This pathway is for patients, carers, patient advocates and community members who want to understand how AI in healthcare is built and how to have a say in it. You do not need a technical background. Where the pages below use technical terms, they explain them.
Why your perspective matters¶
The people who build AI tools for healthcare often have no personal experience of the illness, the treatment or the waiting room. An AI tool learns from past patient records, and those records can leave groups out. If few people like you were in the data, the tool may work less well for you. Records also show what was measured, which is not always what matters to patients: a tool can predict a lab value well and still miss the symptom that worried you most. Past unfairness in care, such as one group being referred less often, can end up built into the tool. And a tool that works on paper can still disrupt the care you rely on, for example when staff spend time answering alerts instead of talking with you.
Patients and carers can spot these problems early, before a tool reaches the ward.
Suggested reading¶
The AI lifecycle in healthcare, which shows the stages where you can take part
Design: stakeholder engagement, on how patients can be involved from the start
Evaluation: usability and user experience, on what testing a tool with its users involves
Evaluation: clinical utility and safety, on what “safe” means for an AI tool
FUTURE-AI: fairness, on why AI tools can work better for some groups than for others
How to contribute to The ELSA Way¶
Your experience can improve this book too. You can point out a page that is hard to follow, or suggest an example from your own care. Contributing to The ELSA Way explains how, including options if you do not use GitHub.
Tools¶
These tools from the Toolbox suit your role.