以下PPT内容由群友 重庆森林(亦简书ID 北欧森林)提供:
![](https://7.dusays.com/2021/01/15/879de4df7ca33.jpg)
![](https://7.dusays.com/2021/01/15/c4e0aad9a387a.jpg)
![](https://7.dusays.com/2021/01/15/6b195e9358706.jpg)
![](https://7.dusays.com/2021/01/15/c3f1f0e8c3b76.jpg)
![](https://7.dusays.com/2021/01/15/233614b31435a.jpg)
![](https://7.dusays.com/2021/01/15/485a400fa1e22.jpg)
![](https://7.dusays.com/2021/01/15/f7f6e2a71f6e4.jpg)
![](https://7.dusays.com/2021/01/15/901333acdc405.jpg)
![](https://7.dusays.com/2021/01/15/a854edafffaeb.jpg)
![](https://7.dusays.com/2021/01/15/b49346ea810a9.jpg)
![](https://7.dusays.com/2021/01/15/70e19146f97f1.jpg)
![](https://7.dusays.com/2021/01/15/71197dabcfd23.jpg)
![](https://7.dusays.com/2021/01/15/e947fc0454e18.jpg)
![](https://7.dusays.com/2021/01/15/7990b0476ff2b.jpg)
![](https://7.dusays.com/2021/01/15/8c6407ba506bf.jpg)
![](https://7.dusays.com/2021/01/15/e9b5e958c9cd0.jpg)
![](https://7.dusays.com/2021/01/15/8ca046f870a36.jpg)
![](https://7.dusays.com/2021/01/15/253c785190689.jpg)
![](https://7.dusays.com/2021/01/15/360fbc2795a75.jpg)
![](https://7.dusays.com/2021/01/15/97d34ed1647c8.jpg)
![](https://7.dusays.com/2021/01/15/ae65dc452e7a1.jpg)
![](https://7.dusays.com/2021/01/15/7e6835241a540.jpg)
![](https://7.dusays.com/2021/01/15/2eb92f467d024.jpg)
![](https://7.dusays.com/2021/01/15/e7fa4f807eb4b.jpg)
![](https://7.dusays.com/2021/01/15/0cfc8fd663e16.jpg)
![](https://7.dusays.com/2021/01/15/da79f6733e74f.jpg)
![](https://7.dusays.com/2021/01/15/69801f0022450.jpg)
![](https://7.dusays.com/2021/01/15/05ea996848391.jpg)
文献报告中提到的主要文章:
扩展阅读(指南解读):
ICC,Bland-Altman plot的实现方法
library(readr)
library(irr)
fpath <- choose.files()
feature_1 <- read_csv(fpath)
fpath <- choose.files()
feature_3 <- read_csv(fpath)
len <- N# N是指标的数量
icc_val<-vector(length=Len)
thr <- 0.75
selected <- feature_1[feature_1$ID %in% feature_3$ID,]#获取id相同的样本
for (i in 2:len){#len是特征的数量,第一列是ID,从第二列开始进行指标的比较
ratings <- cbind(selected[,i],feature_3[,i])
icc <- icc(ratings, model = "twoway",
type = "agreement",
unit = "single", r0 = 0, conf.level = 0.95)
icc_val[i] <- icc$value
}
Index <- which(icc_val > thr)
dim(icc_val)=c(1,len)
write.csv( icc_val,file = "output.csv",row.names = F)
# icc处的参数:type:agreement 是不同评分者的评价是否一致 consistency不同评分者的评分是否存在相关性
# unit:single 是单个评分 average是取平均的评分
# source: https://blog.csdn.net/sinat_34054843/article/details/106005893
有个小尾巴,需要群里的高手协助解决:
python代码批量计算ICC的循环是我自己写的,不够优美,还希望高手完成:
1、批量提取某一个ICC、pvalue和置信区间等,并且统计比如 每个区间的频率;
2、R代码的循环我也不会写,我已经搜到网上的现成代码,但是那个代码有bug,我没有调好,到时候大神给改一下吧;
原作者 @NordicForrest