Personalising Models of Human Blood Sugar Using SciML - JuliaCon 2024

Explore a conference talk on personalizing models of human blood sugar using scientific machine learning (SciML). Dive into the application of precision nutrition for improving metabolic health through personalized dietary recommendations. Learn about the Eindhoven Diabetes Simulator (EDeS), a mechanistic model designed to capture blood glucose and insulin responses to standardized meals. Discover how neural universal differential equations are employed to enhance the model's ability to describe individual variability in meal responses. Examine the integration of neural networks into specific model components to increase flexibility while maintaining biological bias. Gain insights into training universal differential equations on sparse human data using physiology-informed regularization and the recovery of personalized glucose-driven insulin production with conditional neural networks. Understand how this hybrid model approach contributes to a better comprehension of individual variations in blood sugar responses to meals and provides interpretable measures for developing personalized dietary recommendations.