AMBOSS Knowledge Profile, Equated Percent Correct and Score Predictor are not interchangeable names for one score. The Knowledge Profile models learning from question-bank activity. EPC expresses adjusted expected performance. The Exam Score Estimate translates relevant activity onto an exam scale, while Score Predictor uses a separate set of practice-exam results and dates.
That distinction is the useful starting point for reading the dashboard. A number intended to guide today's revision should not automatically become the sole reason to book, postpone or sit an examination.
What changed in August 2026?
AMBOSS's 24 August 2026 announcement introduced a more visible explanation of its Knowledge Profile. The company describes a model of topic and competency proficiency built from Qbank responses, not Library reading or AI Mode conversations. The article also explains that adaptive selection had been operating since April 2026; the whole system was not newly created in August.
Those are product-design descriptions. They do not establish that a specific learner's next examination result will match an estimate, or that using the dashboard causes a particular improvement in learning.
Raw percentage and EPC answer different questions
A raw percentage describes the questions you answered. EPC attempts to account for question difficulty and relevance to the selected objective. According to the analysis guide checked on 6 September 2026, its interpretation depends on the objective: selected board-exam objectives concern expected standardised-exam performance, while others concern the relevant AMBOSS question pool.
Consider an original example. One learner spends a session revisiting familiar questions. Another answers unfamiliar questions in a difficult topic. Their raw percentages cannot, on their own, establish which learner has stronger overall preparation. The selection of questions is part of the explanation.
An adjusted model is intended to address that problem, but adjustment is not magic. It still depends on the available responses, how questions are represented and the relationship between the practice material and the eventual assessment. A more sophisticated estimate remains an estimate.
The Exam Score Estimate is not an official result
For Step 2 CK, AMBOSS describes an estimate based on Qbank activity, accompanied by a range. The same analysis documentation explains that relevant metrics appear after at least 40 questions for the selected objective. That threshold allows a display to appear; it should not be interpreted as evidence that 40 questions are sufficient preparation.
The range deserves attention alongside the central number. A learner near an important decision boundary should not treat a single displayed score as a guarantee. Nor should a small movement be interpreted without considering what changed in recent practice.
Avoid reverse-engineering an official pass decision from a screenshot taken out of context. The appropriate use is to combine the estimate with independent practice assessments, the candidate's preparation history and advice from the relevant educational team.
What Score Predictor uses instead
The Score Predictor guide checked on 6 September 2026 describes entering practice-exam results and the dates they were taken, together with the target examination date. This is separate from the continuously updated Qbank-based profile.
That means two numbers can differ without either interface being broken. They may use different evidence, different time assumptions and different models. First check the inputs before trying to decide which number is "right".
A mistyped practice score, an incorrect date or a result obtained under unusually assisted conditions can undermine the usefulness of the prediction. Preserve how the assessment was completed rather than treating every entered score as equivalent evidence.
A practical way to read the dashboard
Start with the selected study objective. Then ask what information the metric uses and what action it can reasonably support.
For a learner deciding what to study this evening, topic-level weaknesses are often more actionable than a headline score. For someone assessing broader readiness after a period of revision, independent timed practice and the history of recent performance become more important.
Suppose a fictional student sees a weak area in interpreting investigations. They should inspect a few recent errors before committing a week to that label. Were the mistakes caused by unfamiliar terminology, an incorrect physiological model or choosing the wrong next step? Each requires different work.
The dashboard locates a possible problem. Reviewing the attempted questions helps explain it. A plan built only from labels risks prescribing more of the same practice without addressing the reason for the errors.
Do not study merely to improve the display
A reassuring number can become an unhelpful objective if it encourages avoiding weak topics or repeatedly answering familiar material. The practical aim is better independent performance, not a more attractive screenshot.
Use difficult questions to expose gaps, then test whether the correction transfers to an unfamiliar example. If the same misconception returns with a different presentation, the first explanation was not enough.
Likewise, do not repeatedly check the dashboard after every minor change in activity. Decide in advance when to review progress and what would justify changing the plan. This reduces the temptation to treat ordinary variation as an instruction to rebuild the entire revision strategy.
How this compares with iatroX's learning approach
This explanation is published by iatroX and includes iatroX among the learning options discussed. Under September 2026 product information, iatroX combines adaptive question sequencing, spaced repetition, a question-specific Socratic Tutor and an exam-date-aware study planner. These features should not be presented as a validated equivalent of AMBOSS's score-estimation models.
The practical connection is between an attempted question and the next learning action. A learner can investigate a misconception with Tutor, adjust the work scheduled around an exam date and practise application in an appropriate simulation track. The value lies in the learning workflow, not a promise to forecast a pass.
For an AMBOSS user, understand the provenance of each displayed metric before acting on it. For a learner who needs help explaining repeated errors, prioritise a tool that supports that reasoning task. For a high-stakes readiness decision, use more than one kind of evidence and retain the uncertainty that the dashboard itself acknowledges.
Frequently asked questions
Is EPC the same as my percentage correct?
No. Raw percentage summarises answered questions, while EPC adjusts expected performance for factors including difficulty and relevance to the selected objective.
Does Score Predictor use the same inputs as the Knowledge Profile?
No. The documentation checked on 6 September 2026 distinguishes practice-exam results and dates from the ongoing Qbank activity used by the profile.
Should I decide whether to sit the exam from one estimated score?
No single modelled score should carry that decision alone. Consider independent assessments, preparation conditions, the estimate's uncertainty and appropriate educational advice.