368-P: Perceived AID Algorithm Performance and Hypoglycemia Prevalence in PWD

• Published in: Diabetes (New York, N.Y.) Vol. 74; no. Supplement_1; p. 1
• Main Authors: BELTRAN, ALAN, LEE, JACKSON, BELL, TREVOR, BRISTOW, TRACY L., WOOD, RICHARD
• Format: Journal Article
• Language: English
• Published: New York American Diabetes Association 20.06.2025
• Subjects: Algorithms; Diabetes mellitus; Hypoglycemia; Insulin
• ISSN: 0012-1797; 1939-327X
• DOI: 10.2337/db25-368-P

Introduction and Objective: Hypoglycemia, defined as blood glucose (BG) below 70 mg/dL, poses significant risks for people with diabetes (PWD) using insulin. Despite growing adoption of Automated Insulin Delivery (AID) systems, hypoglycemia among their users remains understudied. This study assessed hypoglycemia experiences alongside perceived AID algorithm performance. Methods: In November 2024, US PWD (n=1,384 Type 1 and n=107 Type 2) using AID systems completed an online survey. Respondents rated their AID algorithms as “too conservative,” “just right,” or “too aggressive” in treating high BG. They also reported frequencies of mild (55-70 mg/dL) and moderate (40-54 mg/dL) hypoglycemia and their BG thresholds for treating lows and highs. In analysis, highly impacted individuals were defined as experiencing near-daily milds and occasional (weekly) moderates. Hypoglycemia impact and BG thresholds were compared across algorithm rating groups. Results: Differences in prevalence of “highly impacted” PWD occurred across all algorithm groups (p<0.05). PWD rating algorithms as “just right” had the lowest impact (12%), followed by “too conservative” (17%) and “too aggressive” (35%). The “too conservative” group treated low BG at a lower mean threshold (70.4 mg/dL) than “just right” (72.2, p<0.01) and “too aggressive” (72.2, p<0.05). They also treated high BG at a lower mean threshold (174.1 mg/dL) than “just right” (183.9, p<0.01) and “too aggressive” (185.2, p<0.01). Conclusion: PWD rating their AID algorithms as “just right” experience the lowest hypoglycemia impact. Those perceiving algorithms as “too aggressive” report the highest, suggesting they need more cautious settings/algorithms. The “too conservative” group treats lows and highs at lower mean thresholds than others, suggesting they’re less proactive with lows and/or “rebound” after preemptively correcting highs, leading to hypoglycemia. These findings emphasize the nuances of managing hypoglycemia on an AID system and the importance of aligning patients to optimal settings/algorithms.