Safe and effective digital glycaemic control: Results from a 1,000-patient clinical trial

Objective: Stress-induced hyperglycaemia and glycaemic variability are associated with increased mortality and morbidity in intensive care unit (ICU) patients. Glycaemic control (GC) protocols aimed at maintaining blood glucose (BG) within safe ranges have been shown to improve outcomes, but they are often difficult to implement safely, limiting their adoption in clinical practice.
Methods: This study presents clinical trial results of the STAR-3D protocol, a unique, AI-powered, digital twin-based system that provides real time recommendations for insulin and nutrition therapy. Between April 2019 and December 2024, 1085 patients were enrolled, representing 1430 GC
episodes and 86,010 hours of control after data filtering. Results: STAR-3D delivered effective and safe control, with 71% of BG measurements falling within the normoglycaemic target range (4.4-8.0 mmol/L) and only 3 severe hypoglycaemia events (BG < 2.2 mmol/L). High and personalised
nutrition delivery was also achieved, with a median [IQR] of 89.0 [17.2 – 100.0] %goal feed delivered. Conclusions: These outcomes were consistent at both cohort and per-patient levels, demonstrating STAR-3D’s ability to provide safe and effective GC through personalised management of glycaemia and nutrition. Its adaptability to individual patients’ needs supports broader adoption in ICUs and offers a robust platform for evaluating the clinical impact of GC.