摘要
慢性肾病是全球十大疾病死因之一,发病率很高,且难以发现,如果可以在早期的时候就发现肾病的存在,将会大大降低因肾病而死亡的人数。首先从UCI数据库中获取早期慢性肾病患者的各项生理指标,提取出导致慢性肾病的关键变量,然后基于Logistic回归模型进行数据的拟合。最后通过5折交叉验证,模型的准确率比普通Logistic回归模型约高出了5.6%,召回率值高出了10.4%,精确率高出了1.1%,F1值高出了5.8%,AUC值高出了5.2%。由此可见,可以为慢性肾病的早期筛查提供可靠的科学依据。
Chronic kidney disease(CKD) is one of the top 10 causes of death in the world. It is very common and difficult to detect. If we can detect the presence of CKD at an early stage, the number of deaths from CKD will be greatly reduced. In this paper, we first obtain physiological indicators of patients with early chronic kidney disease from the UCI(University of California Irvine) database, extract the key variables leading to chronic kidney disease,and then fit the data based on Logistic regression model. Finally, with 50% cross validation, the accuracy of the model in this paper is approximately 5.6% higher than that of the ordinary Logistic regression model, with 10.4% higher recall, 1.1% higher accuracy, 5.8% higher F1 value, and 5.2% higher AUC. As can be seen, this can provide a reliable scientific basis for early screening for chronic kidney disease.
作者
韩瑜
王波
周振宇
杜晓昕
HAN Yu;WANG Bo;ZHOU Zhen-yu;DU Xiao-xin(College of Computer and Control Engineering,Qiqihar University,Heilongjiang Qiqihar 161006,China)
出处
《齐齐哈尔大学学报（自然科学版）》
2023年第1期13-19,共7页
Journal of Qiqihar University(Natural Science Edition)
基金
黑龙江省教育厅基本科研业务费面上项目(145209125)
齐齐哈尔大学研究生创新科研项目资助(YJSCX2021014)。
