prisma2020-flow-diagram

# PRISMA 2020 Flow Diagram ## When to use Invoke when the user asks for a PRISMA flow diagram, study-selection flowchart, or screening flowchart for a systematic review / meta-analysis, or when the companion skill `doing-meta-analysis-r` needs a reporting figure. ## Required R packages ```r install.packages(c("PRISMA2020", "DiagrammeR", "DiagrammeRsvg", "rsvg")) ``` ## Input contract (CSV) Required columns (PRISMA 2020 two-arm layout): ``` previous_studies, previous_reports, register_results, database_results, website_results, organisation_results, citation_results, duplicates_removed, excluded_automation, excluded_other, records_screened, records_excluded, reports_sought_retrieval, reports_not_retrieved, reports_assessed, reports_excluded_reason1, reports_excluded_reason2, reports_excluded_reason3, new_studies, new_reports, total_studies, total_reports ``` Three-arm (Previous + New + Other methods) adds the `other_*` column family. ## Canonical pipeline ```r library(PRISMA2020) data <- read.csv("prisma_counts.csv") plot <- PRISMA_flowdiagram( data, interactive = FALSE, previous = TRUE, other = FALSE, fontsize = 12, font = "Helvetica", title = NA ) PRISMA_save(plot, filename = "prisma_flow.pdf", filetype = "PDF") ``` ## IPAI integration - **Primary use case**: PrismaLab R&D systematic reviews, JBLMGH Neurology submissions — CAVE score PSE, toxoplasmosis, HIV-TB co-infection manuscripts. - **Artifact storage**: register counts in `ops.artifacts` with `type='prisma_counts'`; upload PDF to Azure Blob `stipaidevlake/research/<study_id>/figures/prisma_flow.pdf`. - **Companion skills**: `doing-meta-analysis-r` (pooling + diagnostics), `meta-analysis-pipeline` (orchestration). ## Safety - Never fabricate counts. Require an explicit CSV from the user or from upstream search-results extraction. - Row totals must reconcile (identified = screened + duplicates_removed). Flag mismatches — do not silently correct.