LatinPSO: An algorithm for simultaneously inferring structure and parameters of ordinary differential equations...

Area of Science:

• Computational Biology
• Systems Biology
• Mathematical Modeling

Background:

• Inferring Ordinary Differential Equations (ODEs) structure and parameters is crucial for understanding complex dynamic systems.
• Existing methods face challenges due to the complexity of the search space in systems identification problems.
• Simultaneous inference offers a more practical approach but remains computationally intensive.

Purpose of the Study:

• To develop a novel algorithm for simultaneously inferring ODE model structure and parameters.
• To address the challenges posed by complex search spaces in systems identification.
• To provide an effective tool for analyzing time-course data from dynamic systems.

Main Methods:

• Proposed a new algorithm named LatinPSO.
• Integrated Particle Swarm Optimization (PSO) with Latin Hypercube Sampling (LHS).
• Utilized time-course data for model inference and validation.

Main Results:

• LatinPSO demonstrated effectiveness in inferring both structure and parameters of ODE models.
• Evaluated performance using a real Human Immunodeficiency Virus (HIV) model and synthetic datasets.
• Achieved satisfactory candidate ODE models with appropriate structures and parameters.

Conclusions:

• LatinPSO offers a robust solution for the challenging problem of ODE model identification.
• The algorithm successfully balances inferring model structure and estimating parameters.
• This approach holds promise for advancing systems biology and dynamic systems analysis.