The increasing deployment of Artificial Intelligence (AI) systems in healthcare has raised significant concerns about bias, fairness, and ethical implications of automated decision-making. While medical AI systems offer capabilities for diagnosis, treatment planning, and patient care, they also inherit and potentially amplify biases present in training data, particularly affecting underrepresented populations. This paper presents a comprehensive framework for auditing AI systems in medical domains through functional audit processes that systematically evaluate bias and fairness. Our approach integrates three key components: data quality assessment, fairness analysis, and application of explainable AI (XAI) techniques. We demonstrate the practical application through the auditing of a machine learning model to predict COVID-19 patient mortality. The results reveal disparities in model performance across different demographic groups.