Requirements for Human-Centered Artificial Intelligence: A Heart Failure Study Across Europe and Latin America

Abstract

This paper explores the requirements for humancentered artificial intelligence (AI) tools for heart failure (HF) management, focusing on the needs of diverse healthcare settings in selected European countries (Netherlands, Spain, Czech Republic) and a Latin American country (Peru). Clinicians, patients, ethicists, and technical experts were engaged through cocreation workshops, local groups, narrative interviews, and surveys to gather clinical, ethical, and regulatory requirements for AI implementation in HF care. These activities provided input on the intended clinical use of AI tools, as well as patient data privacy and security concerns. Clinical requirements revealed regional differences in AI tool preferences and key predictors. European clinicians favored integration into secondary and tertiary care, focusing on quality of life and comprehensive follow-up measures, while clinicians in Peru prioritized secondary care with an emphasis on treatment adherence and complication management. Ethical considerations, such as data privacy and bias mitigation, were universally important but some context-specific differences emerged. European stakeholders emphasized mitigating biases related to sex, ethnicity, and socioeconomic status under European regulations, whereas Latin American stakeholders focused on context-specific ethics and robust national oversight. By aligning these insights with FUTURE-AI principles, the study ensures the development of effective, human-centered AI tools. This research highlights the importance of continuous stakeholder engagement and contextualizing AI applications to enhance their relevance, usability, and adoption across diverse healthcare settings.

Ähnliche Arbeiten