Investigators will present results from a variety of studies looking at ways to predict, manage, and potentially prevent hypoglycemic events during Monday’s Oral Presentations session Hypoglycemia and Technology, which will begin at 2:15 p.m. in S-10 (South, Exhibition Level). ADAMeetingNews.org asked several of the presenting authors to preview their study presentations.
How to Use rtCGM Data to Predict Future Severe Hypoglycemia?
Norbert Hermanns, PhD, Research Institute Diabetes Academy Mergentheim in Bad Mergentheim, Germany
Study Background/Objectives
Dr. Hermanns: It has repeatedly been shown in patients with diabetes with hypoglycemia problems that continuous glucose monitoring (CGM) can reduce mild as well as severe hypoglycemia without compromising glycemic control. Whereas we have evidence from methodologically sound studies that CGM is an effective treatment for people with diabetes, the diagnostic value of CGM in identifying people who are at risk for severe hypoglycemia is currently not sufficiently explored.
In our study, we combined data of two landmark CGM studies to predict future severe hypoglycemia during the trial from CGM baseline data. We analyzed the data of 127 study participants with type 1 diabetes randomized to the control group using self-monitoring of blood glucose during the trials. We wanted to know which cutoff values of exposure to low glucose could predict future severe hypoglycemia during the trials with an optimal sensitivity and specificity.
Machine Learning to Predict Hypoglycemic Events from Continuous Glucose Monitoring Data
Yuan-Chi Chang, PhD, IBM Research in Yorktown Heights, NY
Study Background/Objectives
Dr. Chang: Our research is focused on the application of machine-learning algorithms to learn and predict elevated risk of hypoglycemia from patients’ CGM data collected in the real world outside of the clinical environment. We believe by providing CGM users with this information well ahead of time, users may be more alerted and watch for their glucose level more attentively. The work is enabled by a large collection of CGM data exceeding 120 million glucose readings from over 10,000 de-identified users, in conjunction with machine learning and real-time analytics processing. This work was developed under the collaboration of IBM Research, IBM Watson Health, and Medtronic Diabetes.
Directness and Sustainability of rtCGM Effects on Hypoglycemia: A Secondary Analysis of the Hypode Study
Dominic Ehrmann, PhD, Research Institute Diabetes Academy Mergentheim in Bad Mergentheim, Germany
Study Background/Objectives
Dr. Ehrmann: For many people with type 1 diabetes, hypoglycemia is still a burdensome and limiting factor of diabetes therapy. Continuous glucose monitoring has proven to be an effective tool to reduce the exposure to hypoglycemic values, as well as severe hypoglycemia. Even in people with type 1 diabetes and heightened problems with hypoglycemia (e.g., reduced hypoglycemia awareness), CGM is effective to avoid hypoglycemia. However, what’s still missing is a better understanding of how and how quickly CGM can reduce the exposure to hypoglycemic values, and how stable such an initial effect is. With outcome studies usually comparing an initial baseline phase with blinded CGM to the outcome phase usually six months later, we wanted to look at the directness and sustainability of CGM effects in the time between these two phases.
Impact of a Hybrid Closed-Loop Insulin Delivery System on Hypoglycemia Awareness in Individuals with Type 1 Diabetes
Marie-Anne Burckhardt, MD, Perth Children’s Hospital and Telethon Kids Institute in Perth, Australia
Study Background/Objectives
Dr. Burckhardt: People with type 1 diabetes who don’t feel when their blood sugar level is low are at risk of developing severe hypoglycemia. This is called impaired awareness of hypoglycemia and develops as a result of recurrent hypoglycemia. We know that hypoglycemia awareness can be restored when hypoglycemia is avoided carefully. Hybrid closed-loop systems with automatic glucose sensing and insulin delivery reducing patient intervention—currently a hot topic in the rapidly developing field of diabetes technology—offer a means to achieve this goal without the need to maintain higher blood glucose targets with less impact on glycemic control.
[gap height=”40″]