Lowest-friction build in the set. Stage 1 is a single-best-answer with hints and a heat-map fallback. Stage 2 is a single-select identification of the one harmful action among five — three attempts and a bonus fourth after the hint modal. Fastest completion; smallest cognitive footprint. Where the teaching point is binary.

Direct extension of Variant 01 within the v5 simple family. Stage 1 is the same single-best-answer with hints and heat-map fallback. Stages 2 and 3 are two consecutive single-select identifications — each isolates one harmful action from a five-option panel, with three attempts plus a bonus fourth after the hint modal. For scenarios where two independent must-never-do tactics live inside the same management pathway and need to be recognized separately. Lowest cognitive load in the three-stage family.

Same Stage 1 as Variant 01. Stage 2 expands into a multi-select: pick the four correct actions out of five while avoiding the one harmful option, with a misclassification penalty. Compact two-stage build with a richer Stage 2 than 01.

The universal two-stage chassis in a mobile-first footprint. Stage 1 picks three findings out of six; Stage 2 picks four interpretations out of eight, with a misclassification penalty. Anti-auto-pass shuffler on every attempt. Teal-marine palette by default; per-buyer CTA labels pre-populated. Three to four minutes per case.

Same universal architecture as Variant 03, scaled to a wide finding panel. Stage 1 picks five findings out of twelve; Stage 2 picks five interpretations out of ten. Maximum reasoning surface in the two-stage family. Warm-gold palette by default. For radiology, dermatology mix, or ECG cases with a rich finding panel.

Stage 0 single-best-answer gate, then a mid-stage teaching narrative with an additional sponsor click-through, then a True/False battery — lower friction than the multi-select builds. All variants include a pre-score sponsor gate by default; this build adds a second click-through mid-flow. Luxury pharma investor palette.

Same chassis as Variant 05, different assessment register. Stage 0 single-best-answer gate, mid-stage teaching narrative with sponsor click-through, then a 2-option MCQ battery — each step presents one correct and one incorrect answer. Marginally higher cognitive load than T/F. Same second mid-flow click-through and luxury pharma palette as Variant 05.

The short-form universal architecture plus a Stage 0 single-best-answer diagnostic gate. The physician commits to a working diagnosis before being released into the findings work. Three attempts, progressive hint disclosure, blocked options across attempts, heat-map mode on attempt three with a Think-again modal. The build where the teaching point is the diagnostic hypothesis itself — anchoring, availability, premature closure.

The reference build for new cases. The three-stage architecture of Variant 07 plus four v6 analytic layers stacked on top: per-pick confidence rating (high · medium · low), passive timing track, a final 2×3 calibration grid that surfaces over-confidence, and a per-card clinical-impact modal with optional CTA. Warm-gold palette by default.