Improving Fairness in Aging-Related AI: A Conceptual Model for Mitigating Biases

Artificial intelligence (AI) has shown immense potential to revolutionize healthcare, particularly within gerontology, by improving accuracy, efficiency, and personalized care through its capacities of analyzing high-dimensional patient-level data. However, significant concerns are emerging about biases embedded in AI systems, which can inadvertently exacerbate existing healthcare gaps associated with protected characteristics such as race, ethnicity, gender, or socioeconomic status. This paper addresses these critical issues by presenting actionable strategies designed to enhance fairness and equity in AI applications within gerontology through an innovative conceptual model of de-biasing within the social context. The transformative potential of AI alongside prevalent biases is illustrated through three representative scenarios—disease diagnosis, chronic condition management, and geriatric rehabilitation—highlighting real-world implications. Furthermore, major sources of bias throughout the AI lifecycle are presented, including biases stemming from unrepresentative training data, inappropriate AI model selection, and insufficient diversity in user feedback. Finally, we introduce comprehensive and evidence-informed de-biasing approaches guided by our proposed model, providing practical frameworks and solutions for creating equitable, effective, and inclusive AI-driven gerontological care.