proportional hazards)构建生存预测模型,通过在训练集和验证集中使用受试者工作特征曲线和校准度曲线评估模型性能。实验结果表明,GBDT在BC患者1、3和5年生存率预测方面具有较高的区分度和校准度。This research focuses on constructing a survival prediction model based on Machine Learning to predict the 1-year,3-year,and 5-year survival rates for patients with Bladder Cancer,aid clinicians in accurately identifying patients with poor prognosis and assist in formulating clinical prognosis plans.Patient data is obtained from the Surveillance,Epidemiology,and End Results(SEER)database.The survival prediction model is constructed based on Logistic Regression(LR),Random Forest(RF),Gradient Boosting Decision Tree(GBDT),and the Cox proportional hazards model.The performance of the model is evaluated using the receiver operating characteristic curve and calibration curve on the training and validation sets.The experimental results demonstrate that GBDT exhibits high discrimination and good calibration in predicting the 1-year,3-year,and 5-year survival rates for BC patients.

关 键 词：膀胱癌 生存预测 机器学习 COX回归 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程] TP391.41[自动化与计算机技术—控制科学与工程]