Predicting Postprandial Glycemic Responses With Limited Data in Type 1 and Type 2 Diabetes.

Area of Science:

• Diabetes management and metabolic research
• Biomedical data science
• Human physiology

Background:

• Predicting postprandial glycemic responses (PPGRs) is crucial for diabetes management.
• Existing methods face challenges with inter- and intraindividual variability.
• Accurate prediction often requires invasive data, such as microbiome analysis.

Purpose of the Study:

• To investigate the predictability of PPGRs using limited, non-invasive data.
• To identify sources of intraindividual variability in glycemic responses.
• To assess the impact of different dietary features on glycemic control.

Main Methods:

• Utilized continuous glucose monitor (CGM) and meal logging data from Type 1 (T1DEXI) and Type 2 (ShanghaiT2DM) diabetes cohorts.
• Employed dietary, demographic, and temporal features to predict 2-hour PPGR and peak glucose rise (Glu _max).
• Evaluated the predictive contribution of food categories versus macronutrients and explored intraindividual variability.

Main Results:

• Achieved prediction accuracy comparable to prior studies for PPGR (R=0.61-0.72) and Glu _max (R=0.64-0.73) without invasive data.
• Incorporating food category features improved prediction accuracy over macronutrients alone.
• Identified time of day (lunch/dinner) and menstrual cycle phase (perimenstrual) as significant sources of intraindividual glycemic variability.

Conclusions:

• Glycemic responses to meals in individuals with Type 1 and Type 2 diabetes can be predicted effectively using limited, non-invasive data.
• Personalized training data and invasive physiological information are not essential for accurate prediction.
• Understanding variability factors like meal timing and physiological cycles enhances diabetes management strategies.