摘要
就最适合应用于疾病风险预测的4种机器学习经典算法,即支持向量机、BP(back propagation)神经网络、随机森林和朴素贝叶斯,对其在疾病风险预测中的前沿应用、方法学特征、优势、缺陷和适用条件进行综述,以期为更合理地应用机器学习方法预测疾病风险提供方法学支持。
Four classical machine learning algorithms,namely support vector machine,back propagation neural network,random forest and naive Bayes,which were most suitable for disease risk prediction,were reviewed in terms of their frontier applications,methodological features,advantages,disadvantages and applicable conditions in order to provide support for more reasonable application of machine learning methods in disease risk prediction.
作者
黄光成
周良
石建伟
黄蛟灵
杨燕
陈宁
刘茜
巩昕
王朝昕
唐岚
俞文雅
HUANG Guangcheng;ZHOU Liang;SHI Jianwei;HUANG Jiaoling;YANG Yan;CHEN Ning;LIU Qian;GONG Xin;WANG Zhaoxin;TANG Lan;YU Wenya(Shanghai Jiao Tong University School of Medicine,Shanghai 200025,China;School of Management and Finance,Tongji University,Shanghai 200092,China;School of Medicine,Tongji University,Shanghai 200092,China;Shanghai Heart Failure Research Center,Shanghai East Hospital,Shanghai 200120,China;Weifang Community Healthcare Center,Pudong New Area,Shanghai 200122,China)
出处
《中国卫生资源》
北大核心
2020年第4期432-436,共5页
Chinese Health Resources
基金
国家自然科学基金面上项目(71774116)
2018年“重大慢性非传染性疾病防控研究”重点专项(SQ2018YFC130057)
上海市卫生和计划生育委员会面上项目(201740202)
上海市浦江人才计划资助(2019PJC072)
2019上海市社区卫生协会社区科研项目(201940052)
2019年度上海市浦东新区卫生科技项目(PW2019A-42)。
关键词
机器学习
疾病风险预测
支持向量机
BP神经网络
随机森林
朴素贝叶斯
machine learning
disease risk prediction
support vector machine
back propagation neural network
random forest
naive Bayes
