ETHAI implementation. A Co-Constructive way for Embedding Ethics in Healthcare AI Development

This paper presents the ETHAI (ETHics of AI) methodology, a novel multidisciplinary approach to integrating ethical values within artificial intelligence systems designed for healthcare applications. While theoretical frameworks for AI ethics have proliferated, significant gaps persist between abstract principles and practical implementation. ETHAI addresses this disconnect through a co-constructive process that brings together ethicists, technical developers, clinical experts, and potential users to collectively define and operationalize ethical requirements. The methodology is characterized by four key principles: contextual awareness, practical application, cross-disciplinary collaboration, and iterative refinement. In specific, this paper reports on the preliminary results of implementation of ETHAI within COMFORTage, a European Commission-funded project developing AI systems for monitoring and supporting patients with dementia and age-related frailty. The findings highlight how the participatory risk assessment provided by the method helped identify domain-specific considerations requiring adaptation of generic ethics principles, while revealing significant cross- disciplinary communication challenges in translating ethical concerns into technical specifications. The semi-structured approach proved particularly effective in balancing normative frameworks with stakeholder engagement, demonstrating that meaningful ethical integration remains possible even when introduced at the design phase instead of project conception. This case study contributes valuable empirical evidence to the field of applied AI ethics and design, where practical implementation experiences remain underreported despite their importance for advancing responsible technology development.