Sustainable and Responsible Artificial Intelligence Implementation in Healthcare

Abstract

Artificial Intelligence (AI) has made significant progress over recent decades. However, its deployment in real-world scenarios has highlighted several risks, ranging from technical deficits to ethical concerns. This has prompted the development of theoretical and normative frameworks for Sustainable and Responsible AI Implementation in various sectors. Such frameworks consider different aspects of the AI lifecycle and deployment, yet their specific application to industrial and service sectors remains scarce. Healthcare is one domain where AI promises significant improvements in outcomes, quality, efficiency, and patient experience.
Nevertheless, best practices for Sustainable and Responsible AI Implementation in healthcare have yet to emerge. The interdisciplinarity and complexity of the domain, coupled with the numerous parallel efforts extending the implementation of established AI and machine learning concepts, call for careful and exhaustive synthesis. The methodology therefore synthesizes existing guidelines on data governance, quality, and privacy; Healthcare AI risk assessment and mitigation; Sustainability and Resource Efficiency in AI; Explainable AI; and AI Audit and Compliance. These aspects are contextualized for AI deployment in Healthcare, and compiled into a set of implementation factors for Sustainable and Responsible Healthcare AI. The implementation of these factors is essential for the deployment of AI solutions that minimize negative impacts on patients, society, and the environment and actively seek to create positive effects.