Your First Comparison
Start with a landscape question and work toward a source-linked comparator set.
On this page
Before you start
Open Atlas from your PhaseFolio product directory. You need access to the product and, for a model-assisted research run, an available run allowance.
Start with a public-source research question. Do not include patient information, confidential trial data, credentials, or material you are not authorized to submit.
Define the landscape
Choose New comparison and enter your landscape question. Include the indication and any population, treatment setting, intervention, or study-design boundaries that matter to the comparison.
For example: “Which studies report on this indication in a previously treated population?” This is a question template, not a claim that any particular study will be found.
Review the proposed search terms before continuing. A broad question and a precise registry expression are different things; make sure the expression captures the scope you intended.
Check sources and prepare the comparison
Inspect the source boundary. Include only available sources that are relevant to your question, then continue with the search/preparation action shown in the product.
Source availability, inclusion in your question, and successful execution are separate states. A source that is unavailable or was not searched cannot support a conclusion that no evidence exists.
Inspect and select comparators
Review the returned candidates. Check the underlying study record, population, design, and other comparison dimensions before selecting a record. A surfaced candidate is a lead for inspection, not a declaration that two studies are interchangeable.
Build or open the evidence packet for your selected records. Check coverage and missing information before proceeding. Read Reading the evidence for details.
Run research when you need synthesis
The model-assisted research action analyzes eligible evidence for the selected records. It is a separate step from retrieving and preparing source evidence, and it can use your run allowance.
Review the selection and scope before starting. Follow the progress and terminal status shown in the product. A pending, failed, or cancelled run is not a completed research answer.
Return to the work
Where the saved-landscape interface is enabled, reopen your landscape from recent comparisons. Use Landscape, Evidence packets, Research, and Activity to inspect its different views.
The history shown in the product determines what was saved. Read Working with landscapes for revision behavior and recovery.