Workshop — the 90-minute run sheet
The facilitator pack for the guided 90-minute collaborator session (first run: 18 September 2026, Singapore): the clock, the checks, and the fallbacks. Learners follow the course page; everything runs on synthetic or public material.
This is a teaching aid, not compliance or legal advice. Review status, honestly stated: no clinician has reviewed this course yet. A review is scheduled. The session presents the course as clinician-review-scheduled, never as clinician-reviewed. Ground rule to state out loud at the start: the workshop accounts are consumer accounts and are not covered for patient data — no real, remembered, or lightly-edited patient detail enters any chat today.
The six blocks
0–10 · Setup, goals, and the data boundary
- Do: access check per setup & help; pair anyone without a working account. State the session's goal (safe, checkable use — not tool marketing) and the ground rule above.
- Materials: setup page; printed one-pagers on the tables.
- Check: every seat has either a working account or a pair; the boundary was said aloud, not assumed.
- Fallback: account trouble → the pair keeps working; the learner follows on the printed sheet and swaps roles at block 4.
10–25 · Privacy and data handling
- Do: the Retention exercise live — every learner finds their own toggle; then walk the consumer-versus-commercial scope split using the current receipted findings on the Retention module.
- Check: the exercise's facilitator check (the feedback-button question), asked to the room.
- Fallback: no account → the learner reads the toggle path from the exercise card and answers that module's self-check quiz instead — it needs no account either.
25–45 · Useful prompts, sources, and uncertainty
- Do: a worked synthetic example (the folder exercise's three files work well); then the Cross-cutting exercise — each learner traces one ✓ claim to its receipt and to the live vendor page.
- Check: at least two learners report what the live page said in its own words; a "moved" report is treated as a win, out loud.
- Fallback: projector-only if the room's connectivity fails — the facilitator traces one claim visibly; learners follow on the receipts page later.
45–70 · Role-based practice in pairs
- Do: pairs pick a role — clinician, researcher, advocate, or technical collaborator — and run the matching drill: the HIPAA boundary drill (clinician/advocate roles) or the Skills exercise (researcher/technical roles). No real patient case work in any role.
- Check: each pair states one thing that failed and what they changed — the Skills exercise's step 3, or boundary card 5's reasoning.
- Fallback: odd numbers → one trio; time pressure → drop to one drill per pair, done well, rather than both done shallowly.
70–85 · A bounded Discovery Lab thesis
- Do: each learner (or pair) writes a one-paragraph thesis: the problem, a useful result in one month, the data it needs, its risks, who would review it, and what evidence of success looks like — the verification habit is that last column.
- Check: the thesis names a reviewer and an evidence standard; "it would be cool" fails the check.
- Boundary, stated as written: the AI4U Discovery Lab is a proposed program; its application is not open, writing a thesis is not an application, and course completion never implies admission or funding.
85–90 · Recap and the self-paced route
- Do: one traced claim from block 3 retold to the room; point at the full course for self-paced completion, the completion checklist, and the open-questions register for what remains honestly unsettled.
- Check: every learner leaves knowing where the course lives and holding a printed one-pager.
Facilitator notes
- Timekeeping: blocks 2 and 4 overrun most; the recovery is shortening block 3's second trace, never skipping block 5's boundary statement.
- Questions you cannot answer: route to the register in front of the room — modelling "I don't know, and here is where unanswered questions live" is course content, not a failure.
- After the session: collect what failed, what confused, and any "moved" source reports — they route into the curriculum's own review loop.