Opportunities for AI-based Model-informed Drug Development: A Comparative Analysis of NONMEM and AI-based Models for...

Area of Science:

• Pharmacometrics
• Computational Biology
• Drug Development

Background:

• Model-informed drug development (MIDD) utilizes mathematical models for optimizing drug dosing.
• Nonlinear mixed effects modeling (NONMEM) is a traditional standard for population pharmacokinetic (PPK) analysis.
• Artificial intelligence (AI) offers potential advancements in predictive performance and efficiency for PPK modeling.

Purpose of the Study:

• To evaluate the effectiveness of AI-based MIDD methods for PPK analysis.
• To compare the performance of AI/ML models against traditional nonlinear mixed-effects (NLME) methods like NONMEM.
• To assess the applicability of AI in enhancing pharmacometrics workflows.

Main Methods:

• Tested five machine learning (ML) models, three deep learning (DL) models, and a neural ordinary differential equations (ODE) model.
• Utilized simulated datasets based on a two-compartment model and a real clinical dataset from 1,770 patients.
• Assessed predictive performance using metrics like root mean squared error (RMSE), mean absolute error (MAE), and R-squared (R²).

Main Results:

• AI/ML models frequently outperformed NONMEM in PPK analysis.
• Performance varied based on AI model type and data characteristics.
• Neural ODE models demonstrated strong performance and explainability, particularly with large datasets.

Conclusions:

• AI/ML methodologies show potential to complement or enhance traditional PPK modeling in MIDD.
• AI approaches offer improved predictive performance and computational efficiency.
• These findings support the integration of AI/ML into future pharmacometrics workflows.