reporting_on_model.Rmd

## Prior: `r names(fits)[index]`

<div class="fold s">
```{r, message=FALSE}
index <- 1

fit <- fits[[index]]
fit %>% summary
plot(fit, pars=effects, plotfun = "areas", transformations="exp")+
  labs(x="Hazard Ratio")+scale_y_discrete(labels=c("E1 vs Control  ","E2 vs Control  "))+
  scale_x_log10(breaks=c(seq(0.4,1,by=0.1),seq(1,3,by=0.2)))+geom_vline(xintercept=c(0.7,1), col="grey")

#plot(fit, pars=effects, "hist")
#plot(fit, pars=effects, "violin")
#plot(fit, pars=effects, "dens_overlay")
plot(fit, pars=effects, "trace")
plot(fit, pars=effects, "acf_bar")

posterior <- as.array(fit)
#dimnames(posterior)

prob_interval <- function(X, interval){
  Y <- interval[1]< X & X <interval[2]
  mean(Y)
}
efficacy <- apply(posterior, 3, FUN=prob_interval, interval=c(-Inf, 0))[effects]
moderate_efficacy <- apply(posterior, 3, FUN=prob_interval, interval=c(-Inf, log(0.8)))[effects]
similarity <- apply(posterior, 3, FUN=prob_interval, interval= log(c(0.8,1.25)))[effects]
harm <- apply(posterior, 3, FUN=prob_interval, interval= c(0,Inf))[effects]
```



Reporting as per section 15.2 of the protocol to give the posterior probability of the HR lieing in various ranges of interest.

```{r}

X <- rbind(efficacy, moderate_efficacy, similarity, harm) %>% as.data.frame
X <- cbind(HR=c("<1","<0.8","0.8-1.25",">1"),X)
X %>% kable(row.names=FALSE, col.names=c("HR","E1 vs Control  ", "E2 vs Control  "))
```
</div>