Background: Artificial intelligence (AI) is predicted to improve healthcare, increase efficiency, save time and resources. However, research shows an urgent need to develop guidance to ensure that the use of AI in healthcare is ethically acceptable.

Purpose: To develop an ethical model to support AI implementation in practice.

Methods: The study used an explorative and empirically driven qualitative design. Individual interviews were conducted with 18 healthcare professionals from two emergency departments in Sweden where the county council has developed an AI application to predict the risk for unexpected mortality within 30 days after visiting an emergency department. A deductive analysis based on ethical theory i.e virtue, deontology and consequentialism, was used.

Findings: The developed model shows how the healthcare professionals use ethical reasoning in relation to the implementation of AI. In relation to virtue ethics, moral considerations in relation to the use of AI were mentioned. In relation to deontology, considerations were mentioned on actions performed based on information acquired from the technology and adherence to specific duties, roles and responsibilities. In relation to consequentialism, considerations about how to provide better resources more rapidly in an equal way and how the technology can be adjusted to each patients’ individual needs and preferences in order to support decisions, self-determination, and actions that are in the patients best interest.

Conclusions: Our findings provide an ethical model demonstrating the relevance of virtue, deontology and consequentialism when AI are to be implemented in practice.