Abstract
OBJECTIVE: The goal of this study was to validate a previously derived and identified physiological subcutaneous (SC) insulin absorption model for computer simulation in a clinical diabetes decision support role using published pharmacokinetic summary measures. METHODS: Validation was performed using maximal plasma insulin concentration (C(max)) and time to maximal concentration (t(max) pharmacokinetic summary measures. Values were either reported or estimated from 37 pharmacokinetic studies over six modeled insulin types. A validation comparison was made to equivalent pharmacokinetic summary measures calculated from model generated curves fitted to respective plasma insulin concentration data. The validation result was a measure of goodness of fit. Validation for each reported study was classified into one of four cases. RESULTS: Of 37 model fits, 22 were validated on both the C(max) and the t(max) summary measures. Another 6 model fits were partially validated on one measure only due to lack of reporting on the second measure with errors to reported or estimated ranges of <12%. Another 7 studies could not be validated on either measure because of inadequate reported clinical data. Finally, 2 separate model fits to data from the same study failed the validation with 90 and 71% error on t(max) only, which was likely caused by protocol-based error. No model fit failed the validation on both measures. CONCLUSIONS: A previously derived and identified model was clinically validated for six insulin types using C(max) and t(max) summary measures from published pharmacokinetic studies. Hence, this article presents a clinically valid model that accounts for multiple nonlinear effects and six different types of SC insulin in a computationally modest form suitable for use in clinical decision support.