Stochastic differential equations in NONMEM: implementation, application, and comparison with ordinary differential...

Area of Science:

• Pharmacometrics
• Pharmacokinetics/Pharmacodynamics (PK/PD)
• Mathematical Modeling

Background:

• Population pharmacokinetic/pharmacodynamic (PK/PD) models are crucial for understanding drug behavior.
• Intra-individual variability presents challenges in accurately modeling PK/PD relationships.
• Nonlinear mixed-effects models are commonly used but can be enhanced with advanced techniques.

Purpose of the Study:

• To explore the application of stochastic differential equations (SDEs) in population PK/PD modeling.
• To implement and evaluate the Extended Kalman Filter (EKF) within NONMEM for SDE-based models.
• To utilize SDEs as a diagnostic tool for assessing model appropriateness.

Main Methods:

• Decomposition of intra-individual variability into measurement and system noise within SDEs.
• Implementation of the Extended Kalman Filter (EKF) for parameter estimation in SDE models using NONMEM.
• Systematic model development using clinical PK data of degarelix.

Main Results:

• Successful implementation of an EKF-based algorithm in NONMEM for SDE models.
• Demonstration of SDEs' utility in pinpointing structural model deficiencies.
• Effective use of dynamic noise estimates to track parameter variations and build an absorption model for degarelix.

Conclusions:

• The EKF-based algorithm is effective for parameter estimation in population PK/PD models using SDEs.
• SDEs provide valuable insights into structural model deficiencies.
• Tracking unexplained parameter variations offers significant information for model refinement.