ML4H 2026 submission 396: preprint v4

Feedback interface mock-up

Version 4 is version 3, which has the corrected Figure 1, plus a design mock-up of the learner-facing feedback loop for OSCE-style stations. The mock-up shows live cues in practice mode, a debrief that sorts skills by evidence, spaced drills on weak skills, a result screen and progress over time. It is one figure in the main text (Figure 2) and an appendix with four more screens.

All learner data in the mock-up are synthetic, for one invented learner. Case titles and the nine feedback domains come from the system; attempts, scores, dates and timings do not. No learner outcomes are reported.

Interactive mock-up
Paper v4 (PDF)
Main LaTeX file
Full LaTeX source (zip), which builds with pdflatex and BibTeX as described in its README

Mock-up of a live goals-of-care OSCE station with a cue tracker, a prompt about an unexplored cue, the transcript and a station checklist. — A live station in practice mode, from Figure 2. Synthetic data.

Mock-up of practice score over time for one specialty track, with hinted, unassisted and delayed re-test attempts, a rolling mean crossing a pass line, and a grid of nine feedback domains for each attempt moving from red and yellow toward green. — Progress over time in one specialty track, with the nine feedback domains for each attempt, from Appendix A. Synthetic data for one invented learner; the pass line is a placeholder.