The disposition index: from individual to population approach

Area of Science:

• Metabolic research
• Biostatistics
• Endocrinology

Background:

• Effective glucose control evaluation necessitates assessing both insulin sensitivity and secretion.
• The disposition index (DI) is a key metric, traditionally calculated using hyperbolic or power function laws.
• Existing curve-fitting methods for DI estimation present limitations, including biased results and assumptions about error-free indices.

Purpose of the Study:

• To critically review traditional DI estimation methods and highlight their inherent issues.
• To introduce a novel nonlinear total least square (NLTLS) approach for improved DI estimation.
• To propose and validate a nonlinear mixed-effects (NLME) model to account for population variability in DI.

Main Methods:

• Review of traditional curve-fitting approaches for disposition index (DI) calculation.
• Development and application of a nonlinear total least square (NLTLS) method.
• Implementation and simulation-based validation of a nonlinear mixed-effects (NLME) model for DI analysis.

Main Results:

• Traditional methods and NLTLS provide biased DI estimates due to unaddressed variability.
• The proposed NLME approach accurately models population hyperparameters and individual DI variations.
• NLME demonstrates superior reliability and robustness compared to curve-fitting methods on simulated and real IVGTT data.

Conclusions:

• The nonlinear mixed-effects approach provides a more accurate and robust method for disposition index estimation.
• Analysis of IVGTT data suggests a power function law with α < 1 may be a better fit than the traditional hyperbolic law.
• This study underscores the importance of accounting for population variability in DI assessment for improved glucose control understanding.