An Interdisciplinary Approach to Bias Management in Health: A Systematic Literature Review

Digital transformation is revolutionizing healthcare by bringing improvements and challenges. Digital Health (DH) solutions have increasingly leveraged Artificial Intelligence (AI) algorithms in recent years. Even though AI is improving healthcare services, adverse effects emerge. This is because AI outputs are not safe from biases, which are systemic errors occurring, for example, within the dataset on which the algorithm is trained. They lead to unfair and ineffective decisions that negatively impact human life and organizational processes, especially in healthcare. For instance, they could reinforce prejudices regarding specific misrepresented or underrepresented groups of people. For this reason, researchers belonging to different disciplines are currently studying how to mitigate and manage AI bias risk. However, literature still lacks a study that explores and systematizes the studies on this topic by adopting an interdisciplinary lens. This study intends to fill this gap. By conducting an interdisciplinary Systematic Literature Review on AI bias management in healthcare settings, this work will expand the comprehension of AI’s light and dark sides, providing some insights into its fair implementation and use in healthcare processes. The interdisciplinary lens will expand the general scientific comprehension of this phenomenon while suggesting how to face it to safeguard healthcare stakeholders’ interests. This research will contribute to the interdisciplinary literature on AI and innovation management in healthcare and provide implications for practitioners and policy-makers involved in AI regulation.