Integrating Participatory Social Innovation Into Requirements Engineering for AI Health Care Solutions: Case Study.

Area of Science:

• Digital Health
• Artificial Intelligence in Healthcare
• Requirements Engineering

Background:

• Effective AI in healthcare demands early, multidisciplinary collaboration.
• Disciplinary differences and communication barriers hinder AI development and implementation.
• A structured approach to AI requirements specification is crucial for shared understanding and successful clinical integration.

Purpose of the Study:

• To present a methodology for AI requirements gathering, specification, mapping, and verification in clinical settings.
• To address the complexities of multistakeholder environments for clinically applied AI.
• To implement and validate this methodology within an international project for an AI-based breast cancer treatment response prediction tool.

Main Methods:

• Conducted 3 iterative rounds of AI requirements gathering with nearly 150 global experts and patients.
• Combined established requirements engineering methods (e.g., MoSCoW) with social innovation techniques.
• Ensured inclusivity and contextual relevance through a participatory, multistakeholder approach.

Main Results:

• Developed a structured framework integrating participatory social innovation and formal requirements engineering for international AI healthcare.
• Generated 184 consensus-based requirements categorized by stakeholder group, offering insights for AI in oncology and digital health.
• Ensured AI requirements align with the FUTURE-AI framework for trustworthiness (fairness, universality, traceability, usability, robustness, explainability).

Conclusions:

• The methodology advances digital health requirements engineering by incorporating nontechnical, global stakeholder needs.
• This unified approach is vital for creating AI solutions that are technically sound, clinically relevant, legally compliant, and socially acceptable.