AI in Healthcare
A postgraduate learning module on applying artificial intelligence in clinical and healthcare settings, built around human-centred design and responsible AI.
This module introduces how artificial intelligence is being applied across healthcare, and, just as importantly, where it should not be applied without care. It is written for working professionals: clinicians, managers, and technologists who need to reason about AI in high-stakes settings rather than build the models themselves.
The through-line is human-centred, responsible design. A model that scores well on a benchmark can still fail a patient. So the module keeps returning to the same question: what does it take for an AI system to earn a place in a real clinical workflow?
What the module covers
The material is organised into a few connected strands.
Where AI fits in healthcare. A grounded tour of the settings where AI is genuinely useful today, triage and prioritisation, imaging and diagnostics support, administrative load, and decision support, separated from the settings where the hype runs ahead of the evidence.
Data, and why it is hard here. Healthcare data is messy, sensitive, and unevenly distributed. This strand covers provenance, consent, de-identification, and the ways bias enters through the data long before any model is trained.
Evaluation that means something. Accuracy is rarely the metric that matters. The module works through calibration, subgroup performance, failure modes, and the difference between a model that is right on average and one that is safe for the person in front of you.
Responsibility and governance. Accountability, transparency, and the human-in-the-loop. Who is answerable when a recommendation is wrong, and how that shapes the way a system should be designed and deployed.
How it is taught
Each topic pairs a concept with a concrete healthcare scenario, so the ideas stay anchored to practice. The emphasis is on judgement: knowing which questions to ask of a vendor, a dataset, or a deployment plan, rather than memorising a fixed answer.
This page is a living document. As the module runs, I will add the slides, readings, and worked scenarios here, and link the related notes as they are written.