The ELSA Way is a free, community-driven handbook for building AI for healthcare with attention to its ethical, legal and societal aspects (ELSA), in every phase from first idea to daily use.
Define the problem, involve the right people, find sources of bias and settle the ethical and legal groundwork before anyone writes code.
Collect representative data, protect privacy, document the model and design how clinicians will work with it.
Validate the tool at other sites and in subgroups of patients, and test its clinical value, safety, fairness, robustness and explanations.
Fit the tool into clinical work, check its inputs and outputs, keep logs and audits, and decide who is accountable.
Why The ELSA Way?¶
Many AI models for healthcare perform well in published studies and are never used on a ward. The technology is often ready before the hospital is: nobody has agreed who checks the alerts, whether the data may be used, or what happens when the model is wrong.
The ELSA Way puts those questions next to the technical steps, phase by phase, so that you deal with them while choices are still cheap to change. It follows the FUTURE-AI guideline Lekadir et al. (2025), which sets out six principles for trustworthy healthcare AI: fairness, universality, traceability, usability, robustness and explainability.
Who is this for?¶
The ELSA Way is written for:
AI developers and data scientists who want practical guidance on responsible design and development;
healthcare professionals who evaluate or use AI tools in their work;
researchers who validate or publish AI systems in clinical settings;
policy makers and managers who set up governance and procurement;
patients and communities who want a say in how AI is used in their care.
The Pathways section suggests a route for each role. The Toolbox lists practical tools you can use in your project.
How to contribute¶
Anyone can suggest a correction, flag a gap or write new content. The Community Handbook explains how.
- Lekadir, K., Frangi, A. F., Porras, A. R., & others. (2025). FUTURE-AI: international consensus guideline for trustworthy and deployable artificial intelligence in healthcare. BMJ, 388, e081554. 10.1136/bmj-2024-081554