patentsStata软件在诊断性研究的meta分析中的命令
在诊断性研究的meta分析中可以计算合并阳性似然比、合并阴性似然比、诊断OR值、ROC值、SROC曲线、HSROC-bivariate meta-analysis等。
Stata进行诊断研究meta分析时的起始命令：
*Variable codes: tp=true positives; fp=fal positives; tn=true negatives;
fn=fal negatives
*add .5 to all zero cells
gen zero=0
replace zero=1 if tp==0|fp==0|fn==0|tn==0
replace tp=tp+.5 if zero==1
replace fp=fp+.5 if zero==1
replace fn=fn+.5 if zero==1
replace tn=tn+.5 if zero==1
gen tpr= tp/(tp+fn)
gen fpr=fp/(fp+tn)
volatilegen logittpr=ln(tp/fn)
gen logitfpr=ln(fp/tn)
gr7 tpr fpr, s(O) noaxis ysize(6) xsize(6) xline(0(.1)1) yline(0(.1)1) tlab(0(.1)1) xlab(0(.1)1) ylab(0(.1)1) t1(1-Specificity) l1(Sensitivity) b2(1-Specificity) b1(ROC Plot of Sensitivity vs Specificity)
英文在线翻译成中文gr7 logittpr logitfpr
spearman logittpr logitfpr2017年注册会计师成绩查询
1.1 合并阳性似然比
命令：metan tp fn fp tn, rr random nowt sortby(author) xlab(.01,1,100) label(namevar=author, yearvar=pubyear) t1(Summary LR+, Random Effects)
2.2 合并阴性似然比
命令：metan fn tp tn fp, rr random nowt sortby(author) xlab(.01,1,100) label(namevar=author, yearvar=pubyear) t1(Summary LR-, Random Effects)
2.3 合并诊断OR值
命令：metan tp fn fp tn, or random nowt sortby(author) xlab(.01,1,100) label(namevar=author, yearvar=pubyear) t1(Summary Diagnostic Odds Ratio, Random Effects)
2.4 ROC值
命令：gr7 tpr fpr, s(O) noaxis ysize(6) xsize(6) xline(0(.1)1) yline(0(.1)1) tlab(0(.1)1) xlab(0(.1)1) ylab(0(.1)1) t1(1-Specificity) l1(Sensitivity) b2(1-Specificity) b1(ROC Plot of Sensitivity vs Specificity)
2.5 SROC曲线
命令：
gen sum= logittpr+ logitfpr
gen diff= logittpr- logitfpr
regress diff sum
predict yhat江西蓝天学院是几本
gr7 diff yhat sum, ylab(3,4,5,6,7,8) xlab(-4,-3,-2,-1,0,1,2) c(.l) s(oi)
gen t=1/(1+(1/(exp(_cons/1-_b)*(fpr/spec)^1+_b/1-_b)))
(constant and b are derived from the above regression model)
*plot SROC curve (generic)
gr7 t fpr, ysize(6) xsize(6) noaxis xline(0(.1)1) yline(0(.1)1) tlab(0(.1)1) xlab(0(.1)1)
ylab(0(.1)1) s(Oi) c(.s) l1(Sensitivity) b2(1-Specificity) ti(Summary ROC Curve) key1(" ")
key2(" ")
advertiment是什么意思
2.6 HSROC-bivariate meta-analysismdn
命令：metandi tp fp fn tn, plot （基于SROC命令）
2.7 发表偏倚
命令：
gen or=(tp*tn)/(fp*fn)
gen lnor=ln(or)
begininggen lnor=(1/tp)+(1/fp)+(1/fn)+(1/tn)
*Begg and Egger test for publication bias with Begg's funnel plot: metabias lnor lnor, graph(begg)
confliction
*Begg and Egger tests for subgroups (eg. Covariate=1)
德文翻译中文metabias lnor lnor if covariate==1, graph(begg)
