This audit is for the MCCQE Part I candidate weighing MedLumen, a low-cost app that pairs a question bank with a "Live Pass Probability Predictor" and performance analytics. Its strength is inexpensive, mobile spaced-repetition drilling. Its principal limitation is that its most eye-catching outputs, a live pass predictor and a headline pass rate, are unvalidated vendor figures, and its question count is not disclosed, so the analytics must not be mistaken for a readiness signal.
What MedLumen offers for MCCQE Part I right now
Figures below are vendor-reported and were checked on 19 July 2026; confirm them on medlumen.io before relying on them.
| Item | Vendor-reported (19 July 2026) | Audit note |
|---|---|---|
| Question volume | Not disclosed on the page | Verify the live MCCQE1 count on the product page before buying |
| AI / analytics features | "AI Clinical Tutor" (24/7), "Live Pass Probability Predictor", "Spaced Repetition", "Performance Analytics" (accuracy, completion, streaks, trends) | The predictor and pass claim need scrutiny |
| Access / pricing | "$25.00/mo" | Low-cost monthly access |
| Marketing claim | "94% First-Attempt Pass Rate" across all exams | Unvalidated; no methodology, denominator or cohort published |
| Content provenance | "Doctor-Written" question bank | Vendor claim; not independently verified |
Two findings belong at the top. First, MedLumen does not publish its MCCQE Part I question count, so you cannot judge coverage before buying, verify it on the product page. Second, the "94% first-attempt pass rate" and the "Live Pass Probability Predictor" are marketing outputs, not validated instruments; there is no published methodology behind either. Treat both as claims to interrogate, not numbers to plan around.
Exam anchor: what the analytics have to cover
Since April 2025 the MCCQE Part I is multiple-choice only: 230 MCQs across two sections of 115 (including unscored pilots), a maximum of two hours 40 minutes per section, and three to five options per item (verify on mcc.ca). Content follows the MCC Objectives across Dimensions of Care (health promotion and illness prevention, acute, chronic, psychosocial) and Physician Activities (assessment and diagnosis, management, communication, and the legal, ethical and organisational CLEO domain). The official MCC preparatory products, the Preparatory Examination and PE-Lite, are the calibration standard; a vendor pass predictor is not.
Define every metric before you trust it
Pin down each MedLumen number before you act on it. First-attempt accuracy is the score that most cleanly tracks learning; repeat accuracy inflates with recognition. A percentile is meaningful only if the comparison cohort and denominator are stated. A predicted score, or a "pass probability", is a model output that is worthless without a published basis, a validation sample and a stated error margin. Coverage should mean proportion of the blueprint attempted, not proportion of the app finished. Difficulty needs a defined scale, and time per item is the pacing metric that most directly maps to a section-timed exam. Where a definition is absent, as it is for MedLumen's pass predictor, the metric is decorative until proven otherwise.
Selection bias: what the engine optimises
Spaced repetition and weak-area targeting optimise your recall of the items the app keeps re-serving. That improves retention, but it over-samples your weak topics and re-serves familiar stems, so your rising in-app accuracy blends genuine learning with recognition, and your percentage stops being comparable to a blueprint-fair, unseen mock. The "Live Pass Probability Predictor" sits on top of exactly this biased stream, so whatever it is computing, it is computing from a mix that does not resemble the real exam. That is the core reason it cannot serve as your readiness verdict.
Blueprint audit: compare your mix, not your average
Ignore the home-screen average and tally your attempts by dimension against the MCC Objectives' weighting. Spaced-repetition apps characteristically leave low-frequency, high-stakes categories thin, CLEO and ethics, population and preventive health, image and data interpretation, and calculation. If the app has drilled you heavily on your worst organ systems and lightly on CLEO, the average and the pass predictor both speak to the former and stay silent on the latter, which the exam still tests. Build the coverage matrix and read the distribution, not the number.
Readiness test: the conditions for a credible signal
A score is a readiness signal only when the items are unseen, the block is timed to section pace, the mix is blueprint-representative rather than weak-area-weighted, you use no assistance mid-block, and the sample is large enough to be stable. MedLumen's spaced-repetition stream breaks the unseen and representative conditions by design, and its pass predictor is not a validated instrument. So measure readiness on unseen, blueprint-fair blocks, and let MedLumen do the drilling job it is actually built for.
Override rules: force what the app under-serves
Because the engine chases retention on its own item set, override it to cover what it under-samples. Force blocks in CLEO, ethics and professionalism; population and preventive health with Canadian screening intervals; image, ECG and data interpretation; and calculation items. Force periodic mixed, full-blueprint blocks even when the app wants to keep reviewing your weakest topic, and never let the pass predictor talk you out of testing a domain it rates you highly on.
Worked dashboard example
Suppose MedLumen shows first-attempt accuracy 66 per cent overall, a "pass probability" of 88 per cent, strong acute care, weak psychosocial and CLEO, and most attempts concentrated in your worst organ systems. The wrong reading is "88 per cent likely to pass, I'm fine." The right reading discards the unvalidated predictor and turns the accuracy breakdown into quotas: cap the over-drilled systems, force psychosocial and CLEO to 40 per cent of next week's volume, add image and calculation blocks, and book an unseen mixed measurement block. No pass prediction is endorsed; the analytics set quotas, and the readiness verdict comes only from unseen material.
Worked example: a seven-day IMG plan
For an international graduate balancing content review with Canadian conventions, give MedLumen the drilling job and iatroX the unseen-measurement job. Days 1 to 2: MedLumen spaced-repetition sessions forced onto psychosocial and CLEO, checking rationales against Canadian norms. Days 3 to 4: MedLumen image, data and calculation blocks. Day 5: consolidate logged errors, no new questions. Day 6: an unseen, timed, mixed iatroX block, no assistance, ignoring MedLumen's pass predictor entirely and reading only the unseen result. Day 7: review iatroX errors and reset next week's quotas. No proprietary-algorithm claim is made for either product; training and measurement stay on separate surfaces.
Decision checklist: continue, supplement, switch or stop
Continue with MedLumen if the low cost and mobile spaced repetition suit you and you treat the predictor as noise. Supplement it with official MCC material and an unseen measurement layer, and verify the undisclosed question count before committing. Switch if the hidden count turns out to be too thin to cover the blueprint, or if the app keeps re-serving recognisable items while your unseen scores stall. Stop adding new questions when unseen blocks plateau across dimensions and errors are careless rather than conceptual. Decide on measured gaps, not on a headline pass rate.
Three mistakes this audit is designed to stop
The first is treating the "Live Pass Probability Predictor" as a verdict. It is an unvalidated model output with no published methodology, error margin or cohort, computed on top of a spaced-repetition stream that does not resemble the real exam, so a reassuring number there is not evidence you will pass and should not change what you study. The second is buying before checking the question count. MedLumen does not disclose its MCCQE Part I volume on the page, and a bank you cannot size is a bank you cannot judge for coverage, so confirm the live count before you commit even at a low monthly price. The third is believing the headline pass rate. A "94% first-attempt pass rate" quoted across all exams, with no denominator, no cohort definition and no independent audit, is marketing, not evidence, and self-selected app users are not a representative sample. Each mistake has the same fix: discard the predictor, verify the count and coverage yourself, and derive your readiness only from unseen, timed, blueprint-representative blocks rather than from any figure the app markets back to you.
Bottom line
MedLumen is a cheap, mobile spaced-repetition drill with a genuinely useful retention engine and two outputs you should not trust: an unvalidated pass predictor and an unsubstantiated headline pass rate, wrapped around an undisclosed question count. Use it to drill and retain, verify its coverage yourself, discard the predictor, and keep your readiness measurement on unseen, blueprint-fair blocks.
Frequently asked questions
Is MedLumen enough for MCCQE Part I on its own? It is unlikely to be sufficient alone, and you cannot even judge that fully because the MCCQE1 question count is not disclosed, so verify it before relying on the app. Used as a low-cost spaced-repetition drill alongside official MCC calibration and an unseen measurement layer it has a place, but its pass predictor and headline pass rate are not evidence of readiness and should not be treated as a substitute for a blueprint-fair mock.
Which MCCQE Part I component does MedLumen not reproduce well? The applied CLEO dimension, the legal, ethical and organisational and population-health items framed to Canadian norms, is the typical weak point of a spaced-repetition app, because those low-frequency categories are exactly what the engine under-samples. Image, data and calculation items are the second common gap, and the timed, blueprint-representative section experience is a third. Force these manually rather than trusting the retention feed to surface them.
How many MedLumen questions should I complete per day for MCCQE Part I? Set the daily count by your review capacity and blueprint spread rather than a fixed number, and because the total bank size is undisclosed, prioritise even coverage of every dimension over raw volume. A smaller number of well-reviewed, well-distributed items, with the low-frequency CLEO and modality categories deliberately forced in, will serve you better than a large daily tally concentrated in the topics the app already favours.
When should I stop using MedLumen and move to mixed mocks? Move to unseen mixed mocks once your forced-blueprint drilling is stable across dimensions and the app is re-serving items you recognise, and disregard the pass predictor when making that call. In the final two to three weeks, prioritise timed, blueprint-representative mocks and the official MCC forms, keeping MedLumen for spot spaced-repetition on any single soft domain rather than for fresh volume or reassurance from its analytics.
How should I combine MedLumen with iatroX without duplicating practice? Give MedLumen the drilling-and-retention job and iatroX the unseen-measurement job, and never run the same item on both. Drill on MedLumen through the week, then sit a fresh iatroX block as your independent read-out, reading the unseen result rather than MedLumen's pass probability; if topics overlap, rotate so the measurement block stays genuinely unseen, following the two-Q-bank rule to keep your scores comparable.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 19 July 2026; MedLumen features and price are vendor-reported as at that date, its MCCQE Part I question count is not disclosed on the product page, and its "94% pass rate" and "Live Pass Probability Predictor" are unvalidated marketing figures, so verify all of these on medlumen.io. Disclosure: iatroX operates a competing MCCQE Part I question bank; this audit confines iatroX to the unseen-measurement job MedLumen does not validly provide, and makes no proprietary-algorithm claim. Corrections are welcome via the feedback route on iatrox.com.
References: Medical Council of Canada, MCCQE Part I multiple-choice and 2025 change pages, and preparatory products (mcc.ca); MedLumen product page (medlumen.io), vendor-reported; iatroX internal resources including Your Q-Bank Percentage Is Not Your Exam Score and the iatroX comparison hub.