PhD Dissertation Proposal by PHAM Hung Manh | Generative AI for Cardiovascular Health

Generative AI for Cardiovascular Health

ABOUT THE TALK

Cardiovascular disease remains a major global health challenge, making reliable and accessible monitoring increasingly important. Two of the most widely used signals for assessing cardiovascular function are the electrocardiogram (ECG), which records the heart’s electrical activity, and the photoplethysmogram (PPG), which measures blood-volume changes through optical sensing. Recent advances in generative AI have opened new possibilities for learning from these signals. However, important challenges remain in ensuring reliable performance on real-world signals, generalizing across diverse populations and healthcare settings, and enabling flexible interaction beyond narrowly defined prediction tasks. In this talk, I will present a progression of methods for building more reliable, transferable, and interactive cardiovascular AI. Drawing on research in signal quality enhancement and disentangled learning, large-scale multimodal representation learning, and language models, I will show how physiology-informed design can improve the full pipeline from sensing to reasoning, ultimately supporting more clinically useful AI systems for cardiovascular healthcare.

SPEAKER BIOGRAPHY

Hung Manh PHAM is a third-year PhD candidate in Computer Science, supervised by Prof. Pan Zhou and co-supervised by Prof. Dong Ma and Prof. Bin Zhu. He is also a visiting student at the University of Cambridge. His research focuses on machine learning for healthcare and biomedicine, particularly physiology-informed learning and medical foundation models. His broader interests include causal inference, interpretable learning, and trustworthy medical AI.