Identification of severe hypoglycemia in adults with type 1 diabetes using CGM-based machine learning: evidence from the FGM-Japan study

Key Points

• Severe hypoglycemia is identified with high accuracy using machine learning techniques and continuous glucose monitoring data.
• The model achieved a sensitivity of 87% and a specificity of 92% while analyzing data from adults with type 1 diabetes.
• Continuous glucose monitoring data was utilized to develop the predictive model for identifying hypoglycemic events.
• This indicates potential for improved management of severe hypoglycemia in type 1 diabetes with real-time monitoring.