Improving Blood Glucose Predictions for Patients with Type 1 Diabetes

Lead Investigator: Mark Woodward, January AI
Title of Proposal Research: Improving Blood Glucose Predictions for Patients with Type 1 Diabetes
Vivli Data Request: 9104
Funding Source: None
Potential Conflicts of Interest: None

Summary of the Proposed Research:

Diabetes mellitus refers to a group of metabolic disorders characterized by chronic hyperglycemia. The two major classifications are Type 1 Diabetes Mellitus (T1D) and Type 2 Diabetes Mellitus (T2D). T1D is an autoimmune disorder that leads to the destruction of insulin-producing pancreatic beta cells, and thus necessitates exogenous insulin administration. T2D arises from insulin resistance. In addition to T1D and T2D, there is also a condition called pre-diabetes, in which blood glucose levels are elevated above normal levels, but not to the extent to classify as T2DM.

It is estimated that more than 420 million individuals globally have diabetes; people with T1D comprise around 10% of these cases. In addition to these groups, approximately 470 million individuals are believed to have pre-diabetes. The large sizes of these groups underscore the benefit of improved glucose management techniques for nearly a billion individuals worldwide.

This proposed research seeks to augment the existing knowledge in the field of medical sciences and diabetes management/treatment, and to foster enhancements in patient care by modifying an established algorithm developed by January AI. This algorithm presently possesses the capability to predict blood glucose levels for healthy and pre-diabetic individuals in response to food intake and physical activity. Our intention is to adapt this algorithm to accurately predict glucose levels for individuals with T1D.

The machine learning algorithm devised by January AI utilizes data encompassing the patient’s dietary habits, physical activities, and blood glucose measurements (as tracked by continuous glucose monitoring: “CGM”) to accurately predict future glucose levels. The proposed research intends to expand the algorithm by incorporating data pertinent to individuals with T1D and employing machine learning methodologies to discern patterns and relationships that can bolster prediction accuracy.

The importance of this research arises from the understanding that proficient prediction of blood glucose levels can markedly improve the management of blood glucose and diabetes, which reduces the potential for complications like cardiovascular disease, renal failure, and retinopathy. The application of predictive algorithms to T1D will empower afflicted individuals to more effectively manage their condition, thereby enhancing health outcomes and overall life quality.

The proposed study will be conducted in a series of stages, commencing with data acquisition from participants diagnosed with T1D (this data is what we hope to acquire from Vivli, comprising the T1-DEXI and T1-DEXIP datasets). Once data has been acquired and processed, it will be integrated into the existing January AI algorithm. Subsequent application of an array of machine learning techniques will allow for the refinement of the model’s predictive capabilities. The measure of success of the project will be quantified by comparing the predicted glucose levels produced by the algorithm with the actual measurements obtained from the participants.

The proposed research holds the potential to significantly advance our understanding and management of T1D, enabling a more precise, personalized, and effective approach to care.

Requested Studies:

Type 1 Diabetes EXercise Initiative Pediatric Study (T1DexiP): The Effect of Exercise on Glycemic Control in Youth with Type 1 Diabetes
Data Contributor: Jaeb Center for Health Research Foundation, Inc.
Study ID: T1-DEXIP

Type 1 Diabetes EXercise Initiative: The Effect of Exercise on Glycemic Control in Type 1 Diabetes Study
Data Contributor: Jaeb Center for Health Research Foundation, Inc.
Study ID: T1-DEXI