Gaussian process modelling of blood glucose response to free-living physical activity data in people with type 1 diabetes

Download

PDF - Publishers print Download (337Kb)

Description/Abstract

Good blood glucose control is important to people
with type 1 diabetes to prevent diabetes-related complications.
Too much blood glucose (hyperglycaemia) causes long-term
micro-vascular complications, while a severe drop in blood glucose
(hypoglycaemia) can cause life-threatening coma. Finding
the right balance between quantity and type of food intake,
physical activity levels and insulin dosage, is a daily challenge.
Increased physical activity levels often cause changes in blood
glucose due to increased glucose uptake into tissues such as
muscle. To date we have limited knowledge about the minute
by minute effects of exercise on blood glucose levels, in part due
to the difficulty in measuring glucose and physical activity levels
continuously, in a free-living environment. By using a light and
user-friendly armband we can record physical activity energy
expenditure on a minute-by-minute basis. Simultaneously, by
using a continuous glucose monitoring system we can record
glucose concentrations. In this paper, Gaussian Processes are
used to model the glucose excursions in response to physical
activity data, to study its effect on glycaemic control