About
I am a physician-scientist and Assistant Professor at Stanford Medicine, jointly appointed in the Department of Epidemiology and Population Health and the Division of Computational Medicine. My work sits at the intersection of medicine, data science and technology, and turns on what it takes for AI to change how care is actually delivered.
I trained as a physician at the University of Edinburgh and later in epidemiology and statistics at Stanford, before working at Google Health and then at Prolaio, where I was the first employee and VP of Clinical Data Science, leading its work on remote patient monitoring and clinical decision support. I returned to Stanford in 2026 to build a lab that translates advances in AI, longitudinal clinical data and sensing into better ways to understand, anticipate and care for patients.
Outside work I cook Greek and Cypriot food, powerlift, play tennis, and explore new places — usually one meal at a time.
The department published a longer Q&A on how I got here.
Vision
Most clinical decisions are still made from intermittent snapshots — a clinic visit, a laboratory test, a measurement taken every few months. Yet patients generate signal continuously, through their physiology, their behaviour, their medical records and the devices they wear. Advances in multimodal and generative AI make it possible to bring those signals together, learn what is normal for a given patient, recognise when something meaningful has changed, and help clinicians act earlier. That is the medicine I want to help build: continuous, personalised and predictive.
My lab develops the methods and systems to make that possible, working across longitudinal clinical data, wearable and at-home sensing, multimodal modelling and clinical decision support. What interests me most is the step from prediction to action. A model that performs well on a benchmark is not the goal; a system that follows a patient over time, surfaces what matters and changes a therapeutic decision is.
Much of my earlier work was in meta-research, reproducibility and the reliability of biomedical evidence. I still care about it deeply, but I now see it as the methodological foundation of the lab rather than its destination. It is why we hold our own clinical AI to the standards we would demand of a trial, and why we try to read the literature honestly enough to act on uncertain evidence without overstating what we know.
The larger ambition is to turn advances in AI into clinical intelligence that is timely, trustworthy and available well beyond the best-resourced hospitals. I want the lab to combine the intellectual freedom of academia with some of the energy of an incubator: pursue ambitious ideas rigorously, test them in real clinical settings, and help the best of them become tools that clinicians and patients actually use. If that is the kind of work you want to spend the next few years doing, write to me.
Papers
- Deep Learning for Epidemiologists: An Introduction to Neural Networks American Journal of Epidemiology · 2023
- Assessment of transparency indicators across the biomedical literature: how open is open? PLOS Biology · 2021
- Altmetric Scores, Citations, and Publication of Studies Posted as Preprints JAMA · 2018
Three of the ones I would pick first. The full record is on Google Scholar and in my CV.
CV
Updated August 2026.