Evangelos Oikonomou

A bit about yourself

I grew up in Athens, Greece, where I completed my medical school training before enrolling in a DPhil in Medical Sciences at the University of Oxford. Following my DPhil studies, I moved to the United States where I have since continued my clinical training in internal medicine and cardiology in addition to research training in data science and biomedical computer vision.

Summarise the research in your DPhil

My doctoral project in Professor Charalambos Antoniades' lab integrated a range of computational techniques, including radiomic and epidemiological analyses, to better understand how the perivascular adipose tissue interacts with the coronary vasculature.

This work established a novel role for perivascular adipose tissue as a sensor of vascular inflammation and characterised the ability of coronary computed tomography imaging as a scalable approach to detect early signs of atherosclerosis. Across epidemiological analyses in large cohort studies we demonstrated the incremental predictive value of this approach for future adverse cardiovascular events.

These analyses were further supported by radio-transcriptomic studies in which we described the dynamic nature of our imaging biomarkers in response to acute coronary syndromes or treatment with anti-inflammatory therapies. This work has formed the basis of a rapidly expanding body of literature on using perivascular fat imaging as a window into vascular disease and biology.

About your current job and the path you took to get there

Following my DPhil studies, I joined the physician-scientist research pathway at the Yale School of Medicine. This track allowed me to complete my clinical training in internal medicine and cardiology while securing time and resources to expand my training in data science, medical informatics, and applied computer vision.

At Yale, I joined the Cardiovascular Data Science (CarDS) Lab founded by my mentor Rohan Khera, where I have since contributed to a broad portfolio of research spanning statistical machine learning, and applied computer vision technologies to diagnose cardiovascular disease through scalable technologies. Our work has focused on the development of statistical machine learning approaches to characterise individualised treatment effects in clinical trials and inform smart adaptive designs through dynamic predictive enrichment.

I have also built on the skills I acquired during my DPhil to develop deep learning methods for the automated characterisation of various forms of cardiomyopathy (i.e. cardiac amyloidosis, hypertrophic cardiomyopathy) as well as valvular disease (i.e. aortic stenosis, rheumatic heart disease) from scalable, point-of-care ultrasound technology.

About what helped you and how you decided to get into this area

During my DPhil studies, I was inspired by the ability of machine learning and computer vision approaches to detect and quantify subtle changes in the radiomic phenotype of various cardiovascular structures, and how these patterns can be used to generate scalable, accessible, objective, and more precise biomarkers of cardiovascular states.

I was fortunate enough to pursue my doctoral project in a supportive environment that encouraged me to explore novel pathways and analytic techniques, and test these both retrospectively and prospectively across large clinical studies to address an unmet clinical need.

Anything extra you found you needed to know, learn along the way or wish you had done differently

Before starting my DPhil studies, I had a very different career path in mind, both in terms of research focus and the kind of skills that I wished to acquire during my studies. But during my doctoral project, I grew to appreciate the dynamic nature of biomedical research and found motivation in exploring ideas and concepts that were previously not as well defined. This prompted me to learn fundamental concepts about my field, but also innovate and define new analytic methods.

Ultimately, the most important skillset I acquired during my DPhil was learning how to be curious and invest in skills that are transferrable across disciplines, thus allowing me to pursue further investigations across more diverse areas in my subsequent steps.