This workflow is for Italian medical graduates using SynapSSM's adaptive feed and analytics to prepare the Concorso SSM's 140-item multiple-choice test. Used well, an adaptive app concentrates effort where you are weak; used blindly, it lets a rising overall score conceal whole low-volume syllabus domains you have barely attempted. The principal limitation to manage is that personalisation optimises for its own item pool, which is not the same thing as transfer to the unseen national paper.
What SynapSSM offers for Concorso SSM right now
These details are vendor-reported from synapssm.com and last checked on 20 July 2026. One figure deserves caution: the site's metadata cites "over 50,000" quizzes while the page body refers only to "thousands", so the exact live total is ambiguous and should be verified on the product page rather than quoted with confidence.
| Item | Vendor-reported detail (20 July 2026) |
|---|---|
| Platform type | Mobile-first adaptive quiz app for the Concorso SSM |
| Question bank | Metadata cites "over 50,000" quizzes; the body says "thousands" — treat the exact total as unverified and confirm on synapssm.com |
| Adaptive/AI | Spaced repetition, "learn from your errors", weak-point analysis and a "data-driven method"; metadata references AI algorithms (vendor-reported, not independently verified) |
| Free tier | 45 quizzes per day and 1 simulation per day |
| Paid | A "Pro" plan around €4.15/month on a promotion from a stated regular €9.99/month, billed monthly, cancel anytime |
| Concorso SSM scope | The 140-MCQ written test; the specialty concorso, not TOLC-MED admission |
The adaptive framing is the whole point of the tool and also the thing to supervise. An engine that feeds you more of what you get wrong is efficient for weak points but structurally inclined to under-sample the areas you rarely see — which is exactly where a broad, fixed national paper can catch you.
The exam the analytics must serve
The Concorso SSM is one computer-based paper of 140 single-best-answer questions, five options each, over 210 minutes, marked +1 correct, −0.25 wrong and 0 blank. The test is worth up to 140 points and the titoli add up to 7, for a maximum of 147, after which a single national graduatoria decides specialty and seat. The exam samples the whole curriculum — medicina interna, chirurgia, farmacologia and the rest — regardless of your personal comfort, and there is no official ministerial question bank to rehearse, only past papers. Any adaptive analytics you rely on must therefore be judged against that fixed breadth and a working pace near 90 seconds an item, not against a personalised feed's internal metrics.
Baseline before you let it personalise
Spend the first week on a small, blueprint-stratified sample of unseen items before you let the app personalise anything. Sampling evenly across domains gives you a genuine coverage map instead of a feed already skewed by your first few answers. Record accuracy by domain, not just overall, because the overall figure is the one most likely to mislead you later; a first-week percentage is a starting map, not a grade, as the percentage article makes clear.
Set domain floors so the feed cannot starve a syllabus area
Set explicit domain floors — a minimum number of attempted items in each syllabus area — so that no adaptive feed can quietly park a low-volume domain while your headline percentage climbs. If the algorithm keeps serving you the topics you fail most, the topics you rarely see never accumulate enough exposure to be safe. The floor is a manual override: you commit to a baseline of attempts per domain, the app optimises within that, and the blueprint-coverage matrix is the tool for tracking that no area falls below the line.
Classify every error before you re-test
Do not send every miss straight back into the feed. Classify it first, because the right response depends on the cause.
| Error type | What it looks like | First response |
|---|---|---|
| Knowledge gap | You never knew the fact or mechanism | Short source read, then a new transfer question later |
| Misread stem | You knew it but misread "except", "most likely" or "next step" | Stem-marking drill; no re-study needed |
| Premature closure | You anchored on the first plausible option | Force a differential before answering |
| Guideline error | You used an outdated threshold or first-line choice | Check a current source; schedule a spaced re-test |
| Calculation error | Arithmetic, units or dose slip | Timed calculation reps |
| Time-pressure error | Right method, but you ran out of time | Pacing block at about 90 seconds an item |
Tagging errors this way stops you "revising" a stem you actually misread, or re-reading a guideline you simply mis-timed.
Choose the review interval by error type
Not every error deserves an immediate repeat. A knowledge gap is better served by a short source read and then a fresh transfer question days later than by re-clicking the same item while the answer is still warm. Misread-stem and time-pressure errors need drills on technique, not re-study. Guideline errors need a source check and a spaced re-test to confirm the correction has held. Letting error type set the interval is what turns analytics into learning rather than into a satisfying but shallow loop of instant corrections.
When to switch from filtered practice to timed random blocks
Reduce topic-filtered, adaptive practice once your domain floors are all met and first-attempt accuracy on unseen sets is stable. The objective signals are concrete: no domain below its floor, filtered accuracy high across areas, and pacing under control. When those hold, more filtered drilling over-fits the app's pool, and the marginal value shifts to timed, random, negative-marked 140-item blocks that reproduce the real paper's unpredictability. If filtered accuracy is high but random-block accuracy or pace lags, that gap — not novelty — is the reason to switch.
Exit criteria: coverage, calibration and retention
You are ready to taper the app when five conditions coincide: every domain has met its coverage floor; first-attempt accuracy on unseen items is stable; your pace holds near 90 seconds an item under negative marking; retention is holding on spaced re-tests rather than decaying between sessions; and you have calibrated against official material — at least one past official paper sat cold — rather than trusting bank completion. Bank completion alone certifies nothing; a full app and a strong unseen, calibrated score are different things.
A seven-day worked example
This is a repeatable template for a graduate mid-preparation. It uses SynapSSM for one job — adaptive drilling with enforced domain floors — and iatroX's Concorso SSM bank to measure transfer on unseen items, with no claim to any proprietary algorithm on either side.
- Day 1 — Blueprint-stratified unseen baseline; record accuracy by domain and set floors.
- Day 2 — Let SynapSSM personalise within those floors; tag every error by type.
- Day 3 — Clear knowledge-gap misses with short source reads; queue transfer questions for later, not instant repeats.
- Day 4 — Top up any domain sitting below its floor, even if the app would not prioritise it.
- Day 5 — Sit a fresh, timed unseen iatroX Concorso SSM block; because these items sit outside SynapSSM's pool, the result measures transfer rather than recognition.
- Day 6 — Run a timed random block under negative marking; review pacing and premature-closure errors specifically.
- Day 7 — Spaced re-test of earlier misses; calibrate against a section of a past official paper; rest.
Decision checklist: continue, supplement, switch or stop
Continue with SynapSSM while it is lifting weak domains and your floors are holding. Supplement it with an unseen measurement bank as soon as your in-app score outpaces your performance on unfamiliar items. Switch emphasis to timed random mocks when filtered accuracy is high but random-block accuracy or pacing lags. Stop adaptive drilling in a domain once it is stable on unseen, calibrated items and the feed is only recycling near-duplicates. Use the iatroX comparison hub to make that call on measurable gaps rather than on the app's streaks or the sunk cost of a subscription.
Frequently asked questions
Is SynapSSM enough for Concorso SSM on its own? As a high-volume adaptive drilling app it can carry a large share of your question practice, particularly for building speed and spacing your weak points. The caveat specific to adaptive tools is that "enough" depends on breadth as well as depth: an engine that feeds you more of what you get wrong can leave low-volume domains under-tested while your headline score rises. It is enough only if you enforce coverage yourself and confirm performance on unseen items outside its pool.
Which Concorso SSM component does SynapSSM not reproduce well? The concorso is wholly multiple choice, so there is no station or viva to miss. SynapSSM's specific weakness is that adaptive personalisation can under-represent the fixed, non-adaptive breadth of the real 140-item paper, which samples the whole curriculum regardless of your comfort. It also cannot reproduce the graduatoria or the titoli points, and — like any single bank — an app tuned to its own items measures recognition of those items rather than transfer to genuinely unseen ones.
How many SynapSSM questions should I complete per day for Concorso SSM? The free tier is vendor-reported at 45 quizzes and one simulation per day, which is a reasonable daily volume for most candidates, and paid access lifts the cap. But raw count is the wrong target: 40–80 items reviewed to the point where you can explain every distractor beats a few hundred skimmed, a point the percentage article makes at length. Let your domain floors, not the app's streaks, decide the mix each day.
When should I stop using SynapSSM and move to mixed mocks? Reduce topic-filtered, adaptive practice once your domain floors are all met and first-attempt accuracy on unseen sets is stable; that is when timed, random, negative-marked 140-item blocks earn their place. If your filtered accuracy is high but random-block accuracy or pacing lags, the adaptive feed has done its job and is now over-fitting, and the marginal value has moved to full mocks and structured error review.
How should I combine SynapSSM with iatroX without duplicating practice? Run them under the two-bank rule with distinct jobs: SynapSSM is the adaptive learning-and-spacing engine, iatroX is the unseen measurement bank you deliberately do not personalise. Never re-answer the same items across both; the point of the second bank is that it is unfamiliar and non-adaptive, so a timed iatroX Concorso SSM block shows whether your SynapSSM gains transfer. Because iatroX core content is English-language, read it as a test of underlying clinical reasoning rather than of Italian wording.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 20 July 2026; SynapSSM figures — the quiz total, free-tier limits and prices — are vendor-reported from synapssm.com on that date, and the total question count in particular is inconsistent between the site's metadata and body, so verify it on the product page. Disclosure: iatroX operates its own Concorso SSM question bank and therefore competes with SynapSSM; this article confines iatroX to the unseen-measurement role that an adaptive app does not claim to fill. Corrections are welcome via the feedback route on iatrox.com.
References: Ministero dell'Università e della Ricerca — portale Scuole di Specializzazione in Medicina and the annual bando/decreto (official format, scoring and titoli); SynapSSM product pages (synapssm.com, vendor-reported); the iatroX Concorso SSM landing page; "Your Q-Bank Percentage Is Not Your Exam Score"; and the iatroX blueprint-coverage matrix pillar.