Design of Adaptive TADF Controller for Glucose-Insulin (GI) Interaction in Type-I Diabetes Patient

Introduction: This study proposes an Elephant Herding Optimization-Tilt Acceleration- Derivative-Filter (EHO-TADF) controller for robust regulation of Blood Glucose (BG) levels in patients with Type-I Diabetes Mellitus (TIDM) using an Artificial Pancreas (AP). TIDM is a chronic condition marked by the body’s inability to produce insulin, leading to dangerous fluctuations in BG levels. Maintaining normo-glycemia (70–120 mg/dl) is challenging due to nonlinear physiological responses and external disturbances like meals and exercise. Existing control strategies such as PID, Fuzzy, LQG, and H∞ controllers often fall short in providing the adaptive tuning, global robustness, and rapid convergence required for accurate BG regulation.Therefore, advanced control strategies integrated with optimization algorithms are essential to improve glycemic control in TIDM patients. This study proposes an EHO-TADF controller for robust regulation of BG levels in TIDM patients with an AP. The primary objective is to develop a controller that ensures normo-glycemia while enhancing stability, accuracy, and robustness under physiological uncertainties and external disturbances. The EHO algorithm tunes the control gains of the TADF controller to minimize the objective function, ensuring optimal insulin infusion in real time. The EHO-TADF controller introduces a filtered structure to improve the dynamic response and noise rejection capability, effectively handling the nonlinear nature of GI dynamics. Methods: Briefly describes the use of the EHO algorithm to tune the proposed TADF controller integrated into an AP framework. Results: Simulation results demonstrate that the proposed approach (EHO-TADF) achieves a faster settling time (230 min), lower peak overshoot (3.9 mg/dl), and reduced noise (1%) compared to other existing control strategies. Discussion: The discussion highlights that the EHO-TADF controller significantly enhances blood glucose regulation in T1DM patients by: • Adapting to various physiological disturbances like irregular meals and physical activity. • Providing faster and more stable responses compared to traditional controllers. • Offering greater robustness and noise immunity, ensuring reliable control even under uncertain conditions. • Maintaining glucose within the normo-glycemic range with minimal insulin usage. The controller’s superior performance in precision, stability, and adaptability confirms its potential for real-time implementation in artificial pancreas systems. Conclusion: The closed-loop patient model achieves a stable BG level of 81 mg/dl under varying meal and exercise disturbances, validating the superior performance and robustness of the EHO-- TADF approach.