In this page, we provide definitions and justifications for each of the six guiding principles and give an overview of the FUTURE-AI recommendations.

The following table provides a summary of the recommendations, together with the proposed level of compliance (i.e. recommended vs. highly recommended).

List of the FUTURE-AI recommendations, together with the expected compliance for both proof-of-concept (Low ML-TRL) and deployable (High ML-TRL) AI tools (+: Recommended, ++: Highly recommended).

  RecommendationsLow ML-TRLHigh ML-TRL
F1Define any potential sources of bias from an early stage++++
2Collect data on individuals’ attributes, when possible++
3Evaluate potential biases and bias correction measures+++
U1Define intended clinical settings and cross-setting variations++++
2Use community-defined standards (e.g. clinical definitions, technical standards)++
3Evaluate using external datasets and/or multiple sites++++
4Evaluate and demonstrate local clinical validity+++
T1Implement a risk management process throughout the AI lifecycle+++
2Provide documentation (e.g. technical, clinical)++++
3Define mechanisms for quality control of the AI inputs and outputs+++
4Implement a system for periodic auditing and updating+++
5Implement a logging system for usage recording+++
6Establish mechanisms for human oversight and governance+++
U1Define intended use and user requirements from an early stage++++
2Provide training materials and activities (e.g. tutorials, hands-on sessions)+++
3Evaluate user experience and acceptance with independent end-users+++
4Evaluate clinical utility and safety (e.g. effectiveness, harm, cost-benefit)+++
R1Define sources of data variation from an early stage++++
2Train with representative real-world data++++
3Evaluate and optimise robustness against real-world variations++++
E1Define the need and requirements for explainability with end-users++++
2Evaluate explainability with end-users (e.g. correctness, impact on users)++
General1Engage inter-disciplinary stakeholders throughout the AI lifecycle++++
2Implement measures for data privacy and security++++
3Define adequate evaluation plan (e.g. datasets, metrics, reference methods)++++
4Identify and comply with applicable AI regulatory requirements+++
5Investigate and address ethical issues+++
6Investigate and address social and societal issues++