随机森林在男性不育症证候学研究中的应用  

The application of random forest in the studies of male infertility syndrome

作　　者：钟晓君[1] 张强 秦国政 ZHONG Xiaoj;ZHANG Qiang;QIN Guozheng(Yunnan Traditional Chinese Medicine Hospital,Kunming 650021,China)

机构地区：[1]云南省中医医院,云南昆明650021 [2]昆明文理学院,云南昆明650221 [3]云南中医药大学,云南昆明650500

出　　处：《现代医院》2022年第10期1626-1628,共3页Modern Hospitals

基　　金：国家自然科学基金委员会地区科学基金项目(81260540)。

摘　　要：目的探索男性不育症患者证候类型的分布,基于随机森林方法探索基本证候类型的重要影响因素。方法利用《男性不育症临床调查表》,调查1000例符合入选标准的男性不育症患者,用R软件通过描述性统计方法分析证候类型的分布,使用随机森林方法计算变量的重要性排序,根据袋外估算误差率筛选变量。结果患者的基本证型和复合证型共有26种组合,基本证型的患者有543例,复合证型的患者共有144例。肾阳虚衰证占比最高。结论随机森林方法预测各基本证型的误判率极低,变量数为4时,袋外估算误差率较低。畏寒肢冷、神疲乏力、阳虚体质、性欲情况是对肾阳虚衰证影响较大的4个因素。Objective To explore the distribution of syndromes of male infertility and important influencing factors of basic syndrome types based on random forest method.Methods The Clinical Questionnaire for Male Infertility was used to investigate 1000 male infertile patients who met the inclusion criteria.The distribution of syndrome types was analyzed by descriptive statistics with R software.The importance of variables was ranked by random forest method.Variables were screened according to the error rate of out of pocket estimation.Results There were 26 basic syndromes and composite syndromes in 1000 cases,of which 543 were basic syndromes and 144 were composite syndromes.Kidney yang deficiency syndrome accounts for the highest proportion.Conclusion The error rate of predicting each basic syndrome type by random forest method is very low.When the number of variables is 4,out-of-bag error is low.Chilly limbs,mental fatigue,yang deficiency constitution and sexual desire are the four factors that have a greater impact on kidney yang deficiency syndrome.

关 键 词：男性不育 证候 随机森林 R软件 

分 类 号：R195.1[医药卫生—卫生统计学]