Table 2 Adjusted \(R^2\) summary for clinical outcomes across six predictive models of varying complexity.

From: Glucodensity functional profiles outperform traditional continuous glucose monitoring metrics

 

Model (1)

Model (2)

Model (3)

Model (4)

Model (5)

Model (6)

HbA1c–5 years \((n=209)\)

0.640

0.645

0.650

0.732

0.743

0.796

HbA1c–8 years \((n=250)\)

0.584

0.579

0.589

0.628

0.656

0.680

FPG–5 years \((n=408)\)

0.434

0.438

0.474

0.518

0.587

0.617

FPG–8 years \((n=223)\)

0.339

0.426

0.385

0.425

0.445

0.517

  1. Model (1) includes baseline age, FPG, and HbA1c. Model (2) adds CGM metrics such as Area Under the Curve (AUC), Mean Amplitude of Glycemic Excursions (MAGE), Continuous Overall Net Glycemic Action (CONGA), and hyperglycemia time-in-range. Model (3) adds CGM-based dynamic markers—entropy rate (ER) and Poincaré ellipse area (S)—to age, FPG, and HbA1c. Model (4) adds the quantile function for the glucose profile to age, FPG, and HbA1c. Model (5) adds quantile functions for glucose speed to Model (4), and Model (6) adds quantile functions for glucose acceleration to Model (5).
Back to article page