ohiot1dm

Patient-specific blood glucose prediction using deep learning, considering the challenges of "small dataset" and "data imbalance"

"Personalised Short-Term Glucose Prediction via Recurrent Self-Attention Network" in CBMS 2021

Glycemic-aware and architecture-agnostic training framework for blood glucose forecasting in type 1 diabetes

"Jointly Predicting Postprandial Hypoglycemia and Hyperglycemia Using Continuous Glucose Monitoring Data in Type 1 Diabetes" in EMBC 2023

Causal mediation analysis of CGM data using autoencoder-learned representations to study how meal carbohydrates affect postprandial glucose through insulin bolus decisions.

Pipeline for Paper 2: CGM peak detection, meal-window exclusion, hourly relative BG change, 24-h composites, medians & boxplots.

Pipeline for Paper 1: CGM/insulin/activity, hourly relative BG change, basal active-insulin simulation, merge & plots.