AI-generated from publicly available materials.A new browser-based tool developed at Anadolu Medical Center in collaboration with Johns Hopkins Medicine enhances insulin infusion safety by transforming traditional paper protocols into an executable decision aid. In a validation study involving 120 cases, the tool successfully matched an independent clinical reference standard, while manual calculations by nursing staff aligned in only 35 instances.
The initiative was spearheaded by Gurkan Camok, who identified six common failure points in manual insulin administration workflows. These included errors in glucose band selection and miscalculations that could lead to incorrect insulin dosing, posing risks of hypoglycemia or prolonged hyperglycemia. The new application streamlines this process by taking four key inputs—current and previous glucose levels, elapsed time, and current infusion rate—to calculate the appropriate insulin adjustment, thereby reducing the cognitive load on healthcare professionals.
During the study, each nurse's manual calculations were compared against the tool's recommendations, which were generated independently the following day to ensure no influence on patient care. The digital tool demonstrated a perfect match with the reference standard, while manual calculations showed significant discrepancies. Importantly, no patient harm was linked to these differences, emphasizing that numerical deviations do not necessarily equate to adverse events.
Since its rollout in January 2026, the tool has been well-received by approximately 400 nurses and physicians, who appreciate its clarity and efficiency. However, Camok warns of potential overreliance on the tool's recommendations and stresses the importance of human oversight at every step. As the software remains a non-commercial internal tool, future plans include formal classification as a medical device and multicenter validation, indicating its potential for broader application in managing other high-risk clinical calculations.