A Scalable Application of Artificial Intelligence-Driven Insulin Titration Program to Transform Type 2 Diabetes...

Area of Science:

• Endocrinology
• Artificial Intelligence in Medicine
• Diabetes Management

Background:

• Long-term type 2 diabetes patients often remain hyperglycemic despite current pharmacotherapy.
• Insulin therapy's efficacy is limited by the demanding nature of frequent physiological titrations.
• Existing treatments struggle to achieve optimal glycemic control in many long-standing diabetes cases.

Purpose of the Study:

• To report real-world outcomes of a scalable transformation in diabetes care using AI-driven autonomous insulin titration.
• To evaluate the efficacy and safety of d-Nav technology in a community endocrinology center.
• To assess the integration of cardiorenal-risk reduction pharmacotherapy alongside AI-guided insulin management.

Main Methods:

• Implementation of d-Nav technology, an AI-based autonomous insulin titration system, in a community endocrinology center.
• Retrospective analysis of treatment efficacy and safety data for the first 600 patients using the system.
• Monitoring of glycated hemoglobin (HbA1c), insulin dosage, and incidence of hypoglycemia.

Main Results:

• High patient retention (82%) with d-Nav use over 8.2 months.
• Significant reduction in HbA1c from 8.6% to 7.3%, with a decrease in patients having HbA1c >9% from 29.3% to 5.7%.
• Low rates of hypoglycemia (0.4/month) and severe hypoglycemia (1.7/100 patient-years), with some patients experiencing reduced insulin doses.

Conclusions:

• AI-driven autonomous insulin titration (d-Nav) offers a scalable solution for improving diabetes management.
• The technology facilitates appropriate use of both insulin and non-insulin pharmacotherapies, including cardiorenal risk agents.
• This approach enhances glycemic control and safety, addressing limitations of traditional insulin titration methods.