From PubMed search to forest plot in under a day

1. Start with a focused question

Use PICO or PEO to define eligibility criteria before searching so screening decisions are faster and more consistent. Specify your population, intervention (or exposure), comparator, and primary outcomes up front — this becomes the backbone of your inclusion criteria.

2. Search and import results

You can add studies individually by NCT ID or PMID, batch import from a search export, or use the Search Agent to automatically build and execute optimized PubMed and ClinicalTrials.gov queries from your PICO/PEO criteria. The agent screens scouted results against your criteria in real time, measuring query precision before committing — so only relevant studies enter your library.

3. Screen with explicit rationales

Capture exclusion reasons during screening to reduce rework during PRISMA reporting. AI-assisted screening suggests include/exclude decisions with confidence scores — always review before accepting.

4. Batch extract endpoint fields

Use Auto‑Extract All Outcomes to process all eligible studies in parallel. The system handles document acquisition, companion paper resolution, and multi‑agent extraction — with confidence scoring and provenance tracking on every value. Keep extraction schemas tight to avoid unnecessary variables; the math engine validates consistency, detects arm swaps, and derives missing quantities automatically.

5. Run model and sanity checks

On the Analysis page, the Stats Agent shows a data summary with your study count, outcomes, and meta-readiness at a glance. Use starter prompts or type a natural-language request to run your first model. The agent selects the right effect measure, calls R/metafor in the browser, and returns a Run Summary card with a colour-coded heterogeneity badge (Low / Moderate / High).

After a model runs, inspect the forest plot for study-level effects, check I² for heterogeneity, review funnel plot asymmetry for publication bias, and consider subgroup analyses for pre-specified hypotheses. The Run Summary card shows outcome, method, k, and heterogeneity at a glance; the Model Fit Summary card shows the full pooled estimate, 95% CI, I², τ², and p-value.