Prof Mark Kelson
Associate Professor of Statistics for Health
Telephone: 01392 722562
Extension: (Streatham) 2562
I am a statistician. A biostatistician as it happens. A medical statistician really. Or should it be data scientist? Am I in data analytics? I’m a biostatdatscientanalyst.
Now that’s cleared up I can tell you what I’m interested in: data. And people*. And science.
Our progression as a species is due to our aptitude for handling information, from the pattern recognition that helped us spot predators in prehistory to our capacity for abstract thought that allowed us solve our most difficult problems (someday we may even figure out how Dwayne "the Rock" Johnson is our highest paid actor). All this requires an ability to process and make sense of data. This is the work of a biostatdatscientanalyst.
I work in between clinical trials, medical statistics, causal inference, reproducibility and data science. My work focusses on two main application areas: physical activity and mental health.
I currently work on a project exploring drug-vaccine interactions, a project exploring the impact of violence reduction units in the UK and a systematic review exploring interventions for peopl experiencing homelessness.
I am interested in bringing the rigour, methodology and philosophy of clinical trials into settings outside of trials, as well as delivering the insights that causal inference brings into the trials setting. If this sounds good to you, do get in touch.
My vision is to bring these ideas to bear on the field of physical activity. Physical activity is fascinating. Everyone knows it’s good for you right? But how good? And for what? And how do you measure it anyway? All of this interests me. I come from a clinical trials background and many of the trials I work(ed) on (have) include(d) elements of physical activity for health, be that functional ability, weight loss or mental health. I am particularly interested in the analysis of accelerometry data.
I am also interested in mental health. Mental health measurement and intervention is another difficult problem and so is grist to the mill of a biostatdatscientanalyst. I have milled the grist out of: common mental disorders, bipolar disorder, suicide and post-traumatic stress disorder.
Reproducibility is another idea that interests me. How do we ensure that the work that we do is of high quality, can be repeated by others and is openly available? It all sounds like sensible stuff, but the practicalities of making this happen are not straightforward.
Methodologically I am interested in multilevel modelling, meta-analysis and R programming.
I contribute to teaching on a second year mathematics prgramme co-delivering “Statistical modelling and Inference”. I lead an MSc module called “Working with Data”. I lead the Applied Data Science and Statistics programme. I am the academic lead for an undergraduate degree apprenticeship in Data Science run in partnership with Exeter College. I also run meta-analysis training with Cardiff University and University of Exeter.
I am a research Fellow with the Alan Turing Institute.
I am the Reproducibility Network institutional lead for the University of Exeter and the Research Integrity Officer for the department of mathematics.
I am an Assistant Director for the Institute of Data Science and Artificial Intelligence with the remit of partnerships. I sit on the IDSAI management group.
I am an editor for Statistical Methods in Medical Research http://journals.sagepub.com/home/smm and statistical editor for the International Journal of Behavioral Nutrition and Physical Activity https://ijbnpa.biomedcentral.com/
I am on the middle career researcher section of the Society for Social Medicine.
I am available to supervise undergraduate projects on physical activity or mental health.
I blog infrequently at SignificantlyStatistical
I sit on a number of committees for trials including:
PACE-HD: Physical Activity and Exercise Outcomes in Huntington’s Disease
Society for Social Medicine
European Public Health Association
* In so far as they produce data