Algorithmic fairness and bias mitigation in clinical machine learning for equitable patient outcomes

Abstract

In recent years, the integration of machine learning algorithms into clinical settings has shown immense potential for improving healthcare outcomes. However, concerns regarding fairness and equity in machine learning models have garnered increasing attention, particularly in healthcare where biased algorithms can perpetuate existing disparities. This thesis investigates the role of fairness-aware algorithms in addressing these issues within clinical machine learning applications. Through case studies and empirical analyses, this research explores how biases manifest and impact model performance across diverse patient populations, highlighting the challenges and opportunities in promoting fairness within clinical machine learning. Subsequently, drawing on datasets from multiple healthcare institutions, we propose and assess the effectiveness of fairness-aware techniques in advancing equitable healthcare outcomes. Ultimately, this thesis contributes to the ongoing dialogue on fairness in machine learning, providing insights and recommendations for the development of ethically sound and socially responsible machine learning algorithms in healthcare.

Ähnliche Arbeiten