Artificial Intelligence (AI)-based systems are rapidly revolutionising the process of healthcare delivery, introducing both opportunities and challenges. Innovation in the United States (US) to maximise the opportunities of improved diagnosis, treatment and management of disease has, however, created a gap between utility and responsible AI practices. There are several challenges that require determined efforts to solve, including data diversification and non-biased models; model explainability and algorithmic transparency; healthcare provider and patient education of AI systems; and overall human-centred considerations of ethics, fairness and human well-being. In particular, as patient-facing decision support smart systems like mobile digital applications and online sources integrate with clinical decision support systems, responsible AI practices become fundamental to not just improving the healthcare outcomes of all, but protecting patients from any possible harms and violation of their rights resulting from the use of such technologies. In this article, I argue that building responsible mobile digital AI-based health technologies across the US healthcare pipeline necessitates a revision of human values, ie one grounded in a human-centred framework whereby humans unequivocally remain at the centre of the AI lifecycle, to underpin a regulatory approach that augments both current broad US initiatives in AI ethics development and integration, and more specific Food and Drug Administration efforts.
Keywords: human-centred AI; responsible AI; healthcare; mobile digital health technology