Researchers have developed a custom algorithm that predicts the impact of certain foods on an individual's blood sugar, according to a new study published in PLOSComputational Biology.
The algorithm has been integrated into an application, Glucoracle, which will allow individuals with type 2 diabetes to maintain a closer control of their glucose levels - the key to preventing or controlling the major complications of a disease that affects 8 % Americans.
Glucoracle allows the user to upload fingerstick blood measurements and a photo of a particular meal to the application, as well as a rough estimate of the nutritional content of the meal. This estimate provides the user with an immediate prediction of post-meal blood glucose levels. The estimate and forecasts are then adjusted for accuracy. The application starts generating predictions after it has been used for a week, allowing the data assimilator to learn how the user responds to different foods.
The algorithm uses a technique called data assimilation, in which a mathematical model of a person's response to glucose is regularly updated with observation data - blood sugar measurements and nutritional information - to improve the predictions of the model, explained the head of co-study George Hripcsak, MD, MS, Professor Vivian Beaumont Allen and President of Biomedical Informatics at the CUMC.
"Although we know the general effect of different types of food on blood glucose, the detailed effects can vary greatly from one person to another and for the same person over time," said lead author David Albers, PhD, Associate Researcher in Biomedical Informatics at the Columbia University Medical Center (CUMC).
"The data assimilator is continually updated with the user's food intake and blood glucose measurements, customizing the model for that individual," said Lena Mamykina, Ph.D., Assistant Professor of Biomedical Informatics At CUMC, whose team designed and developed the Glucoracle Application.
The researchers first tested the data assimilator on five people using the application, including three with type 2 diabetes and two without the disease. The predictions of the application were compared to actual measures of post-meal blood glucose and predictions of diabetic educators.
For the two non-diabetic individuals, predictions of application were comparable to actual glucose measurements. For the three subjects with diabetes, application predictions were slightly less accurate, possibly due to fluctuations in the physiology of patients with diabetes or parameter error, but were still comparable to predictions of diabetes educators.
"This assessment was designed to prove that it is possible, using routine self-monitoring data, to generate real-time glucose forecasts that people could use to make better nutritional choices. Make an aspect of self-management of diabetes that was almost impossible for people with type 2 diabetes more manageable.Now, our task is to make the data assimilation tool feeding the application even better " , said Dr. Albers.
The researchers believe that the application could be ready to be widely used within two years.
