A probabilistic computation framework to estimate the dawn phenomenon in type 2 diabetes using continuous glucose monitoring

Souptik Barua,Namino Glantz,Arianna Larez,Wendy Bevier,Ashutosh Sabharwal,David Kerr

DOI: https://doi.org/10.1038/s41598-024-52461-1

IF: 4.6

2024-02-06

Scientific Reports

Abstract:In type 2 diabetes (T2D), the dawn phenomenon is an overnight glucose rise recognized to contribute to overall glycemia and is a potential target for therapeutic intervention. Existing CGM-based approaches do not account for sensor error, which can mask the true extent of the dawn phenomenon. To address this challenge, we developed a probabilistic framework that incorporates sensor error to assign a probability to the occurrence of dawn phenomenon. In contrast, the current approaches label glucose fluctuations as dawn phenomena as a binary yes/no. We compared the proposed probabilistic model with a standard binary model on CGM data from 173 participants (71% female, 87% Hispanic/Latino, 54 ± 12 years, with either a diagnosis of T2D for six months or with an elevated risk of T2D) stratified by HbA 1c levels into normal but at risk for T2D, with pre-T2D, or with non-insulin-treated T2D. The probabilistic model revealed a higher dawn phenomenon frequency in T2D [49% (95% CI 37–63%)] compared to pre-T2D [36% (95% CI 31–48%), p = 0.01] and at-risk participants [34% (95% CI 27–39%), p < 0.0001]. While these trends were also found using the binary approach, the probabilistic model identified significantly greater dawn phenomenon frequency than the traditional binary model across all three HbA 1c sub-groups (p < 0.0001), indicating its potential to detect the dawn phenomenon earlier across diabetes risk categories.

multidisciplinary sciences