Leeno: Type 1 diabetes management training environment using smart algorithms

Abstract

A growing number of Type-1 Diabetes (T1D) patients globally use insulin pump technologies to monitor and manage their glucose levels. Although recent advances in closed-loop systems promise automated pump control in the near future, most patients worldwide still use open-loop continuous subcutaneous insulin infusion (CSII) devices which require close monitoring and continuous regulation. Apart from specialized diabetes units, hospital physicians and nurses generally lack necessary training to support the growing number of patients on insulin pumps. Most hospital staff and providers worldwide have never seen or operated an insulin pump device. T1D patients at nurseries, schools, in hospital emergency rooms, surgery theatres, and in-patient units all require close monitoring and active management. The lack of knowledge and necessary training to support T1D patients on pumps puts them at life-threatening risks. In this work, we develop a training simulation software for hospitals to educate and train their physicians and nurses on how to effectively operate a T1D pump and reduce hypoglycemia events. The software includes clinically validated T1D virtual patients that users can monitor and adjust their pump settings to improve glycemic outcomes. We develop a Fuzzy-Logic learning algorithm that helps guide users learn how to improve pump parameters for these patients. We recruited and trained 13 nurses on the software and report their improvement in pump administration, basal rates adjustments, and ICR modulation.