We talk to John Dennis, a data science researcher working in the University of Exeter. He used data from 1 million people with Type 2 diabetes to discover how the clinical characteristics of a patient alter the glucose lowering with different treatments. John’s 5 drug model, published in the Lancet in late February 2025, uses simple clinical information to identify, for the first time, the best glucose lowering treatment for a patient. The model selected best treatment lowers the glucose most, doubles the time before another medication is needed and reduces long term complications of diabetes.
John Dennis is an Associate Professor of Health Data Science and Wellcome Trust Fellow, working in the Clinical Diabetes Group at the University of Exeter Medical School. He originally came from working for 5 years in finance/insurance industry role in London using data to assess actuarial risk of clinical features. He moved to the University of Exeter and did a PhD with Andrew Hattersley and Bev Shields as supervisor. His talents as a data scientist were immediately obvious and he started to apply innovative solutions to the problem of how to choose the optimal drug to lower an individual patient’s blood glucose. Even as a PhD student he was leading the work of the UK MASTERMIND consortium that examined how to bring precision prescribing to Type 2 diabetes by using routine clinical features available on all patients.
John is applying data science and advanced statistical methods to better understand the benefits and risks of each drug option, and to develop decision aids to help health professionals choose the best drug for individual patients. He is also developing a more general methodological framework for precision medicine research combining data from electronic health records and clinical trials. He has published extensively see here https://scholar.google.com/citations?user=llh3TCkAAAAJ&hl=en and has been invited to give lectures throughout the world including at the main diabetes conferences the ADA and EASD. He has won the EASD rising star award.
This episode is timed to be close to the publication in the Lancet of February 26th of his outstanding paper on a 5-drug model allowing the optimising of the treatment of people with Type 2 diabetes to have the best lowering of blood glucose levels. John has showed that only 20% of patients receive the optimal therapy The paper will have a link to it here on this website when it is published
The model is simple to use, utilises simple routine data and has been shown to be robust and reproducible in data from hundreds of thousands of Type 2 patients. It can be tried using this link here T2D treatment selection
