尔云间  一个专门做科研的团队

小伙伴们大家好啊，小云今天要和大家讲的是决策曲线的绘制，小云之前了解了一下诊断模型，感觉和之前小云接触的预后模型有点像，但不一样的是，小云今天接触的诊断模型只有一个分组，即高低分组与各基因表达量进行计算，而预后模型则是时间和状态两个量。小云今天要说的是决策曲线的绘制。

 

这个诊断模型有五个基因，然后这里的分组是用的高低分组

library(ggDCA)

library(rms)

library(nricens)

library(foreign)

library(rmda)

library(regplot)

 

data3 <- read.table(file = "input.txt",header = T,sep = "\t",row.names = 1)#载入数据

#1.诊断列线图构建

dd <- datadist(data3)

options(datadist="dd")

RAB24<- decision_curve(Risk_Score~ RAB24,data = data3,

                        family = binomial(link ='logit'),#模型类型，这里是二分类

                        thresholds= seq(0,1, by = 0.01),

                        confidence.intervals =0.95,#95可信区间

                        study.design = 'cohort')#研究类型，这里是队列研究

DNAJB9<- decision_curve(Risk_Score~ DNAJB9,data = data3, family = binomial(link ='logit'),

                      thresholds= seq(0,1, by = 0.01),

                      confidence.intervals =0.95,study.design ='cohort')

TOMM5<- decision_curve(Risk_Score~ TOMM5,data = data3, family = binomial(link ='logit'),

                      thresholds= seq(0,1, by = 0.01),

                      confidence.intervals =0.95,study.design ='cohort')

STAT3<- decision_curve(Risk_Score~ STAT3,data = data3, family = binomial(link ='logit'),

                      thresholds= seq(0,1, by = 0.01),

                      confidence.intervals =0.95,study.design ='cohort')

PHF23<- decision_curve(Risk_Score~ PHF23,data = data3,

                            family = binomial(link ='logit'),#模型类型，这里是二分类

                            thresholds= seq(0,1, by = 0.01),

                            confidence.intervals =0.95,#95可信区间

                            study.design = 'cohort')#研究类型，这里是队列研究

nomogram<- decision_curve(Risk_Score  ~ RAB24+DNAJB9+TOMM5+STAT3+PHF23,data = data3,

                          family = binomial(link='logit'),

                          thresholds= seq(0,1, by = 0.01),

                          confidence.intervals =0.95,study.design ='cohort')

 

List<-list(RAB24,DNAJB9,TOMM5,STAT3,PHF23,nomogram)

plot_decision_curve(List,curve.names= c('RAB24','DNAJB9','TOMM5','STAT3','PHF23','nomogram'),

                    cost.benefit.axis =T,col = c('red','blue','green','yellow','brown'),

                    confidence.intervals =FALSE,standardize = F,

                    legend.position = "topright")#legend.position = "none"

 

最后画出来的曲线是这样的

感觉不是很好看，但差不多就是这么个意思，大家明白流程就好。小伙伴们，你们画出来了吗。