An Efficient Methodology for PK-PD Data Analysis Based on Novel Bound Selection and Modified Spiral Dynamic Optimization Methods

Compartmental Pharmacokinetic and Pharmaco dynamic (PK-PD) data analysis plays a pivotal role in the developability assessment of drug candidates and eventual bedside applications of drugs. It employs laborious procedures for the selection of initial parameter estimates and parameters bounds and further, is critically dependent on PK-PD domain expertise. In order to overcome these bottlenecks, we report in this paper, an efficient methodology for performing compartmental PK-PD data analyses and it involves the following three key stages: A) the selection of parameter bounds based on the changes in the characteristics of the PK-PD data graphs, B) global optimization of the parameter bounds to obtain the initial parameters using modified spiral dynamics optimization algorithm(MSDO) and C) local optimization of the initial estimates using Generalized Reduced Gradient (GRG) non-linear and Nelder-Mead Simplex methods. The efficiency of the methodology is established using sixteen Pharmacokinetic and Pharmacodynamic datasets. Based on the results, we conclude that the new methodology is i) superior over other reported approaches in providing parameter estimates with minimal WRSS etc. ii) faster to perform successful analysis iii) easier to use, even for beginners in PK-PD data analysis, as it does not require experience in PK-PD data analyses, iv) widely applicable to a variety of PK-PD modelling scenarios involving non-linear and non-monotonic functions, and vi) a promising methodology to reduce cost and time of drug discovery and development.