Go beyond the limits of Genetic Algorithm in daily covariate selection practice

Abstract

Covariate identification is an important step in the development of a population pharmacokinetic/pharmacodynamic (PK/PD) model. Among the different available approaches, the stepwise covariate model (SCM) is the most used. However, SCM is based on a local search and on the assumption of independent covariate effects. Moreover, it is manly designed to test few covariate-parameter relationships, often chosen ad hoc or based on prior knowledge. The application of genetic algorithms (GA) for covariate selection has been proposed as a possible solution to overcome these limitations. However, their actual use during model building is limited by the extremely high computational costs and convergence issues, both related to the number of models being tested. In this paper, we proposed a new genetic algorithm for covariate selection to address these challenges. The GA was first developed on a simulated case study; then, its performances were evaluated on a real-world problem related to Remifentanil. The implemented GA showed good results both in terms of correctness of the selected model and fitness optimization, and it was able to outperform the SCM in both scenarios. Furthermore, the heuristics introduced to overcome the limitations affecting currently available GA approaches allowed to reduce convergence times, obtain replicable results, and limit the number of selected covariates.

Reinforcement Learning and PK‐PD Models Integration to Personalize the Adaptive Dosing Protocol of Erdafitinib in Patients with Metastatic Urothelial Carcinoma

“…AI/ML are now present throughout the entire continuum of pharmacology research, from the early drug discovery phases [1][2][3] to the clinical practise. 4,5 Recently, hybrid approaches integrating AI/ML with mathematical modeling and simulations have been proposed and gained great momentum in the pharmacometric landscape to solve different tasks [6][7][8][9][10] such as pharmacokinetic (PK) and pharmacodynamic (PD) predictions, 6,7 covariate selection 6,7,10,11 as well as treatment optimization, especially in a precision dosing context. 12 Precision dosing refers to tailoring the dose on each individual patient at each given time, in order to ensure the greatest benefits and least risks.…”