Mathematical modelling in diabetes: the ups and downs, peaks and pitfalls

1 Nov 2013

David Matthews
Oxford Centre for Diabetes Endocrinology and Medicine, and
Harris Manchester College
University of Oxford


Every medical clinician knows about insulin and blood sugar. Blood sugar is controlled by insulin. The more insulin that you give the lower the blood glucose will go. Given how simple this seems it is remarkable that in practice insulin treatment becomes so complex that it is often managed only by specialists.

The simplistic view about the interrelationships between insulin and glucose fall apart almost immediately one meets clinical reality. The effect of insulin differs dramatically from person to person. Some people with type 1 diabetes need 4 units of insulin a day and some people with type 2 diabetes might need 400 units a day. Insulin has multiplicity of actions – it not only lowers glucose clearance but affects triglyceride metabolism, hepatic glucose output, hormone sensitive lipase, growth, appetite and renal and vascular function. In the body in normal subjects and those with type 2 diabetes insulin secretion is modulated not only by glucose but also by amino acids, non esterified fatty acids, vagal stimuli, glucagon, somatostatin and steroids.

Understanding how the body achieves even a semblance of homeostasis turns out to be a complex problem. Modelling insulin secretion in man is a challenge that yields many insights and some solutions. Without mathematical models measuring insulin concentrations in blood turns out to be almost futile, and is never routinely undertaken in clinical practice (in contrast to, say, measuring thyroxin in hypothyroidism).

I will discuss the paradox of insulin and insulin resistance, non steady state insulin secretion and some insights from mathematical modelling. I will outline a view about cellular democracy in the otherwise autocratic world of homeostasis.

current theory lunch schedule