This audit is for FACEM trainees considering PrimeX for the ACEM Fellowship exam who want to know how faithfully its simulator reproduces the real assessment before they rely on it. PrimeX genuinely targets the Fellowship — AI-graded short-answer questions, an MCQ bank and an OSCE simulator, all vendor-reported. Its principal limitation is intrinsic to any automated simulator: it can rehearse structure and content, but it cannot reproduce examiner variability, live role-player dynamics, non-verbal behaviour or a real physical examination. One honest note up front: iatroX covers ACEM Primary, not the Fellowship, so it appears here only as a foundation-knowledge and unseen-MCQ layer, never as a Fellowship SAQ or OSCE product.
What PrimeX offers right now
PrimeX is an Australian platform marketed as AI-graded SAQ, viva and MCQ practice for specialist exams, including the ACEM Fellowship. The figures below are vendor-reported as of 19 July 2026; confirm the live numbers and price on primexstudy.com.au.
| Feature | Vendor-reported position (19 July 2026) |
|---|---|
| SAQ practice | Full library with AI grading and examiner-style debrief (Fail / Borderline / Pass / Distinction), marking checklists and model answers |
| MCQ / SCQ bank | Fellowship-level questions with explanations for all options; described as unlimited |
| OSCE simulator | Vendor states all six ACEM OSCE station types are covered, with a voice mode and a 4-minute reading + 7-minute assessment structure |
| Study notes and flashcards | A study-note library and several thousand spaced-repetition flashcards (vendor-reported in the thousands) mapped to the FACEM 2021 curriculum |
| Curriculum mapping | Progress tracked across the ~206 FACEM 2021 learning objectives and their sections |
| Access and price | 7-day free trial; monthly, quarterly and yearly plans in Australian dollars — verify current pricing on the product page |
Credit where due: reproducing the 4-plus-7-minute station cadence and grading SAQs against examiner-report themes is a serious attempt at fidelity, and the free public SAQ grader lowers the barrier to trying it. The audit below is about where automated fidelity necessarily stops.
The ACEM Fellowship exam it has to model
The FACEM assessment has two independent components. The Written comprises two 180-minute papers — a Short Answer Questions (SAQ) paper and a Single/Select Choice Questions (SCQ, multiple-choice) paper — six hours in total. The Clinical exam is an OSCE of up to twelve stations, each eleven minutes (four minutes reading plus seven minutes of assessment), 132 minutes in all, run over two consecutive days. Station types span history-taking, examination, communication, resuscitation and management discussion, teaching, and structured case-based discussion, with live role-players in the interactive stations and examiners scoring against station-specific guides. ACEM does not publish a fixed numerical pass mark; standard-setting uses examiner judgement, and published examiner reports are more informative than headline pass rates. Confirm the current format on acem.org.au, which is the authority; PrimeX's coverage claims are vendor mappings onto that structure.
Format map: what the simulator reproduces
Mapped against the exam, PrimeX's simulator reproduces the following, per the vendor: the SAQ writing task and its timed structure; the SCQ/MCQ recognition task; the station cadence (reading then assessment) for the OSCE; a spoken interaction via voice mode; and a scored debrief tiered like an examiner outcome. What it models as text-and-voice rather than as a physical event: the examination station (you describe or narrate rather than perform), the role-player relationship in communication stations, and the teaching or case-based discussion dynamic where a real examiner adapts to your answers in real time. That distinction — reproduced task versus reproduced conditions — is the core of a simulator audit.
Fidelity test: timing, interface, scoring, branching, permitted actions
On timing, the vendor's 4+7 structure matches the ACEM station clock closely, which is the single most useful thing a simulator can get right. On interface, an app is not an exam-day station with a real examiner in the room; rehearse the mechanics but do not mistake comfort with the app for comfort with the setting. On scoring categories, PrimeX's Fail/Borderline/Pass/Distinction tiers echo examiner language, but an automated grade is a model's estimate against a checklist, not an examiner's holistic judgement — treat it as directional. On branching, a genuine OSCE branches through a role-player's responses and an examiner's follow-ups; an automated station branches through a model, which is more consistent but less unpredictable than a human. On permitted actions, be clear about what you are actually allowed to do in the app versus the real station, especially for examination and procedural discussion. Where automated scoring and human assessment diverge, the human wins on exam day; calibrate the automated score against real examiner reports and, where you can, against a supervising FACEM's feedback.
Case-mix audit
A simulator is only as representative as its case library. Check that PrimeX's SAQ and OSCE content spans the realistic proportions the exam samples: common high-stakes presentations (undifferentiated chest pain, the crashing airway, sepsis, major trauma, the unwell child) alongside the rarer but examinable material; acute and chronic-on-acute presentations; and — easily under-weighted — the communication, ethics, teaching and patient-safety stations that carry real marks in the OSCE. If the library skews toward resuscitation and MCQ-friendly diagnostics and thin on error disclosure, consent, end-of-life discussion and teaching, your rehearsal will be lopsided in exactly the domains candidates find hardest to self-assess.
Feedback audit
Automated feedback blends three different things, and you should separate them. First, observable behaviours the system can legitimately detect — did you cover the checklist items, did you state the safety-critical step. Second, inferred competence, where the model guesses at reasoning it cannot actually see. Third, model-generated commentary, which can be fluent, plausible and occasionally wrong. Trust the first, sanity-check the second, and verify the third against a primary source before you internalise it. The general principle for any AI-graded SAQ or OSCE is to calibrate the automated score against known-standard material before you trust it — do not let a confident tier label overwrite a supervisor's assessment.
Repetition risk
The specific hazard of a finite simulator library is false fluency: after enough passes you are recognising cases, not reasoning through them, and your tier grades rise for the wrong reason. Guard against it by rotating unseen scenarios, keeping a reserve of cases you have never attempted for late calibration, and periodically writing an SAQ answer cold before you see any model answer. If your PrimeX grades climb while your performance on genuinely novel material stalls, you are memorising the library.
What it cannot test
Be explicit about the gaps, because they are where candidates are blindsided. An automated simulator cannot test true non-verbal behaviour and rapport with a real role-player; it cannot reproduce examiner-to-examiner variability or the pressure of a human observer; it cannot replicate a hands-on physical examination or the tactile and interpretive parts of a real station; and it cannot rehearse your local departmental workflow or the exam-day logistics. These are not PrimeX failings — they are the boundary of the modality. Cover them with human practice: supervised mock OSCEs, peer role-play and real examiner feedback.
A seven-day worked example
One week for a trainee revising around clinical shifts, using PrimeX for one defined job — timed SAQ and OSCE-station rehearsal with its AI debrief — and iatroX only as an unseen ACEM-Primary-level foundation-knowledge check. No proprietary-algorithm claims are made about either tool.
- Monday — Two PrimeX SAQ stations under the clock; read the AI debrief, then verify every clinical claim against a primary source before accepting it.
- Tuesday — One PrimeX OSCE communication or ethics station in voice mode; note where the model's grade and your own judgement diverge.
- Wednesday — 30 unseen ACEM-Primary foundation items in iatroX (physiology, pharmacology, anatomy that Fellowship answers assume), timed; compare the error type with your SAQ misses, not the wording.
- Thursday — One PrimeX resuscitation or management-discussion station; write your structure cold first, then compare with the model answer.
- Friday — SCQ/MCQ block on PrimeX in your weakest curriculum section.
- Saturday — A supervised, human mock station with a FACEM or peer — the part no simulator replaces.
- Sunday — Reconcile the week: which gaps were knowledge (fix with Primary-level revision and iatroX measurement) versus performance (fix with human practice).
iatroX's role here is deliberately narrow and stated plainly: it is a Primary-level foundation-knowledge and unseen-MCQ layer that keeps the underlying science sharp and measurable. It is not a Fellowship SAQ or OSCE product, and it does not replace PrimeX's simulator or a human mock.
Decision checklist: continue, supplement, switch or stop
- Continue PrimeX while it is still surfacing unseen cases and your cold SAQ performance is improving.
- Supplement with supervised human OSCE practice always — the simulator cannot cover role-player and examiner realism — and with a foundation-knowledge check if Primary-level gaps are driving Fellowship errors.
- Switch or add another resource only for a measurable gap: a station type the library underserves, or feedback you cannot calibrate.
- Stop leaning on the simulator once your grades reflect recognition of a familiar library rather than fresh reasoning; move to unseen cases and human mocks.
Bottom line
PrimeX is a credible, well-structured attempt to simulate the ACEM Fellowship, and its timing fidelity and examiner-style SAQ grading are real strengths worth using. Audit it for what it is: a content-and-structure rehearsal tool whose automated feedback needs human calibration and whose finite library invites false fluency. Pair it with supervised human OSCE practice for the conditions it cannot reproduce, and keep your Primary-level foundation knowledge measurable on unseen items — the job iatroX can honestly do — rather than expecting any single tool to carry the whole Fellowship.
Frequently asked questions
Is PrimeX enough for ACEM Fellowship on its own? For the written components and structured SAQ rehearsal PrimeX can do a lot, and its AI grading is a reasonable first-pass mirror of examiner themes. It is not enough on its own for the OSCE, because no automated simulator reproduces live role-players, examiner variability, non-verbal rapport or a hands-on examination. Treat PrimeX as one layer and add supervised human practice for the clinical exam.
Which ACEM Fellowship component does PrimeX not reproduce well? The OSCE conditions, more than the OSCE content. It can rehearse station structure, timing and a spoken answer, but the human elements — a role-player's unscripted responses, an examiner's holistic judgement and follow-ups, a genuine physical examination — sit outside what a text-and-voice simulator can deliver. Those are best rehearsed with a FACEM supervisor or peers.
How many unseen PrimeX cases or stations should I preserve for final ACEM Fellowship calibration? Keep a deliberate reserve you have never attempted — a practical target is at least eight to ten unseen SAQ stations and several unseen OSCE scenarios — for the final one to two weeks. Calibration only works on material your recognition has not already contaminated, so ring-fence those cases from the start and do not dip into them for practice.
When should I stop using PrimeX and move to mixed mocks? Move toward full, mixed, human-supervised mocks once your cold SAQ performance is stable and your OSCE-station structure is automatic, and certainly once your PrimeX grades are rising mainly because you recognise the library. The simulator's job is rehearsal and diagnosis; the final weeks belong to unseen material and realistic mock conditions.
How should I combine PrimeX with iatroX without duplicating practice? Keep them in strictly separate roles. PrimeX handles Fellowship-level SAQ and OSCE rehearsal; iatroX handles unseen ACEM-Primary-level foundation-knowledge measurement — the physiology, pharmacology and anatomy that Fellowship answers assume. Because iatroX is not a Fellowship SAQ or OSCE product, there is no overlap to duplicate: you use it to confirm that the underlying science is secure and to compare error types, not to re-rehearse Fellowship cases.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 19 July 2026. Question counts, station coverage, AI-grading behaviour, access periods and prices are vendor-reported and change without notice; verify the current details on primexstudy.com.au before relying on them. Disclosure: iatroX operates a competing question bank, but for the ACEM Fellowship iatroX covers only ACEM Primary — it is not a Fellowship SAQ or OSCE product and is not a simulator, so this audit confines iatroX's role to unseen Primary-level foundation-knowledge measurement and makes no proprietary-algorithm claims. Corrections are welcome via the feedback route on iatrox.com. References: ACEM Fellowship examination format and curriculum (acem.org.au); PrimeX ACEM Fellowship product pages (primexstudy.com.au/acem-fellowship); iatroX ACEM Primary bank (iatrox.com/australia/exam/au-acem); "Your Q-Bank Percentage Is Not Your Exam Score" (iatrox.com/blog/qbank-percentage-not-your-exam-score); and calibrating AI-graded SAQs and OSCEs before you trust the score (iatrox.com/blog/ai-graded-saqs-and-osces-how-to-calibrate-automated-feedback-before-you-trust-the-score).