New guidelines to build trusted AI in healthcare research

A new set of guidelines have been launched to create trustworthy AI systems in healthcare. The first of its kind, the FUTURE-AI guideline provides recommendations covering the entire lifecycle of medical AI, from design, development and validation to regulation, deployment, and monitoring.

In recent years, artificial intelligence (AI) has made significant strides in healthcare, helping with tasks like disease diagnosis and predicting treatment outcomes. However, despite these advances, many healthcare professionals and patients are still hesitant to fully embrace AI technologies. This hesitation largely stems from concerns about trust, safety, and ethics.

In particular, existing research has shown that AI tools in healthcare can be prone to errors and patient harm, biases and increased health inequalities, lack of transparency and accountability, as well as data privacy and security breaches.

To overcome these challenges the FUTURE-AI Consortium has developed a comprehensive set of guidelines published in the BMJ. Developed by an international consortium of 117 experts from 50 countries the new guidelines called FUTURE-AI provide a roadmap for creating trustworthy and responsible AI tools for healthcare.

The FUTURE-AI guidelines are built around six guiding principles:

1. Fairness: AI should treat all patients equitably and without bias.
2. Universality: AI solutions should be applicable across different healthcare contexts and populations.
3. Traceability: It should be possible to track how AI systems make decisions.
4. Usability: AI tools must be user-friendly for healthcare professionals and patients alike.
5. Robustness: AI systems should perform reliably under various conditions.
6. Explainability: Patients and clinicians need clear explanations of how AI arrives at its conclusions.

These guidelines fill an important gap in the field of healthcare AI to give clinicians, patients, and health authorities the confidence to adopt AI tools knowing they are technically sound, clinically safe, and ethically aligned. The FUTURE-AI framework is designed to evolve over time, adapting to new technologies, challenges, and stakeholder feedback. This dynamic approach ensures the guidelines remain relevant and useful as the field of healthcare AI continues to rapidly advance."

Gary Collins, Professor of Medical Statistics at the Nuffield Department of Orthopaedics, Rheumatology and Musculoskeletal Sciences (NDORMS), University of Oxford, and author of FUTURE-AI