bland

Bland-AltmanPlots（⼀致性评价）在python中的实现

假设有reader1和reader2，分别对⼀定数量病⼈的某⼀影像指标进⾏评分，现在想看⼀下这两位研究者评分的⼀致性，绘制Bland-Altman图是

⼀种较为直观、简单的⽅式。python代码实现⽅法如下：

⾸先读⼊数据

folderPath="/Urs/.../ICC/features4ICC/"

data1=_excel((folderPath,"reader1_"))

data2=_excel((folderPath,"reader2_"))

(0,"reader",([0]))

(0,"reader",([0])*2)

(0,"target",range([0]))

(0,"target",range([0]))

data=([data1,data2])

print(data)

input_

Method1

importnumpyasnp

importpingouinaspg

ax=_blandaltman(data1['original_shape_Elongation'],data2['original_shape_Elongation'])

method_

Method2

#pipinstallpyCompare#forthefirsttime

importpyCompare

ltman(data1['original_shape_Elongation'],data2['original_shape_Elongation'],

percentage=Fal,

title='Bland-AltmanPlot2',

limitOfAgreement=1.96)

method_

ltman(data1['original_shape_Elongation'],data2['original_shape_Elongation'],

savePath='',

figureFormat='tiff')

Method3

asplt

importnumpyasnp

defbland_altman_plot(data1,data2,*args,**kwargs):

data1=y(data1)

data2=y(data2)

mean=([data1,data2],axis=0)

diff=data1-data2#Differencebetweendata1anddata2

md=(diff)#Meanofthedifference

sd=(diff,axis=0)#Standarddeviationofthedifference

r(mean,diff,*args,**kwargs)

e(md,color='gray',linestyle='-')

e(md+1.96*sd,color='gray',linestyle='--')

e(md-1.96*sd,color='gray',linestyle='--')

importrandom

bland_altman_plot(data1['original_shape_Elongation'],data2['original_shape_Elongation'])

('Bland-AltmanPlot3')

()

method_

参考资料：

Pingouinofficialdocumentation

Why&HowtoutheBland-AltmanplotforA/Btesting|Python+code

Bland-Altmanplotwithconfidenceintervalboundaryinpython