HCVMultiscaleFit: A Simulator For Parameter Estimation in Multiscale Models Of Hepatitis C Virus Dynamics

Area of Science:

• Mathematical Biology
• Computational Science
• Virology

Background:

• Parameter estimation (calibration) is crucial for mathematical models based on differential equations.
• Sophisticated models, such as age-structured and multiscale hepatitis C virus (HCV) dynamics models, present significant calibration challenges.
• Existing fully numerical methods for partial differential equations (PDEs) are often too slow for practical use due to precision issues.

Purpose of the Study:

• To develop a computationally efficient method for parameter estimation in multiscale differential equation models.
• To adapt existing methods to overcome speed limitations and enable practical application.
• To create a user-friendly tool for simulating viral dynamics during antiviral treatment.

Main Methods:

• Implemented constrained optimization to incorporate boundary value constraints for model parameters.
• Utilized derivative-free optimization methods to avoid computationally expensive numerical derivative approximations.
• Developed a simulator, HCVMultiscaleFit, with a Graphical User Interface (GUI) for efficient parameter estimation.

Main Results:

• Reduced the running time of parameter estimation methods by orders of magnitude, making them practical.
• Successfully demonstrated the adaptability of these efficient methods to other multiscale models and viruses.
• Provided a functional GUI simulator (HCVMultiscaleFit) for parameter estimation in HCV dynamics.

Conclusions:

• The combination of constrained optimization and derivative-free methods significantly enhances the efficiency of parameter estimation for complex differential equation models.
• These improved methods are applicable beyond HCV dynamics, offering a practical solution for various sophisticated mathematical models.
• HCVMultiscaleFit provides a valuable tool for researchers simulating viral dynamics and evaluating antiviral treatments.