mlegp

mlegp implements Gaussian process emulation and sensitivity analysis for R-based computer models of complex biological systems characterized by high-dimensional, non-linear, and computationally demanding behavior.

Key Features:

• Gaussian Process Modeling: Fits Gaussian process (GP) models to predict outputs from computational models and serve as statistical emulators.
• Sensitivity Analysis: Performs sensitivity analysis by using GP fits to identify and quantify the influence of input parameters on model outputs.
• Handling Complex Systems: Targets high-dimensional, non-linear biological system models that are computationally demanding.
• R Package Implementation: Provided as an R package for fitting GPs to outputs of computer simulations.
• Computational Efficiency: Uses surrogate GP models to reduce computational overhead when exploring and validating complex models.

Scientific Applications:

• Biological System Analysis: Enables analysis of dynamics in complex biological models by applying GPs to simulation outputs.
• Model Optimization and Validation: Supports refinement and validation of models by pinpointing key parameters via sensitivity analysis.
• Resource Efficiency in Modeling: Facilitates more efficient exploration and validation of computationally intensive biological models by reducing direct simulation costs.

Methodology:

Fits Gaussian processes to outputs generated by computer simulations and performs sensitivity analysis by systematically varying input parameters and observing resulting changes in outputs.