pyDarwin

pyDarwin performs nonlinear mixed-effect model selection by integrating machine learning algorithms with NONMEM to conduct global searches over user-defined model spaces.

Key Features:

• Integration with Machine Learning: Incorporates machine learning algorithms to navigate complex nonlinear mixed-effect model search spaces.
• Utilization of NONMEM: Uses NONMEM (Nonlinear Mixed Effects Modeling) for model evaluation and parameter estimation.
• Global Search Capability: Performs a global search across the entire user-defined model space rather than relying on stepwise selection.
• Efficiency and Objectivity: Automates model evaluation using predefined criteria to reduce manual selection bias and improve selection reproducibility.

Scientific Applications:

• Pharmacometrics: Supports nonlinear mixed-effect modeling for pharmacokinetic and pharmacodynamic analyses in drug development.
• Clinical data analysis: Applicable to clinical datasets with complex inter- and intra-individual variability and large sample sizes.
• Quetiapine case study: Has been applied to quetiapine clinical data as an example of the model search and selection workflow.

Methodology:

Define the search space for potential models, employ machine learning techniques alongside NONMEM to explore that space, and evaluate model configurations based on predefined criteria to select the optimal model.