Understanding Review Phases

Reviews Happen in a Single Conversation

A review in Scholara is one continuous conversation with the AI assistant. You don't navigate between separate forms, pages, or tabs — the assistant guides you through each step in order, asks for what it needs, and produces artifacts (your PICO summary, search strategy, screening results, analyses) that appear in a side panel as you go.

The chat is the review. Anything you'd traditionally do across multiple screens — refining a research question, building search strings, screening abstracts, running a meta-analysis — happens by talking to the assistant.

The Phase Indicator

At the top of every review you'll see a phase indicator with eight steps: PICO, Confirm, Protocol, Strategy, Search, Abstract, Fulltext, and Analysis. The current step is highlighted; completed steps are filled in. The indicator is your map — it shows where you are in the review and what's coming next.

The Eight Phases

1. PICO

The assistant extracts the core elements of your research question — Population, Intervention, Comparison, Outcomes (or Population, Concept, Context for scoping reviews). If your question is ambiguous, the assistant will ask clarifying questions before proceeding.

2. Confirm

Review the extracted PICO summary, which appears as an artifact in the side drawer. Edit any field by telling the assistant what to change — for example, "narrow the population to adults aged 18–65" or "add placebo as the comparator."

3. Protocol

The assistant builds your inclusion and exclusion criteria, study design preferences, and other protocol fields based on your confirmed PICO. Edits happen conversationally — ask for changes and the protocol artifact updates in the side drawer in real time.

4. Strategy

Generate Boolean search strings translated for each database. The assistant produces queries optimised for PubMed, OpenAlex, Europe PMC, and ClinicalTrials.gov, and lets you choose which databases to run.

5. Search

Scholara executes your searches across the selected databases, deduplicates results across sources, and reports the article count. You don't need to do anything during this phase — progress streams into the chat as it runs.

6. Abstract

AI screens every abstract against your inclusion criteria. The screening drawer opens automatically and shows each article's decision (Include / Exclude / Maybe) with the reasoning. You can override any decision, ask the assistant questions about borderline cases, or bulk-exclude flagged categories.

7. Fulltext

For studies that pass abstract screening, upload PDFs (or let Scholara fetch open-access versions automatically) and the assistant screens the full text against your criteria. Override decisions and add structured exclusion reasons as needed.

8. Analysis

Use the chat to extract structured data from included studies, generate characteristics tables and outcome plots, run meta-analyses, and create your PRISMA flow diagram. Each output appears as an artifact you can open, copy, or export.

Moving Between Phases

Phases progress in order, but you can revisit earlier ones at any time. To edit your PICO after starting screening, scroll back in the chat or just ask the assistant — for example, "go back to the protocol and add quasi-experimental designs." Scholara tracks artifact versions so you always know which results are based on the current protocol.

What You'll See on Screen