[Paper Review] FUTURE-AI: International consensus guideline for trustworthy and deployable artificial intelligence in healthcare

Karim Lekadir, Aasa Feragen+118 more|arXiv (Cornell University)|Aug 11, 2023

Artificial Intelligence in Healthcare and Education38 citations

TL;DR

FUTURE-AI presents an international, consensus-based framework with six guiding principles and 28 recommendations to develop, deploy, and monitor trustworthy medical AI across the full lifecycle.

ABSTRACT

Despite major advances in artificial intelligence (AI) for medicine and healthcare, the deployment and adoption of AI technologies remain limited in real-world clinical practice. In recent years, concerns have been raised about the technical, clinical, ethical and legal risks associated with medical AI. To increase real world adoption, it is essential that medical AI tools are trusted and accepted by patients, clinicians, health organisations and authorities. This work describes the FUTURE-AI guideline as the first international consensus framework for guiding the development and deployment of trustworthy AI tools in healthcare. The FUTURE-AI consortium was founded in 2021 and currently comprises 118 inter-disciplinary experts from 51 countries representing all continents, including AI scientists, clinicians, ethicists, and social scientists. Over a two-year period, the consortium defined guiding principles and best practices for trustworthy AI through an iterative process comprising an in-depth literature review, a modified Delphi survey, and online consensus meetings. The FUTURE-AI framework was established based on 6 guiding principles for trustworthy AI in healthcare, i.e. Fairness, Universality, Traceability, Usability, Robustness and Explainability. Through consensus, a set of 28 best practices were defined, addressing technical, clinical, legal and socio-ethical dimensions. The recommendations cover the entire lifecycle of medical AI, from design, development and validation to regulation, deployment, and monitoring. FUTURE-AI is a risk-informed, assumption-free guideline which provides a structured approach for constructing medical AI tools that will be trusted, deployed and adopted in real-world practice. Researchers are encouraged to take the recommendations into account in proof-of-concept stages to facilitate future translation towards clinical practice of medical AI.

Motivation & Objective

• Address the limited real-world adoption of medical AI by establishing a formal, international consensus on trustworthy AI in healthcare.
• Define a comprehensive framework covering the full AI lifecycle—from design and development to deployment and monitoring.
• Establish six guiding principles (Fairness, Universality, Traceability, Usability, Robustness, Explainability) and actionable best practices.
• Provide a risk-informed, assumption-free guideline to facilitate translation of medical AI into clinical practice.

Proposed method

• Assembled a 118-member, interdisciplinary, international consortium from 51 countries.
• Employed a multi-stage Delphi process and online consensus meetings over 24 months.
• Conducted an in-depth literature review to identify dimensions relevant to trustworthy AI in medicine.
• Developed six guiding principles inspired by FAIR data principles.
• Generated and iteratively refined a set of recommendations (54 → 22 → 30 → final 28).
• Applied a machine-learning-technology-readiness-level (ML-TRL) lens to distinguish recommendations for proof-of-concept versus deployable AI.

Experimental results

Research questions

• RQ1 What guiding principles are necessary for trustworthy AI in healthcare?
• RQ2 What concrete best practices ensure fairness, universality, traceability, usability, robustness, and explainability across the AI lifecycle?
• RQ3 How should recommendations differ between early-stage research and clinical deployment (ML-TRL distinctions)?
• RQ4 How can international consensus be achieved and operationalized across diverse settings and domains in medicine?

Key findings

• Established six guiding principles for trustworthy medical AI: Fairness, Universality, Traceability, Usability, Robustness, Explainability.
• Defined 28 concrete recommendations spanning technical, clinical, legal, and socio-ethical dimensions.
• Demonstrated that the recommendations apply across the full AI lifecycle (design, development, validation, regulation, deployment, monitoring).
• Introduced an ML-TRL-based distinction to tailor recommendations for proof-of-concept versus deployable tools.
• Achieved broad international consensus through iterative surveys, expert feedback rounds, and four consensus online meetings.