A Wavelet AI Algorithm to Automatically Identify Postprandial Glucose Responses From Continuous Glucose Monitoring Profiles in People Without Diabetes
Authors' conclusion
The wavelet AI algorithm accurately identified PPGRs from CGM data in people without diabetes, offering a novel automated approach to monitor early signs of postprandial dysglycemia in this population.
Study context
- Population
- Adults without diabetes (normoglycemia and prediabetes); public continuous glucose monitoring (CGM) dataset of 25 normoglycemic adults (age 40 ± 14 years, BMI 26 ± 6 kg/m², HbA1c 5.4 ± 0.4%) and three validation cohortspublic dataset
- Participants
- 25 adults
- Comparison
- AI-predicted vs ground-truth mealtimes for PPGR identification from CGM
- Outcome
- Mealtime prediction error and tAUC and iAUC of the PPGR
Methodological detailswavelet transform-based AI algorithm, public dataset plus three independent validation cohorts
- Wavelet transform-based AI algorithm
- Public dataset plus three independent validation cohorts
Performance
As reported in the abstract
PPGR start time prediction error vs ground-truth mealtimes
- Median error of PPGR start time, minutes [IQR]
- 10 [IQR: 4, 19]
- tAUC and iAUC from ground-truth vs AI-predicted mealtimes similar
- all P > .1
Further findings
Association of AI-derived PPGR iAUC with laboratory markers (adjusted analysis)
- HbA1c: β (95% CI), P
- β = 0.57 [95% CI: 0.19, 0.95], P = .006
- Fasting glucose: β (95% CI), P
- β = 0.52 [95% CI: 0.12, 0.92], P = .013
- Adjusted analysis
Performance in validation cohorts
- Algorithm performance remained consistent across the three validation cohorts
Original publication
A Wavelet AI Algorithm to Automatically Identify Postprandial Glucose Responses From Continuous Glucose Monitoring Profiles in People Without Diabetes. J Diabetes Sci Technol. Published 29 March 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →