摘要
目的对阿尔茨海默病(Alzheimer′s disease,AD)进程[认知正常(cognitive normal,CN)、早期轻度认知障碍(early mild cognitive impairment,EMCI)、晚期轻度认知障碍(late mild cognitive impairment,LMCI)和AD]进行多分类预测,为制定个性化诊疗方案提供参考。方法利用阿尔茨海默病神经影像学计划(Alzheimer′s disease neuroimaging initiative,ADNI)数据库中的527例个体的27个变量,进行特征选择筛选特征子集、SMOTE过采样处理类别不平衡后构建两个集成分类模型XGBoost和Bagging,并将分类性能与朴素贝叶斯和K-近邻进行比较。结果使用经SMOTE过采样后构建的Bagging集成模型准确率最高(94.40%);Bagging对EMCI、LMCI和AD的类准确率较高,分别为100.00%、88.00%和87.00%,Bagging模型性能较优。结论本文构建的AD进程多分类Bagging模型,不仅可实现直接多分类,而且有较高的准确率,可为临床AD的诊疗工作提供借鉴。
Objective To perform multi-classification prediction of cognitive normal(CN),early mild cognitive impairment(EMCI),late mild cognitive impairment(LMCI)and Alzheimer′s disease(AD),and to provide reference about personalized care plans.Methods The data derived from 27 variables of 527 individuals in the Alzheimer′s disease neuroimaging initiative(ADNI)database.Feature selection was performed to screen feature subsets,and SMOTE oversampling was used to deal with class imbalance problems of data.We constructed two integrated multi-classification models(XGBoost and Bagging),and compared the classification performance with Naive Bayes and K-nearest neighbor.Results The Bagging had the highest accuracy of 94.40%.In the Bagging model,the class-accuracy of EMCI,LMCI and AD were 100%,88%and 87%respectively,so Bagging model was optimal.Conclusion Our study constructed a multi-classification model-Bagging for AD progression,which can perform multi-classification directly,and has high accuracy.This model provides suggestion for clinical diagnosis and treatment of AD.
作者
张嘉嘉
易付良
杨慧
陈杜荣
秦瑶
崔靖
白文琳
韩红娟
葛晓燕
余红梅
Zhang Jiajia;Yi Fuliang;Yang Hui(Department of Health Statistics,School of Public Health,Shanxi Medical University,030001,Taiyuan)
出处
《中国卫生统计》
CSCD
北大核心
2022年第5期675-679,684,共6页
Chinese Journal of Health Statistics
基金
国家自然科学基金资助项目(81973154)。
