Early detection of prediabetes using blood glucose remains difficult as individuals often present with slight elevations or even normoglycaemia. However, their insulin profiles usually show marked elevations, suggesting that insulin is a more sensitive biomarker for prediabetes. We developed a mechanistic whole-body model that can predict an individual’s insulin time-course and insulin sensitivity from a given glucose time-course. The model describes the absorption, distribution, metabolism and excretion of glucose and the relevant hormones, while various homeostatic processes such as insulin secretion were dynamically represented by sub-models. Matched glucose and insulin data were obtained from 7 oral glucose tolerance test studies. Model predictions of insulin time-course using glucose data as inputs were compared with observed data for validation (Fig 1). Both population and individual simulation results indicated robust prediction of insulin profiles and demonstrate that it is possible to utilize glucose data (easy to obtain) to predict insulin data (difficult to obtain) for prediabetes screening.
K.Hor cheng: None. J.C.Y.Chan: None.
