摘要
目的分析不同健康青年人群分类间的差异菌,以此构建基于肠道菌群的健康青年人群分类模型,说明不同分类的健康青年人肠道菌群的分布特征,为青年人的健康管理提供策略。方法将肠道菌群种层面相对丰度数据转变为定性数据,并运用聚类算法将受试人群分成不同类;分析不同类人群的差异菌,依据NB、IBK、J48、R分类算法和属性筛选算法,得到以不同类人群的肠道菌群分布差异为客观指标的人群分类模型;并对各模型性能相关的多项评价指标及各模型的ROC曲线图进行分析,以获得最优模型。结果对二类、三类、四类、五类分类人群所建立的模型筛选出了各自最优模型,所有模型测试的准确性在82.50%~100%,平衡精度在0.672~1.000,受试者工作特征曲线下面积在0.842~1.000。通过模型间的比较基于肠道菌群将受试人群分为四类时,所建模型为最优。结论基于肠道菌群种层面数据构建的健康青年人群分类模型具有可行性,通过分析不同分类健康青年人群肠道菌群的分布特征,可为其健康管理提供参考。
Objective To analyze the difference in the classification of intestinal flora among different healthy young people,build a classification model for them,explain the distribution characteristics of different types of intestinal flora,and provide strategies for the health management of young adults.Methods The relative abundance data at the species level of the intestinal flora were transformed into qualitative data.Clustering algorithm was used to classify the subject population.The NB,IBK,J48,R classification algorithms and attribute screening algorithms were used to obtain a population classification model with difference in the intestinal flora distribution as the objective indicator.Multiple evaluation indexes and ROC graph related to the performance of each model were analyzed to obtain the optimal model.Results The optimal models were screened out for different classified populations.The accuracy of all tested models ranged from 82.50%to 100%,the equilibrium accuracy ranged from 0.672 to 1.000,and the area under the subject working characteristic curve ranged from 0.842 to 1.000.The proposed models were optimal when the subject population was divided into four categories.Conclusion It is feasible to construct classification models for healthy youth adults based on the species level of intestinal flora,to help with the health management of young adults.
作者
田学梅
王慧
梁静宣
王耘
TIAN Xue-mei;WANG Hui;LIANG Jing-xuan;WANG Yun(Research Center of Chinese Medicine Information Engineering,Beijing University of Chinese Medicine,Beijing 102488)
出处
《中南药学》
CAS
2021年第11期2291-2299,共9页
Central South Pharmacy
基金
国家自然科学基金项目(No.81673697,No.81373985)。
关键词
肠道菌群
分布特征
模型构建
intestinal flora
distribution characteristics
model building