Minh-Khoi Pham

I'm an pursuing my Ph.D. of Computer Applications at Dublin City University, Dublin, Ireland. My current research focuses on developing and critically evaluating foundation model approaches for clinical risk prediction using large-scale structured electronic health records. I am particularly interested in how pretrained models can support robust zero-shot and few-shot inference for rare and imbalanced outcomes, such as healthcare-associated infections, while reducing reliance on task-specific labeling and retraining. My work examines how temporal structure, feature heterogeneity, patient interaction networks, and retrieval mechanisms influence model performance, generalisability, and failure modes in real hospital settings.

I'm passionate about discovering new methods, state-of-the-art models, and applying or reimplementing them. Technology evolves daily, so I dedicate effort to exploring as many new innovations as possible. All my code is open source on my GitHub for educational purposes and for fellow researchers and developers to reference.