Fair machine learning in Healthcare

Machine learning is increasingly used in healthcare for disease diagnosis and risk prediction, but biased data can lead to unfair outcomes. This project explores how biased labels in electronic health records impact model performance, particularly across different demographic groups. By analyzing embeddings, evaluating model fairness, and assessing prediction disparities, we highlight key challenges in ensuring equitable data-driven healthcare analysis. Our findings emphasize the need for improved feature selection and bias mitigation strategies to promote fairness in medical machine learning models.

Click presentation slides and project GitHub page to view it.