Talk title: Unsupervised Learning of Health Through Molecular and Physical Activity Trajectory Patterns
Speaker biography: Dennis Wang is the Academy of Medical Sciences Professor (Chair in Data Science) at Imperial College London. He is also a Senior Principal Scientist at A*STAR research institutes in Singapore. Dennis’ team focuses on identifying biomarkers of health outcomes across the human lifespan and computational predictions to recommend diagnoses and treatments for multimorbidities (cardiovascular, neurological, infectious and oncological conditions). They are also interested in developing machine learning approaches that leverage diversity from large cohort datasets to accelerate risk assessment and drug development
Talk summary: Diagnosing individuals with complex or overlapping health conditions remains a major challenge, as symptoms are often non-specific and do not map neatly onto known disease categories. Recent advances in omics technologies and wearable
devices now allow us to collect detailed molecular and physical activity data at scale. However, interpreting this high-dimensional information, especially when clinical labels are uncertain, is not straightforward. In this talk, I will present how unsupervised machine learning methods can help uncover hidden patterns in such data, enabling the classification of heterogeneous health states and the discovery of meaningful biomarkers. I will illustrate these approaches using examples from child growth and cardio-respiratory health studies conducted in population cohorts in the UK and Singapore, and discuss how they may support earlier detection and better targeting of interventions in real-world settings.