The terms bias and fairness are often mistakenly used interchangeably in discussions about AI systems. This confusion is particularly problematic in healthcare, where biased data and models can lead to unfair treatment recommendations and diagnostic inaccuracies. Measuring bias and fairness is crucial to ensure equitable patient care and accurate diagnostics. This paper presents an approach to (1) quantify data biases using appropriate data-based metrics, and (2) use the findings of this analysis to guide a bias-aware synthetic data generation process. We also explore (3) the use of bias-aware synthetic data generators to address biases in three different use cases: one focusing on balancing labels, one on balancing sensitive attributes, and one addressing both simultaneously. Lastly, we propose an evaluation framework to assess data- and model-based metrics for quantifying fairness improvements. Our preliminary findings suggest that synthetic data generated using bias-aware generators can improve inclusivity, balance performance with fairness, and reduce representation bias, potentially contributing to more equitable healthcare outcomes. This study underscores the importance of aligning synthetic data generation with the intended research objectives and highlights the potential of robust synthetic data generators to meet diverse healthcare needs, thereby improving their overall utility.