Bias Audit Framework and Data Diversification for AI Led Diagnostics

My project will consist of two components. The first part will be a practicum with a berlin-based startup that is building out a health AI compliance platform to serve as an intermediary layer for hospitals/clinics, payors, and HR within Germany and the EU. Through this practicum, I will gain hands-on experience regarding how to create AI regulatory and governance frameworks, including in areas such data privacy, risk management (for compliance regulations such as EU AI Act, GDPR, European Health Data Space) as well bias mitigation. The second part will consist of independent research that will focus on understanding the current data training methods for health/medical AI models to determine what the biggest challenges are to obtaining or creating (synthetic) diversified data, what risks this can pose, and how to develop a protocol for ethical, efficient and accurate data diversification.