---
title: "Guarded evidence synthesis with imedR"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{Guarded evidence synthesis with imedR}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

The package separates exploration from strict quantitative synthesis.

```{r, eval=FALSE}
library(imedR)
db <- imed_data()
audit <- imed_validation_report(db)

x <- imed_analysis_set(db, comparability_class="CMP-W-VOL")
imed_check_comparability(x)
imed_meta_eligibility(x)

# Only if eligibility succeeds:
fit <- imed_meta_mean(x)
imed_results_table(fit)
imed_leave_one_out(x)
imed_forest(fit)
imed_funnel(fit)

# Moderator analysis is guarded by independent-effect count.
# imed_meta_regression(x, moderators=c("Year"))
```

A failed eligibility check is not an inconvenience to bypass: it is evidence
that the proposed synthesis requires redesign, stratification, additional
uncertainty information, or a dependence-aware model.