A prediction method for plasma concentration by using a nonlinear grey Bernoulli combined model based on a self-memory algorithm.

The goal of this work is to present and explore the application of a novel nonlinear grey Bernoulli combined model based on a self-memory algorithm, abbreviated as SA-NGBM, for modeling single-peaked sequences of time samples of acetylsalicylate plasma concentration following oral dosing. The self-memorization SA-NGBM routine reduces the dependence on a solitary initial value, as the initial state of the model utilizes multiple time samples. To test its forecasting performance, the SA-NGBM was used to extrapolate the plasma concentration predicted data, in comparison with the later time samples. The results were contrasted with those of the traditional optimized NGBM (ONGBM), exponential smoothing (ES) and simple moving average (SMA) using four popular accuracy and significance tests. That comparison showed that the SA-NGBM was much more accurate and efficient for matching the individual, nonlinear-system stochastic fluctuations than the existing ONGBM, ES and SMA models. The findings have potential applications for signal matching to similar small sample size, single-peaked, plasma concentration series.