For most working doctors and clinical-year students, the mythical free evening never arrives. Revision happens in fragments: the bus, the coffee queue, the twenty minutes after handover. The good news from learning science is that fragments are not a compromise; short, spaced retrieval sessions are close to optimal for retention. The tools just have to be built for them.
What fragment-first design means
Judge a revision app on six unglamorous properties. Session length: can it deliver value in five minutes, or does everything assume a desk hour? One-handed usability: full vignettes and lab tables readable with a thumb on a crowded train. Question loading: instant, because a spinner eats a third of a micro-session. Continuity: a set started on the phone visible in tonight's laptop analytics. Notifications: due reviews resurfaced at sane times, since unprompted spaced repetition is where streaks die. And offline behaviour: what genuinely works in a basement mess room, as opposed to what loads a cached shell.
The contenders
iatroX is built around the fragment: adaptive sets sized to the time you have, spaced reviews batched for exactly these windows, the Socratic Tutor available on a missed question in the queue, and the planner adjusting when a shift swallows a session, all synchronised across native iOS and Android apps. The range inside one app matters on the ward: question set, quick guideline answer, calculator, next patient.
Geeky Medics is the other app clearly designed for clinical life, and its offline OSCE guides are the best answer on this list to hostile hospital Wi-Fi.
UWorld and AMBOSS both offer mature native apps with offline question modes, polished for their US-centred banks; excellent commute tools if those are your exams.
Osmosis is the strongest choice for turning a commute into video time, with questions and flashcards alongside.
Browser-based platforms, including Neural Consult by current listings, are at a structural disadvantage here regardless of desktop quality: the tab is simply where micro-sessions go to die.
Workflows by window size
Five minutes: spaced reviews only. They are due, they are short, and clearing them is the highest-yield thing five minutes can buy. Ten minutes: a five-question adaptive set, plus the Tutor on anything missed. Twenty minutes: a ten-question themed set on your current weakest domain, ending with one guideline check on whatever felt shaky. The common thread is that every window starts with retrieval, not reading; reading is what the evening is for, if the evening comes.
The honest conclusion
The best commute app is the one that removes every step between unlocking your phone and answering a question you were about to forget. On that measure, for UK and international exam candidates, iatroX's combination of short adaptive sessions, automated review scheduling and Tutor follow-up in one native app is the strongest package; Geeky Medics wins the offline ward; UWorld and AMBOSS win the US commute. Whichever you choose, let the notification win occasionally. That is the whole system working.
A one-week commute experiment
Rather than take any app comparison on faith, including this one, run the experiment: one week, your real commute, one candidate app, three rules.
Rule one: the phone opens to the app, not to the feed. Move it to the home screen's first position and let the notification, not willpower, initiate the session; you are testing the app's ability to own the window, which is most of what you would be paying for.
Rule two: every window starts with retrieval. Reviews first if any are due, then a short adaptive set; reading only if the window outlasts the questions. Log nothing manually, because the app's own analytics are part of what is under test.
Rule three: score the week on four numbers. Sessions that actually happened out of commutes available. Questions answered. Misses that received tutor follow-up rather than a shrug. And the Friday check: does the app's picture of your weak topics match your own sense, or has it seen something you had not?
A good fragment-first platform, and this is where iatroX's short adaptive sets, batched reviews and in-queue Tutor are designed to shine, will post embarrassing improvements on the first two numbers simply by removing friction, and the fourth number is where adaptive machinery proves it was watching. An app that loses the week to loading screens, re-logins or sessions that die with the signal has told you everything, cheaply.
Then, and only then, buy the annual plan. The experiment costs one week and nothing else, which is the correct price for information this decision usually lacks.
One caveat to keep the experiment honest: judge the week on the numbers, not on how studying felt, because friction-free practice feels lighter precisely when it is working and effortful sessions flatter themselves in memory. The analytics are the referee, which is, conveniently, also the first lesson the exam itself will teach.