Harry Triantafyllidis

Modules:

• Data Science (MSc)

I hold a PhD in Applied Computer Science (Operations Research) from the University of Macedonia, Greece, with co-supervision from the Massachusetts Institute of Technology. Following this, I held research positions at University College London and the University of Oxford, before being awarded an MRC Early Career Research Fellowship at the School of Public Health, Imperial College London. 

My research develops advanced optimisation, network modelling, and graph-based machine learning methods to improve decision-making in healthcare and global health. One strand of my work focuses on computational biomedicine, including disease phenotyping, functional stratification, and the integration of genomic and transcriptomic data to study heterogeneity across cancer and other non-communicable diseases. A second strand focuses on dynamic health systems modelling, where I develop digital twin frameworks and temporal graph learning methods for operational problems such as hospital antimicrobial resistance, surveillance, and intervention planning. Together, these strands aim to bridge methodological innovation and practical impact, producing analytical tools that inform policy, improve patient stratification, and support clinically relevant decision-making.

My research combines optimisation, network modelling, and machine learning to study complex biomedical and healthcare systems. I work across two connected areas: computational biomedicine, where I develop methods for disease phenotyping and multi-omic integration; and dynamic health systems modelling, where I build digital twin and temporal graph learning frameworks for problems such as hospital antimicrobial resistance.

I have developed HarmonizeR, a platform in R and Python for integrating genomic and transcriptomic data with statistical, optimisation, and graph-based learning methods to study disease heterogeneity, especially in cancer. In parallel, I am developing a hospital AMR digital twin platform that combines mechanistic simulation and temporal graph learning to forecast AMR dynamics and support intervention planning in healthcare settings