This study investigates the use of Reinforcement Learning (RL) to optimize insulin infusion in Automated Insulin Delivery (AID) systems. The Deep Deterministic Policy Gradient (DDPG) algorithm is evaluated as a potential control strategy and compared to Model Predictive Control (MPC) and PID controllers using a highly realistic in silico simulation model. Simulations for adult Type 1 Diabetes (T1D) patients demonstrate that the DDPG controller effectively reduces hypo- and hyperglycemic episodes and improves glucose stability.
A Deep Deterministic Policy Gradient control algorithm for Automatic Insulin Delivery / Lops, Giada; Racanelli, Vito Andrea; Manfredi, Gioacchino; De Cicco, Luca; Mascolo, Saverio. - STAMPA. - (2025). (Intervento presentato al convegno 1st IFAC Workshop on Engineering Diabetes Technologies (EDT 2025) tenutosi a Valencia nel May 8-9 2025) [10.1016/j.ifacol.2025.06.003].
A Deep Deterministic Policy Gradient control algorithm for Automatic Insulin Delivery
Vito Andrea Racanelli;Gioacchino Manfredi;Luca De Cicco;Saverio Mascolo
2025
Abstract
This study investigates the use of Reinforcement Learning (RL) to optimize insulin infusion in Automated Insulin Delivery (AID) systems. The Deep Deterministic Policy Gradient (DDPG) algorithm is evaluated as a potential control strategy and compared to Model Predictive Control (MPC) and PID controllers using a highly realistic in silico simulation model. Simulations for adult Type 1 Diabetes (T1D) patients demonstrate that the DDPG controller effectively reduces hypo- and hyperglycemic episodes and improves glucose stability.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
