UNSTRUCTURED
Type 1 diabetes (T1D) results from an autoimmune response, which causes the destruction of insulin-producing pancreatic β cells. This condition necessitates lifelong insulin therapy to manage glucose levels. Recent advancements in technology, particularly in the realm of machine learning (ML), have revolutionized T1D care. This article presents a comprehensive review of ML applications in T1D management, focusing on glucose level prediction, insulin dose control, and the development of closed-loop artificial pancreas systems. ML, particularly deep neural networks (DNNs), is introduced as a potent tool with diverse applications, including medical diagnostics and electromechanical system control. The closed-loop systems, where insulin delivery is connected with the real-time CGM systems data and controlled by specialized algorithms, are very useful tools, which bring new possibilities for better T1D treatment like improving glycemic control, increasing time in range, and reducing hypoglycemic episodes. Although there are still many challenges in T1D treatment, the development of ML and more and more advanced insulin pumps is the future of T1D management.