Resumen de Development and validation of strategies to mitigate the risk of hypoglycemic events in type 1 diabetes mellitus
Type 1 diabetes mellitus (T1DM) is a chronic condition resulting from the destruction of insulin-producing pancreatic beta-cells due to an autoimmune process that may occur at any age, but tends to develop in childhood and adolescence. The causes of this destructive process are not fully understood, and it results in an impairment of insulin secretion by the pancreas. Therefore, individuals afflicted with T1DM require an uninterrupted supply of exogenous insulin to survive. If insulin is not provided, blood glucose levels tends to rise to dangerous high levels (hyperglycemia) and subjects can face several complications associated with the damage and failure of various organ systems. In the 90's, the diabetes control and complications trial demonstrated that intensive insulin therapy is able to reduce the risk of these complications, but at the same time increase the risk associated with the occurrence of low blood glucose levels (hypoglycemia), which in severe cases may lead to seizures and even death.
Current intensive insulin therapy is based on the delivery of insulin through multiple daily injections (MDI) or through continuous subcutaneous insulin infusion by an insulin pump. Combining an insulin pump and blood glucose measurements provided by a continuous glucose monitoring (CGM) system arose the first automated devices to manage T1DM, by interrupting insulin infusion to reduce the risk of hypoglycemia. More recently, the so-called Artificial Pancreas emerged, which is a closed-loop system to regulate insulin infusion automatically based on a control algorithm. However, the great variability observed in glucose metabolism and big disturbances caused by meals and physical activity are major hurdles to achieve tight glycemic control.
The proposals of this work pay special attention to improve glycemic control in patients with T1DM under real-life conditions, focusing in hypoglycemia avoidance. Therefore, novel strategies are presented to improve the performance of the Artificial Pancreas device under development by the Spanish Consortium on Artificial Pancreas and Diabetes Technology, to cope with announced meals and aerobic exercise. Firstly, a novel methodology to adjust in real time a specific constraint of the control system, taking into account a previous approach validated in clinic, have been presented and tested in simulation. Secondly, mitigation strategies to reduce the risk of exercise-induced hypoglycemia have been proposed and extensively validated in silico and preliminary results of a clinical trial with real patients are also presented. Additionally, a new decision support system is proposed to predict the occurrence of nocturnal hypoglycemia in adult patients under MDI therapy. Data from ten adults have been collected in order to obtain predictive models using machine learning algorithms, considering CGM, insulin and physical activity data.
