Skip to main content

How it works

How our questions are written & reviewed

An algorithm is only as good as the questions beneath it — so content comes first. Every item is written by a clinician, verified by a second, traced to the standard references, and re-examined as residents answer it. This is that process, in full — nothing hidden.

In one line: written by an ophthalmologist, verified by a second, sourced to the BCSC and Kanski, and corrected the moment a reader flags a flaw.

How a question is made
Drafted by anophthalmology-trained…Independent clinicalreview for accuracyEvery fact checkedagainst BCSC & KanskiPublished to thequestion bankDifficulty calibrated onreal resident answers

The pipeline, end to end. Each stage is explained below.

1

Authored by clinicians — never crowdsourced

Every question is written by an ophthalmology-trained author — the stem, the options, and the explanation of why each distractor is wrong. They follow the one-best-answer item-writing standards used for medical licensing exams, so a question tests clinical reasoning — not recall trivia or test-taking tricks.

2

Verified by a second clinician

No question goes live on one person's judgment. A second clinician re-checks the stem, the keyed answer, and the explanation — confirming the medicine is correct, current, and unambiguous, and that exactly one option is defensibly best. Anything contestable is revised or cut before it reaches you.

3

Traced to the standard references

Every clinical claim is anchored to an authoritative source — so an answer is defensible, not merely plausible. Each question is written and cross-checked against:

  • AAO Basic and Clinical Science Course (BCSC)

    Ophthalmology's official curriculum. Every question maps to a BCSC section (2023–2024 edition), so coverage mirrors the syllabus your exam is drawn from.

  • Kanski's Clinical Ophthalmology

    Cross-referenced for signs, imaging, and management — so the reasoning holds against more than one authoritative text.

  • Primary literature & landmark trials

    Where a fact rests on evidence — drug dosing, trial thresholds, first-line management — it is traced to the source trial or guideline, not to memory.

4

Calibrated on real performance

Publishing isn't the end. As residents answer, each question's difficulty is calibrated on how the cohort actually performs — not the author's guess. That real-world signal feeds your readiness index and flags any item that behaves oddly — too easy, too hard, or ambiguous — for another review.

5

Corrected in the open

Flag any question in a single tap. Reports go straight to the clinical team, who triage, fix, clarify, or retire the item — and the correction reaches everyone at once. A question bank that self-corrects is the entire point of working in the open.

When a reader flags a question
A resident flags aquestionClinical team reviewsthe reportFix, clarify, or retirethe itemCorrection goes live foreveryone

Where AI fits — and where it doesn't

Tooling can help authors draft faster and catch typos or formatting slips. But a clinician is accountable for every published question — the medicine, the keyed answer, and the explanation are written and reviewed by humans. Nothing is auto-published from a model.

Our commitment: quality is a process, not a badge. We watch how every question behaves in the cohort, revise the ones that don't hold up, and would sooner retire a weak item than defend it. Find one that's wrong — and it's fixed for everyone.

Content is the foundation. Two engines sit on top of it: adaptive spaced repetition (what to study, when) and the readiness index (how ready you are). See all three on the how-it-works overview.

The standards behind it

Our writing and review follow the same item-writing standards used to build medical licensing exams: