CDOoDocuments.StdDocumentDescDocuments.DocumentDescContainers.ViewDescViews.ViewDescStores.StoreDescDocuments.ModelDescContainers.ModelDescModels.ModelDescStores.ElemDesc TextViews.StdViewDescTextViews.ViewDescTextModels.StdModelDescTextModels.ModelDescQTextModels.AttributesDesc*<model{ #__________________________________________ OVARIAN ___________________________________________ #--------------------------------------------- ###1. Likelihoods #--------------------------------------------- # only head to head comparison for(i in 1:NO) { yO[i]~dnorm(deltaCrudeO[i],wO[i]) deltaCrudeO[i]<-deltaO[i]+0.5*betaO[i]*varO[i] } #--------------------------------------------- ###2.Meta-analysis model #--------------------------------------------- # only on head to head for(i in 1:NO){ deltaO[i]~dnorm(meanO[i],precisionO) betaO[i]~dnorm(Beta,precBeta) } #------------------------------------------------------------ ###3. Useful parametrisation of the means #--------------------------------------------------------------- for(i in 1:NO){ meanO[i] <- dO[bO[i]] - dO[tO[i]] } dO[8]<-0 for (k in 1:7) {dO[k] ~ dnorm(0,.0001) } dO[9] ~ dnorm(0,.0001) #All pairwise mean differences for (c in 1:(NTO-1)) { for (k in (c+1):NTO) { mean.diffO[c,k] <- dO[c] - dO[k] hrO[c,k]<-exp(mean.diffO[c,k]) }} for(k in 1:9){HRcomp8[k]<-exp(dO[k] - dO[8])} #Ranking for(k in 1:NTO){orderO[k]<-rank(dO[],k) most.effectiveO[k]<-equals(orderO[k],1) for(j in 1:NTO){ effectivenessO[k,j]<-equals(orderO[k],j) }} #deviances for(i in 1:NO) {DO[i]<-wO[i]*(yO[i]-deltaCrudeO[i])*(yO[i]-deltaCrudeO[i])} D.barO<-sum(DO[]) #Cumulative effectiveness for(k in 1:NTO){for(j in 1:NTO){ cumeffectivenessO[k,j]<-sum(effectivenessO[k,1:j])}} for(k in 1:NTO){CRASO[k]<-sum(cumeffectivenessO[k,1:(NTO-1)]) /(NTO-1)} ###### PRIORS################################## precisionO<-1/pow(sdO,2) precBeta<-1/pow(sdBeta,2) sdO~dnorm(0,1)I(0,) sdBeta~dnorm(0,1)I(0,) Beta~dnorm(0,0.0001) #_____________________________________ COLORECTAL ________________________________________________________ #--------------------------------------------- ###1. Likelihoods #--------------------------------------------- # only head to head comparison for(i in 1:NHtHC) { yC[i]~dnorm(deltaCrudeC[i],wC[i]) deltaCrudeC[i]<-deltaC[i]+0.5*betaC[i]*varC[i] } #More than two treatments yC[(NHtHC+1):NC]~dmnorm(deltaCrudeC[(NHtHC+1):NC],PRECC[,]) for(i in (NHtHC+1):NC){ deltaCrudeC[i]<-deltaC[i]+0.5*betaC[i]*varC[i] } #--------------------------------------------- ###2.Meta-analysis model #--------------------------------------------- # only on head to head for(i in 1:NHtHC){ deltaC[i]~dnorm(meanC[i],precisionC ) } #More than two treatments deltaC[(NHtHC+1):NC]~dmnorm(meanC[(NHtHC+1):NC],KC[,]) for(i in 1:(NC-NHtHC)){for(j in 1:(NC-NHtHC)){ KC[i,j]<-precisionC*HC[i,j]}} # H is data, assuming common heterogeneity #------------------------------------------------------------ ###3. Useful parametrisation of the means #--------------------------------------------------------------- for(i in 1:NC){ meanC[i] <- dC[tC[i]] - dC[bC[i]] } dC[2]<-0 dC[1] ~ dnorm(0,.0001) for (k in 3:NTC) {dC[k] ~ dnorm(0,.0001) } for(i in 1:NC){ betaC[i]~dnorm(Beta,precBeta) } #All pairwise mean differences for (c in 1:(NTC-1)) { for (k in (c+1):NTC) { mean.diffC[c,k] <- dC[k] - dC[c] HRC[c,k] <- exp(mean.diffC[c,k] ) }} for(k in 1:NTC){HRcomp2[k]<-exp(dC[k]-dC[2])} #Ranking for(k in 1:NTC){ orderC[k]<-rank(dC[],k) most.effectiveC[k]<-equals(orderC[k],1) for(j in 1:NTC){ effectivenessC[k,j]<-equals(orderC[k],j)} } #Cumulative effectiveness for(k in 1:NTC){for(j in 1:NTC){ cumeffectivenessC[k,j]<-sum(effectivenessC[k,1:j])}} for(k in 1:NTC){CRASC[k]<-sum(cumeffectivenessC[k,1:(NTC-1)]) /(NTC-1)} #deviances for(i in 1:NHtHC) {DC[i]<-wC[i]*(yC[i]-deltaCrudeC[i])*(yC[i]-deltaCrudeC[i])} #More than two treatments for(i in (NHtHC+1):NC){DC[i]<-(yC[i]-deltaCrudeC[i])*(yC[i]-deltaCrudeC[i])*PRECC[i-NHtHC,i-NHtHC]} D.barC<-sum(DC[]) ###### PRIORS################################## precisionC<-1/pow(sdC,2) sdC~dnorm(0,1)I(0,) #_____________________________________________________________________BREAST #--------------------------------------------- ###1. Likelihoods #--------------------------------------------- # only head to head comparison for(i in 1:NHtHB) { yB[i]~dnorm(deltaCrudeB[i],wB[i]) } #More than two treatments yB[(NHtHB+1):(NHtHB+N3armsB)]~dmnorm(deltaCrudeB[(NHtHB+1):(NHtHB+N3armsB)],PREC1B[,]) #More than three treatments yB[(NHtHB+N3armsB+1):NB]~dmnorm(deltaCrudeB[(NHtHB+N3armsB+1):NB],PREC2B[,]) for(i in 1:NB){ deltaCrudeB[i]<-deltaB[i]+0.5*betaB[i]*varB[i] } #--------------------------------------------- ###2.Meta-analysis model #--------------------------------------------- # only on head to head for(i in 1:NHtHB){ deltaB[i]~dnorm(meanB[i],precisionB) } #More than two treatments deltaB[(NHtHB+1):(NHtHB+N3armsB)]~dmnorm(meanB[(NHtHB+1):(NHtHB+N3armsB)],K1B[,]) for(i in 1:N3armsB){for(j in 1:N3armsB){ K1B[i,j]<-precisionB*H1B[i,j]}} # H1 is data, assuming common heterogeneity #More than three treatments deltaB[(NHtHB+N3armsB+1):NB]~dmnorm(meanB[(NHtHB+N3armsB+1):NB],K2B[,]) for(i in 1:(NB-NHtHB-N3armsB)){for(j in 1:(NB-NHtHB-N3armsB)){ K2B[i,j]<-precisionB*H2B[i,j]}} # H2 is data, assuming common heterogeneity #------------------------------------------------------------ ###3. Useful parametrisation of the means #--------------------------------------------------------------- for(i in 1:NB){ meanB[i] <- dB[bB[i]] -dB[tB[i]] betaB[i]~dnorm(Beta,precBeta)} for (k in 1:17) {dB[k] ~ dnorm(0,.0001) } dB[18]<-0 for (k in 19:NTB) {dB[k] ~ dnorm(0,.0001) } #All pairwise mean differences for (c in 1:(NTB-1)) { for (k in (c+1):NTB) { mean.diffB[c,k] <- dB[c] - dB[k] hrB[c,k] <- exp(mean.diffB[c,k] )}} for (c in 1:NTB) { HRcomp18B[c] <- exp(dB[c]-dB[18]) } #Ranking for(k in 1:NTB){ orderB[k]<-rank(dB[],k) most.effectiveB[k]<-equals(orderB[k],1) for(j in 1:NTB){ effectivenessB[k,j]<-equals(orderB[k],j)} } for(k in 1:NTB){for(j in 1:NTB){ cumeffectivenessB[k,j]<-sum(effectivenessB[k,1:j])}} for(k in 1:NTB){CRASB[k]<-sum(cumeffectivenessB[k,1:(NTB-1)]) /(NTB-1)} for(i in 1:NHtHB) {DB[i]<-wB[i]*(yB[i]-deltaCrudeB[i])*(yB[i]-deltaCrudeB[i])} #3treatments for(i in (NHtHB+1):(NHtHB+N3armsB)){ DB[i]<-(yB[i]-deltaCrudeB[i])*(yB[i]-deltaCrudeB[i])*PREC1B[i-NHtHB,i-NHtHB]} #3treatments for(i in (NHtHB+N3armsB+1):NB){ DB[i]<-(yB[i]-deltaCrudeB[i])*(yB[i]-deltaCrudeB[i])*PREC2B[i-NHtHB-N3armsB,i-NHtHB-N3armsB]} D.barB<-sum(DB[]) ###### PRIORS################################## precisionB<-1/pow(sdB,2) sdB~dnorm(0,1)I(0,) } TextControllers.StdCtrlDescTextControllers.ControllerDescContainers.ControllerDescControllers.ControllerDesc TextRulers.StdRulerDescTextRulers.RulerDescTextRulers.StdStyleDescTextRulers.StyleDescZTextRulers.AttributesDesc$ ZGo * ,[ @Documents.ControllerDesc Ws,! [h$