A patient moves from a symptom to a treatment through five steps. Digital systems have entered all five, and each raises a different question — which is why a single verdict on “AI in medicine” is always weak.
Where a system sits in a patient's path, and whether a person still decides after it.
Interactive diagram
Explore the labelled diagram, charts and maps for this topic in full study mode.
Free preview
This is the free notes preview
You're reading the free notes. Aimnova Pro unlocks the full study experience — and you can try it with your first topic free to keep:
- FlashcardsLock in vocabulary and key terms with spaced repetition.
- Practice questionsAnswer exam-style questions and get instant AI marking.
- Mock exams & past-paper vaultSit full mocks and see exactly how examiners award marks.
- Personalised study planA daily plan built around your exam date and weak areas.
It proposes
- A clinician sees the suggestion
- They can disagree, and sometimes do
- Responsibility stays with a person
It decides
- The outcome happens automatically
- Nobody reviews the ordinary case
- An error has no owner
The line is not always where it looks: A system that formally proposes can decide in practice: if a clinician must justify every disagreement and has thirty seconds per case, the suggestion is being followed rather than considered.
Study smarter, not longer
Most students waste 40% of study time on topics they already know. Our AI tracks your progress and optimizes every minute.
Real-world examples you can name
Consumer wearables detecting atrial fibrillation — Apple Heart Study published 2019
A smartwatch notifies the wearer of an irregular pulse. A large study found most people notified did have an irregular rhythm on follow-up, but only a small fraction of wearers were ever notified — so it screens, it does not diagnose.
Who it affected: Wearers who found a condition early, and health systems handling the consultations that false alarms generate.
AlphaFold and the protein structure database — AlphaFold 2 published 2021; over 200 million predicted structures released July 2022
A machine-learning system that predicts a protein's three-dimensional shape from its amino-acid sequence, a problem that had resisted fifty years of work. The predictions were released openly rather than licensed.
Who it affected: Biologists worldwide, including laboratories that could never afford the equipment to determine structures experimentally.
School closures and remote learning — 2020–2022
At the peak, UNESCO counted over 1.5 billion learners out of school. Where teaching moved online it kept going for households with a device and a connection and stopped for those without, turning the digital divide into a measurable gap in learning.
Who it affected: School-age children, most sharply those in low-income and rural households.
Where the strongest case is: Detection. Consumer wearables detecting atrial fibrillation (USA, then widely, Apple Heart Study published 2019) finds a heart rhythm problem that a yearly check would miss entirely, and it hands the finding to a person rather than acting on it.
How this is tested — health is one of the most common Paper 2 and Paper 3 contexts, and it rewards precision about where the system sits. It comes up two ways:
Paper 1 — structured question
- Part c: evaluate an automated system used in healthcare
- Part b: explain a risk of automated triage
Paper 2 — source-based question
- Q4: synthesise sources on health technology
- HL Paper 3: recommend, with a human review route
The trap: “AI in medicine is dangerous”: Say which step. Detection, triage and prescribing carry completely different risks, and a general verdict covers none of them.
Explain one risk of using an automated system to decide which patients are seen first.
Model answer plan
See the mark-by-mark plan — for / against / judgement, with marking guidance — in study mode.
Evaluate the use of automated image analysis in diagnosis.
Model answer plan
See the mark-by-mark plan — for / against / judgement, with marking guidance — in study mode.