Heart rate monitoring demonstrated superior sensitivity for detecting physical activity preceding hypoglycemia, capturing 80.2% of cases compared to 53.5% for patient declarations and 37.6% for steps.
Observational (n=497)
Does heart rate monitoring improve the detection of physical activity preceding hypoglycemia compared to patient declarations or step counts in adults with Type 1 Diabetes?
Heart rate monitoring is more sensitive than patient declarations or step counts for detecting physical activity that precedes hypoglycemia in type 1 diabetes, potentially enabling proactive automated insulin delivery adjustments.
Introduction and Objective: Physical activity (PA) is a well-known precipitant of hypoglycemia in type 1 diabetes (T1D), yet the temporal relationship between PA detection signals and hypoglycemic events remains incompletely characterized. We aimed to identify which PA indicators—patient declarations, heart rate elevation, or step counts—most frequently precede hypoglycemia, and quantify the timing between PA onset and hypoglycemia occurrence. Methods: We analyzed data from the T1-DEXI database, comprising 497 virtual adult patients with T1D (mean age 37 ± 14 years; mean HbA1c 6.6 ± 0.8%). PA was identified in the 120 minutes preceding hypoglycemia using three detection methods: patient declarations (intensity levels 1-3), heart rate exceeding a personalized threshold (resting heart rate + 0.4 × heart rate reserve) for 10 minutes, and step count 10 steps for 10 minutes. The maximum heart rate was calculated using the Tanaka formula. Results: Heart rate elevation was the most sensitive indicator, capturing 80.2% of PA cases. Patient declarations captured 53.5% of cases, while step counts captured 37.6%. With heart rate as the only active indicator, 37.2% of PA cases were detected, compared to 9.8% for patient declarations alone and 3.9% for step counts alone. PA detected using all indicators combined preceded 40.3% of hypoglycemic events, with a mean interval of 65.5 ± 0.4 minutes between PA onset and hypoglycemia. Conclusion: Heart rate monitoring demonstrates superior sensitivity for detecting PA preceding hypoglycemia compared to patient declarations or step counts alone. The observed delay between PA onset and hypoglycemia (65.5 minutes) aligns with the peak action window of rapid-acting insulin analogues (30-90 minutes), suggesting that timely insulin dose reduction upon PA detection could effectively prevent subsequent hypoglycemia. Integrating continuous heart rate data into automated insulin delivery systems may therefore enable proactive algorithmic intervention. Disclosure P. Gauthier: Employee; Current; Diabeloop SA. A. Adenis: Employee; Current; Diabeloop SA. S. Lachal: Employee; Current; Diabeloop SA. P. Gimenez: Employee; Current; Diabeloop SA. C. Desir: Employee; Current; Diabeloop SA. T. Le Roux-Mallouf: None. E. Huneker: Employee; Current; Diabeloop SA. P.Y. Benhamou: Employee; Current; Diabeloop SA. Advisory Panel; Ended; Eli Lilly and Company, Novo Nordisk.
GAUTHIER et al. (2026) conducted an observational in Type 1 Diabetes (n=497). Heart rate monitoring vs. Patient declarations and step counts was evaluated on Sensitivity for detecting physical activity preceding hypoglycemia. Heart rate monitoring demonstrated superior sensitivity for detecting physical activity preceding hypoglycemia, capturing 80.2% of cases compared to 53.5% for patient declarations and 37.6% for steps.