Lead Investigator: JOSE GARCIA-TIRADO, University of Bern
Title of Proposal Research: Automatic Recognition of Main Glycemic Disturbances in Type 1 Diabetes (T1D)
Vivli Data Request: 9103
Funding Source: None
Potential Conflicts of Interest: None
Summary of the Proposed Research:
Type 1 diabetes (T1D) is a lifelong chronic metabolic disorder with a high economic, physical, social, and mental toll on both people with the condition and their caregivers. This autoimmune condition results in absolute insulin deficiency and a lifelong need for external insulin (injections) to regulate blood glucose concentration. Effective management of diabetes becomes challenging when confronted with external factors, such as meals, physical activity, and stress, which significantly impact blood glucose levels. Despite promising results on glucose management achieved from existing treatments, mechanisms to mitigate disruptions caused by these events remain elusive. Current findings estimate that almost 70% of people with T1D (80% in the US) do not reach recommended glycemic targets according to the American Diabetes Association (ADA). As such, we need to work tirelessly to offer reliable and affordable technological solutions to improve public health.
This project aims to develop an engineering methodology to digest a wealth of data in the T1D population (continuous glucose monitors -CGM-, insulin injections, activity trackers -heart rate, steps, and accelerometry data-, meal intake, and demographics) to recognize behavioral patterns with respect to meal intake and physical activity. Such information will allow us to tailor better treatments given the characteristics and behavior of the user.
We will accomplish the above by integrating mathematical models (models predicting glucose concentration) with sensor data to reconstruct the magnitude and duration of such glycemic disturbances (effects of meals and physical activity over glucose concentration). This will hold potential for personalized treatment strategies, facilitating timely interventions based on real-time data and paving the way for improved treatments on diabetes technologies.
Statistical Analysis Plan:
(1) Data will be first analyzed for completeness and continuity. CGM gaps <2h will be interpolated using a Piecewise Cubic Hermite Interpolating Polynomial (PCHIP). Insulin data gaps <2h will be interpolated using a zero-order holder (ZOH) with the last insulin value. Data gaps >2h will be excluded from the analysis. Missing meal estimates can be imputed at the investigator’s discretion with averaged meal values.
(2) Data will be pre-processed to guarantee format compatibility and alignment in time.
(3) We will reconstruct meal-like disturbances retrospectively from CGM and insulin records and physical activity-related disturbances from CGM, insulin records, and activity tracker data. To filter out any small deviation and focus on meaningful disturbances, we propose an automated disturbance detector using features characterizing the estimated disturbance values, d, and continuous glucose measurements, cgm, for each day of historical data collected. Once features are generated for each five-minute interval, the probability of a large disturbance (Pi_meal or Pi_activity), will be determined using logistic regression.
(4) Once major glycemic disturbances are detected, daily indicator signals are defined to group similar days into clusters (equal to one in the two hours following disturbance detections and zero otherwise). Using k-means with the hamming distance measure, these signals will be clustered. The number of clusters,
k, for each individual is based on which produced the highest Calinski-Harabasz score, maximizing cluster separation and cohesion.
This procedure will then produce disturbance signatures that can be fed into mathematical models to mimic the glycemic response to meal- and exercise-related events.
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
Public Disclosures:
CYPRIS, L.E. and GARCIA-TIRADO, J.F., 2024. 243-OR: Characterization of Behavioral Patterns Involving Hypoglycemic Episodes in Type 1 Diabetes—Results from the T1DEXI Study Cohort. Diabetes, 73(Supplement_1). Doi : 10.2337/db24-243-OR