Journal

PARA candidates face a thin resource market, so structured diagnosis matters more — work out whether breadth

FFICM failure usually clusters in physiology, ventilation, shock and data interpretation — rebuild around integrated ICU reasoning, not isolated facts.

FRCA failure is usually rote facts without mechanisms — rebuild the underlying physiology and pharmacology so applied and viva-style questions follow.

Paper A and Paper B fail for different reasons — sciences and statistics versus clinical judgement, risk and law — so diagnose which before you rebuild.

After failing MRCP Part 1 the question is not 'which Q-bank is best' but 'which failure mode did my last Q-bank fail to detect'.

A calm, diagnostic seven-day plan for UKMLA and finals resits — work out what actually went wrong before you rebuild.

Most trainees already revise with AI. Five habits to make it build recall instead of an illusion of competence — and why question-first tools enforce them for you.

ChatGPT, Pastest Tutor, Osmosis or iatroX — an honest, evidence-based comparison of AI tutors for medical exams, including who each one is (and is not) for.

ChatGPT writes a perfect explanation of your wrong MRCGP AKT answer — and that is exactly why it does not teach you. What real tutoring does differently.

The iatroX Socratic Tutor asks you first instead of handing you the answer — question-first, grounded in NICE/CKS/SmPC, calibrated per exam. Here is why we built it that way.

Used as an answer machine, AI can quietly worsen your exam retention while feeling like it helps. The evidence — and what a learning-safe tutor does instead.

Doximity's model shows a flywheel: clinician network → workflow tools → AI usage → data → insight products → partnerships. iatroX can build a UK version around clinical questions, ...