Population insulin sensitivity from sparsely sampled oral glucose tolerance tests.

Area of Science:

• Metabolic research
• Statistical modeling
• Endocrinology

Background:

• Estimating population insulin sensitivity (S_I) is crucial for metabolic health assessment.
• Traditional methods often require numerous data points, limiting their application.
• Sparsely sampled oral glucose tolerance tests (OGTT) present a challenge for accurate S_I estimation.

Purpose of the Study:

• To develop and validate a novel method for estimating population-level insulin sensitivity (S_I) from limited OGTT samples.
• To assess the efficacy of a nonlinear multilevel (NLML) statistical model combined with the Dalla Man OGTT model.
• To compare the NLML approach with traditional methods using both simulated and real-world data.

Main Methods:

• Combined the Dalla Man OGTT mathematical model with a nonlinear multilevel (NLML) statistical model.
• Utilized sparsely sampled datasets (3-4 samples per subject within 120 min).
• Validated the model using simulated subjects and a cohort of prediabetic and type 2 diabetic (T2D) patients.

Main Results:

• The NLML model accurately estimated population S_I from simulated data with 3 or 4 time points.
• Significantly lower insulin sensitivity was detected in simulated subjects with below-average S_I (P<0.001).
• The NLML model successfully differentiated insulin sensitivity between prediabetic and T2D subjects (P<0.001).

Conclusions:

• Population S_I estimation from OGTT data using the NLML model is effective, even with sparse sampling.
• This approach allows for the assessment of population insulin sensitivity differences not detectable with individual S_I calculations.
• The method provides a valuable tool for metabolic research when extensive sampling is not feasible.