Improved Detection of Decreased Glucose Handling Capacities via Novel Continuous Glucose Monitoring-Derived Indices: AC_Mean and AC_Var
Sugimoto, H.; Hironaka, K.-i.; Nakamura, T.; Yamada, T.; Miura, H.; Otowa-Suematsu, N.; Fujii, M.; Hirota, Y.; Sakaguchi, K.; Ogawa, W.; Kuroda, S.
Show abstract
BackgroundEfficiently detecting impaired glucose control abilities is a pivotal challenge in public health. This study assessed the utility of relatively easy-to-measure continuous glucose monitoring (CGM)-derived indices in estimating glucose handling capacities calculated from resource-intensive clamp tests. MethodsWe conducted a prospective, single-center, observational cohort study of 64 individuals without prior diabetes diagnosis. The study utilized CGM, oral glucose tolerance test, and hyperglycemic and hyperinsulinemic-euglycemic clamp tests. We validated CGM-derived indices characteristics using an independent dataset and mathematical model with simulated data. FindingsA novel CGM-derived index, AC_Var, was significantly correlated with insulin sensitivity (r = -0.31; 95% CI: -0.52 to -0.06), insulin clearance (r = -0.31; 95% CI: -0.54 to -0.06), and disposition index (DI) (r = -0.31; 95% CI: -0.52 to -0.07) calculated from clamp tests. AC_Var was also significantly correlated with insulin resistance (r = 0.48; 95% CI: 0.23 to 0.68) in an independent dataset. Multivariate analyses indicated AC_Vars contribution to predicting reduced blood glucose control abilities independent from conventional CGM-derived indices. The prediction models accuracy utilizing CGM-measured glucose standard deviation and AC_Var as input variables, with clamp-derived DI as the outcome, closely matched that of predicting clamp- from OGTT-derived DI. Mathematical simulations also underscored AC_Vars association with insulin clearance and DI. InterpretationCGM-derived indices, including AC_Var, can be useful for screening decreased blood glucose control ability. We developed a web application that calculates these indices (https://cgm-ac-mean-std.streamlit.app/). FundingThe Japan Society for the Promotion of Science KAKENHI, CREST, Japan Science and Technology Agency, and Uehara Memorial Foundation.
Matching journals
The top 7 journals account for 50% of the predicted probability mass.
Similar papers in this journal
Similar papers in this journal
- Precision Diagnostic Approach to Predict 5-year Risk for Microvascular Complications in Type 1 Diabetes 95%
- Daily variation in blood glucose levels during continuous enteral nutrition in patients on the Intensive Care Unit: a retrospective observational study 92%
- Discovery of therapeutic agents targeting PKLR for NAFLD using drug repositioning 91%
Similar papers in this journal
- Metabolic and proteomic signatures of type 2 diabetes subtypes in an Arab population 93%
- AI-Enhanced Integration of Genetic and Medical Imaging Data for Risk Assessment of Type 2 Diabetes 93%
- Genetic Architecture and Risk Prediction of Gestational Diabetes Mellitus in over 116,144 Chinese Pregnancies 93%
Similar papers in this journal
- Plasma proteomic signatures of adiposity are associated with cardiovascular risk factors and type 2 diabetes risk in a multi-ethnic Asian population 91%
- The Anna Karenina model of beta cell maturation in development and their dedifferentiation in type 1 and type 2 diabetes 91%
- Risk-stratified lifestyle intervention to prevent type 2 diabetes 90%
Similar papers in this journal
- Intra- and inter-individual metabolic profiling highlights carnitine and lysophosphatidylcholine pathways as key molecular defects in type-2 diabetes 93%
- Identification of metabolomics biomarkers for type 2 diabetes: triangulating evidence from longitudinal and Mendelian randomization analyses 92%
- Can machine learning improve risk prediction of incident hypertension? An internal method comparison and external validation of the Framingham risk model using HUNT Study data 91%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.