First session#
Use a fictional scenario and a lab you control. OpenHealthSim runs a local server with a browser interface. The website explains the workflow; your lab application keeps the scenario, session record and configuration.
This guide describes the tested Windows/Chrome and Linux MVP. Actual Safari/iPad, macOS, external services, clinical approval and release qualification remain separate. Human review is part of the workflow.
Before you begin#
- Open the lab URL supplied by its administrator and sign in with your own account. There is no universal default username or password.
- Ask the administrator to enable the authoring, session and debrief features and configure a model provider if you want AI extraction. Keys belong in the host’s key settings, never in a scenario or a public demo.
- Choose Scenarios → Scenario authoring for source work and Simulation → Sessions for running. Page availability follows your account’s permissions.
- For a tour without setup, open the demonstration or watch the workflow. Demo records are fictional.
Import a Word source#
Download the fictional Word practice source, or read its plain-text version. It contains a small manual course and simulated observation table for software practice, with clinical scoring deliberately omitted.
Open Scenario authoring, select a supported .docx source and import it. Open the saved source and check the parsed paragraphs and table cells against the original file before extraction. Detailed import instructions explain the two extraction paths and how to recover when a proposal is incomplete.
Start AI course analysis or the available Extract scenario path. Select the configured provider and inspect its disclosure before sending source material. A model proposal is an unapproved draft.
Review the draft#
Compare patient details, learning outcomes, states, transitions, observations, units, dialogue and end criteria with the source. Correct mistakes, resolve the findings the app requires and record the requested reviews. Inspect scores and their scale separately. Approve only the saved version you intend to use. A machine check or successful save does not approve clinical content.
If approval remains blocked, read the exact finding rather than skipping it. Unverified correction drafts can have a limited rehearsal path with scores withheld; that is distinct from an approved learner session.
Prepare and create#
In Sessions, choose Prepare a session. Pick the approved scenario version, room, purpose, facilitator details, learner roles and available stations. Inspect routing and compatibility. Capture is optional; it requires consent and explicit browser permission. Choose Create session when the setup is ready.
Creating a session saves its setup. It does not start the scenario clock. Keep the created session’s console link or return to it from the Sessions list.
Start, pause and return#
Open the created session’s Live console, verify its identity, then choose Start session. Use the displayed actions and manual transition controls to progress the scenario. Confirm observations on the intended simulated patient and add a marker for a useful debrief moment.
Choose Pause session to pause scenario progression. Return through the same session link, verify its current state and choose Resume session. Browser navigation is not a stop command; the session guide explains re-entry and recording limits.
End and debrief#
Choose End session and confirm it when the run is finished. Open Debrief for that exact session. Review the factual timeline and any suggested narrative, then save your edits. Select the facilitator or learner view before exporting JSON or PDF. Debrief instructions explain review, export and missing evidence.
Where to go next#
- Authoring — source review, approval and extraction recovery.
- Sessions — setup, manual controls, learners and re-entry.
- Interfaces — simulated devices, paired screens and routing.
- Lab administration — accounts, providers and data management.
- Community plugins — inspect, install inactive and explicitly enable.
- Developer guide — a complete downloadable author kit and coding-agent skill.