Intelligent Diabetes Management
Diabetes is a global health challenge, affecting more than 590 million people worldwide. Many individuals with diabetes eventually require insulin therapy, but insulin dosing remains highly complex due to substantial variability in insulin sensitivity, glucose dynamics, lifestyle factors, and treatment adherence. Current dosing approaches, such as standard basal titration algorithms, are often rigid, slow to adapt, and insufficiently personalised, leading to suboptimal glucose control and an increased risk of both hyperglycaemia and hypoglycaemia.
This project, taking a whole systems approach, seeks to develop and validate an individualised and intelligent adaptive Model Predictive Control (MPC) strategy for insulin dose guidance and diabetes management in people with diabetes. The proposed control system will be architected and designed to update patient-specific parameters online, managing disturbances such as meals and physical activity, and ensuring safety against hypoglycaemia.
Key objectives of this project are: (a) to capture personalised glucose–insulin models; (b) to develop adaptive and robust MPC algorithm capable of addressing intra- and inter-day variability in insulin sensitivity and lifestyle factors; (c) to incorporate safety layers and fault-tolerant filtering to manage sensor delays, noise, and uncertainties; and (d) to evaluate proposed controllers through in silico simulations and, where feasible, clinical datasets or collaborations with healthcare partners.
This research has the potential to transform insulin therapy in diabetes by enabling intelligent, patient-centred dose guidance systems. The outcomes could inform the next generation of digital health tools, smart insulin delivery devices, and clinical decision-support platforms, ultimately improving quality of life and reducing complications for millions of people living with diabetes.