Artificial Pancreas Control Algorithms Using the Hovorka Equations for Personalized Insulin Delivery in Children with T1DM

Abstract

Type 1 Diabetes Mellitus (T1DM) requires precise insulin regulation to prevent complications such as hyperglycemia and hypoglycemia. This study compares the control algorithms of an artificial pancreas by integrating both the Hovorka equation and its improved version to enhance personalized insulin delivery for adolescents with T1DM. A mathematical model was developed in MATLAB to simulate glucose-insulin dynamics and assess performance under real-world conditions, including meal intake and insulin sensitivity. The Ordinary Differential Equation Solver 45 (ODE45) was used to model the dynamic interaction between glucose levels and insulin infusion over 24 hours. The standard Hovorka equation established baseline insulin delivery, while the improved version introduced a trial-and-error approach for bolus insulin adjustments and refined carbohydrate intake calculations. Performance was evaluated based on the ability to maintain blood glucose within the normoglycemic range (4.0–7.0 mmol/L). Simulation results showed that both models effectively regulated glucose levels, but the improved equation provided better glycemic stability by reducing postprandial spikes more efficiently. Blood glucose, which initially rose to 10–12 mmol/L after meals, returned to normal faster with the improved model. This enhanced adaptability suggests its potential for real-time continuous subcutaneous insulin infusion (CSII) systems. The findings emphasize the importance of refining control algorithms for personalized insulin delivery. Future research should incorporate additional physiological factors such as stress and exercise to improve real-world applicability and optimize automated diabetes management.