Interview University of Manchester
Professor Niels Peek
University of Manchester
Leader of Work Package 1 “Risk Score Development”
Background: Niels Peek is Professor of Health Informatics at the University of Manchester. He has a background in Computer Science and Artificial Intelligence, and his research focuses on translational data science for clinical risk prediction, personalised and precision medicine, patient safety, and multimorbidity. He is the former President of the Society for Artificial Intelligence in Medicine, a member of the European Lab for Learning and Intelligent Systems (ELLIS), a fellow of the Alan Turing Institute, and a fellow of the American College of Medical Informatics. Since 2021, he directs the Christabel Pankhurst Institute for Health Technology Research and Innovation, which promotes needs-led health technology research and innovation and provides end-to-end support for translation of new technologies into healthcare practice.
In the PROFID project, Niels Peek is the lead for Work Package 1 “Risk score development”.
1. Could you tell us a bit about yourself?
“I am a Dutch computer scientist and grew up in the south of the Netherlands (near Eindhoven), did my training in Utrecht, and later worked at the Academic Medical Center in Amsterdam. I moved to Manchester, UK, seven years ago, which has been a great experience. I very much like the people and the “Northern” spirit here. North England is also fantastic for hiking, and Manchester is of course brilliant if you like football.”
2. Could you describe your work and how it is related to PROFID?
“In Manchester, I lead a large group of health data scientists, with risk prediction being one of the main focus areas. We conduct both theoretical and applied research, and very often there is cross-fertilization between the two, which is also why we believe that it is important to do both. PROFID is one of our more applied projects, but it has already provided us with many ideas for theoretical work in the future.”
3. What is your role in the project?
“I lead Work Package 1 “Risk score development”, where we apply cutting-edge analytical methods from machine learning and statistical modelling to a very large dataset of patients with myocardial infarction or ischemic cardiomyopathy and reduced left ventricular ejection fraction. Our goal is to develop a new model for predicting the risk of sudden cardiac death in these patients.”
4. Why do you believe PROFID is important?
“Sudden cardiac death is one of most common causes of death, especially among people with myocardiac infarction. There exists a very effective preventative treatment for sudden cardiac death (implantable cardioverter defibrillators, ICDs) but we are not very good at deciding who should receive that treatment, and we cannot give an ICD to everyone. Now is the time to change that: using large datasets and advanced analytical methods we can assess which patients are really at risk of sudden cardiac death and therefore need ICD treatment.”
5. What is the best part of your work in PROFID?
“In PROFID we have the strongest possible consortium of researchers to address this problem, and I do consider it a true privilege to work with this consortium. I am also privileged to have fantastic people in my team in Manchester, and to work with brilliant people from a London based SME, Spectra Analytics. Working with so many bright people is most inspiring, and it gives confidence that we will achieve good results and have real impact on patient outcomes.”
6. What are the main challenges in your work in PROFID?
“The data that we work with are highly complex and heterogeneous. We spent most of our time understanding, describing, and summarizing the data. That is laborious, time consuming, and often boring. But we have to do this thoroughly and make sure that there are no mistakes. Also, the fact that the datasets are derived from many different trials, registries, and cohort studies means that they are all very different and a lot of work is needed to harmonise them. Finally, the size of the datasets poses challenges: some of our analytical methods take several days, if not longer, to complete – and we often have to repeat them many times.”
7. What do you believe to be the biggest milestone PROFID has achieved so far?
“We have unequivocally shown that left ventricular ejection fraction is a poor predictor of risk of sudden cardiac death in the PROFID patient population. This means that existing clinical guidance, which is entirely based on left ventricular ejection fraction, is inadequate and urgently needs to be informed by better risk stratification methods.”