Equity within AI systems: What can health leaders expect?

Abstract

Artificial Intelligence (AI) for health has a great potential; it has already proven to be successful in enhancing patient outcomes, facilitating professional work and benefiting administration. However, AI presents challenges related to health equity defined as an opportunity for people to reach their fullest health potential. This paper discusses the opportunities and challenges that AI presents in health and examines ways in which inequities related to AI can be mitigated.

Advancing Health Equity Through Artificial Intelligence: An Educational Framework for Preparing Nurses in Clinical Practice and Research

“…The integration of AI into nursing practice requires careful attention to avoid worsening existing inequities (Gurevich et al, 2023). Unequal access to care can result in a lack of diverse and representative health data for training AI algorithms, inadvertently perpetuating biases, particularly if the available data primarily reflects a mostly White population which can result in biased outcomes for non-White populations (Gurevich et al, 2023). Although AI offers numerous opportunities, it requires frequent assessment for potential ethical, privacy, and security concerns .…”

“…By analyzing extensive datasets, ML-driven analytics provides insights that can identify health risks, refine diagnoses and treatments, and prevent adverse health events (Chilla, 2023). This technology is increasingly used in nursing for patient monitoring, chronic disease management, and emergency preparedness (Hwang et al, 2022; Pailaha, 2023). These developments highlight the significant impact of ML-driven predictive analytics on health care and its potential to revolutionize nursing practice.…”

Community perspectives on AI/ML and health equity: AIM-AHEAD nationwide stakeholder listening sessions

“…SDoH are social factors such as built environment, access to healthcare, education, and distribution of resources [10,11]. SDoH influence health directly, and there are structures and systems that disadvantage certain groups of people in SDoH domains which can lead to health disparities.…”

“…Additionally, when thinking about the AI/ML data pipeline, it is not necessarily that the algorithms are being intentionally created with bias, but that the data feeding into the algorithms is inherently biased due to a lack of data from underrepresented groups, or data absenteeism, and an overrepresentation of data from more populous groups [9,13]. Therefore, it is important that AI initiatives keep an eye toward health equity and advocate for system level changes to better govern implementation of data science approaches [11] Algorithms have the potential to significantly impact patients and communities, and it is vital that equity is a top priority [5]. ML approaches to equity can be described through the concept of "fairness" [9].…”

“…An iterative codevelopment process between stakeholders, end-users, technical experts, and researchers can be key in developing AI/ML solutions to health disparity questions [17]. Although this can be more resource intensive for data scientists and researchers developing algorithms on the frontend, the opportunity to find innovative ways to dismantle health inequities would likely save on downstream healthcare costs once implemented [11]. Nevertheless, engaging stakeholders in AI/ML development has been a challenge thus far [18].…”