A glucose-insulin pharmacodynamic surface modeling validation and comparison of metabolic system models J. Geoffrey Chase a, *, Steen Andreassen b , Ulrike Pielmeier b , Christopher E. Hann a , Kirsten A. McAuley c , J.I. Mann c a Mechanical Eng., Centre for Bio-Engineering, University of Canterbury, Christchurch, New Zealand b Center for Model-based Medical Decision Support (MMDS), Aalborg University, Aalborg, Denmark c Edgar National Centre for Diabetes Research, University of Otago, School of Medicine, Dunedin, New Zealand 1. Introduction Type 1 and Type 2 diabetes are epidemic [1–3] with significant economic cost [4–7], driven by the inability of individuals and their clinicians to achieve consistent adequate control of blood glucose levels [8–10]. Hence, the rate of costly chronic complications is rising. In addition, critical care and surgical studies have shown that successfully providing tight glycaemic control can signifi- cantly reduce mortality [11–14] and cost [15,16]. However, achieving it has proven difficult, even though initial results show tighter control may be better [17,18]. The potential of models for managing glycaemic levels in any insulin resistant cohort is thus of growing import. However, relatively few models have been clinically validated. For most models, the primary form of validation has been simple fitting of the model to match clinical data [19]. Occasionally, more rigorous prediction validation, which tests the models ability to predict the outcome of a known intervention on retrospective clinical data (or in a clinical trial) is used. However, only a few clinically validated models can predict within clinically acceptable ranges [20–26]. This paper presents a new form of model validation that examines the steady state pharmaco-dynamic (PD) surfaces, including underlying pharmaco-kinetics (PK). It is thus a means of analysing the fundamental model dynamics used over a range of input and response variables. Comparison to clinical data points on this surface can provide a means of validating the model’s underlying dynamics. In this research, a 3D surface of plasma insulin (x), plasma glucose (y) and resulting rate of change in endogenous glucose balance (z) is compared to 77 sets of glycaemic clamp data from individuals with relatively high insulin sensitivity [20]. The Biomedical Signal Processing and Control 4 (2009) 355–363 ARTICLE INFO Article history: Received 3 November 2008 Received in revised form 11 April 2009 Accepted 20 April 2009 Available online 17 May 2009 Keywords: Pharmacodynamics Pharmacokinetics Metabolism Diabetes Insulin sensitivity Modeling Euglycaemic clamp Hyperinsulinaemic clamp Model validation ABSTRACT Metabolic system modeling for model-based glycaemic control is becoming increasingly important. Few metabolic system models are clinically validated for both fit to the data and prediction ability. This research introduces a new additional form of pharmaco-dynamic (PD) surface comparison for model analysis and validation. These 3D surfaces are developed for 3 clinically validated models and 1 model with an added saturation dynamic. The models include the well-known Minimal Model. They are fit to two different data sets of clinical PD data from hyperinsulinaemic clamp studies at euglycaemia and/or hyperglycaemia. The models are fit to the first data set to determine an optimal set of population parameters. The second data set is used to test trend prediction of the surface modeling as it represents a lower insulin sensitivity cohort and should thus require only scaling in these (or related) parameters to match this data set. This particular approach clearly highlights differences in modeling methods, and the model dynamics utilized that may not appear as clearly in other fitting or prediction validation methods. Across all models saturation of insulin action is seen to be an important determinant of prediction and fit quality. In particular, the well-reported under-modeling of insulin sensitivity in the Minimal Model can be seen in this context to be a result of a lack of saturation dynamics, which in turn affects its ability to detect differences between cohorts. The overall approach of examining PD surfaces is seen to be an effective means of analyzing and thus validating a metabolic model’s inherent dynamics and basic trend prediction on a population level, but is not a replacement for data driven, patient-specific fit and prediction validation for clinical use. The overall method presented could be readily generalized to similar PD systems and therapeutics. ß 2009 Elsevier Ltd. All rights reserved. * Corresponding author. Tel.: +64 3 364 7001. E-mail addresses: geoff.chase@Canterbury.ac.nz (J.G. Chase), sa@hst.aau.dk (S. Andreassen), Kirsten.Mcauley@stonebow.otago.ac.nz (K.A. McAuley). Contents lists available at ScienceDirect Biomedical Signal Processing and Control journal homepage: www.elsevier.com/locate/bspc 1746-8094/$ – see front matter ß 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.bspc.2009.04.002