Projects
- Machine learning-assisted continuous glucose and ketone monitoring for diabetic ketoacidosis - University of Waterloo
- Supervisor - Prof. Mahla Poudineh
- September 1, 2022 - August 13, 2024
- Thesis link
- Type 1 diabetes impacts millions worldwide, requiring rigorous monitoring of blood glucose to avoid severe complications like hyperglycemia and diabetic ketoacidosis. Continuous glucose monitoring (CGM) devices measure glucose levels from interstitial fluid (ISF) in real-time, facilitating effective treatment. While existing machine learning models excel in short-term glucose predictions, they often overlook the need for long-term forecasting essential for optimizing insulin therapy. This work introduces an encoder-decoder model that extends the forecasting horizon from 1 to 3 hours, potentially improving insulin delivery accuracy. Additionally, ISF-based sensors face challenges with sensing delays due to varying glucose and other analyte transfer times from blood to ISF. This work also examines these delays using decision-tree algorithms and an in vivo diabetic rat study using the cases of continuous glucose and continuous ketone monitoring to enhance sensor accuracy and enable more personalized monitoring of such analytes.
- Simulation of atrial fibrillation in a computer model of the heart - University of Montreal