Skip to main content

Starting a Research Project from the AI Assistant

Dotan Asselmann
Updated

You can start a research study straight from the AI Assistant. When you ask a research question in chat — the kind that compares groups, tests a correlation, or asks what predicts an outcome — the assistant hands you off to the Research Hub, Theator's dedicated research workspace, with your question already set up. The AI Assistant is where the study begins; the Research Hub is where you build the cohort, run the statistics, and write the paper.

Ask a research question in chat

Open the AI Assistant from the left sidebar and describe the study you want to run, in plain language. For example:

  • "Start a research project on ICG fluorescence imaging and anastomotic leak outcomes in my colorectal cases."
  • "Can you create a research hub to investigate the correlation between the Parkland scale and length of stay and complications?"
  • "I want to study whether my faster operative times hold after adjusting for case complexity — start a research project."

A good research prompt names a question you want tested across a group of cases — a comparison ("is X different between group A and group B?"), a correlation ("are X and Y related?"), or a predictor ("what predicts outcome Z?") — rather than a single number you could look up directly.

What the assistant does with a research question

When the assistant recognizes a research question, it does not try to run the whole study inside the chat. Depending on how concrete the study idea is, it offers one of two cards:

  • Continue in the Research Hub — for open-ended research questions or ideation (for example, "what could I study?" or "are X and Y correlated across my cases?"). Clicking the card opens the Research Hub home page with a summary of your question pre-filled in the message box, where you can edit it. Nothing is created until you press Send; sending the question starts your conversation with the research assistant, and where research projects are enabled it also creates a new project to hold it.
  • Create a research project — when you have a specific study in mind (for example, "Start a research project on ICG fluorescence and anastomotic leak outcomes"). The assistant proposes a named project; when you confirm it, the Research Hub creates the project, pins your research question in the workspace, and opens it with a starting brief already prepared in the chat.

From there you continue in the Research Hub, where the research assistant builds cohorts, runs multi-step statistical analyses, and drafts your abstract and article.

The AI Assistant can run individual statistical tasks in chat — a comparison, a correlation, or a regression against your case data. What it does not manage is the full study workflow: defining a persisted cohort, sequencing analyses across it, and drafting a manuscript. That end-to-end work belongs in the Research Hub.

Quick checks you can do in chat first

Before you commit to a project, you can use the AI Assistant to sanity-check the scope of your idea right in the chat. You can ask it to:

  • Count a cohort — for example, "How many colorectal cases with ICG did I do last year?"
  • Analyze your surgical data — trends, benchmarks, and comparisons across your cases.
  • Export a table — ask it to export the results to a spreadsheet to review the underlying cases.

These quick answers help you decide whether a research question is worth turning into a full project before you open the Research Hub.

Continuing in the Research Hub

Once the project is open, the Research Hub guides you through a study one approved step at a time:

Question → Cohort → Data points → Analysis → Article

You stay the researcher — you set the question, decide which cases count, and choose what belongs in the paper. For the full walkthrough of the workspace, its tabs, and how to export your manuscript, see Using the Research Hub.

Who can use it

If Research Hub does not appear in your left sidebar under Clinical Research, contact your Customer Success Manager to confirm access.

What's Next?

Comments

0 comments

Please sign in to leave a comment.