UncertainSCI

UncertainSCI quantifies uncertainty in computational biomedical simulations by constructing polynomial chaos (PC) emulators to analyze how parameter variability affects model outputs.

Key Features:

• Advanced Parameter Sampling Techniques: Employs state-of-the-art sampling methods to construct polynomial chaos (PC) emulators.
• Polynomial Chaos Emulation: Models parameters as random variables and enables efficient computation of output statistics and sensitivities.
• Weighted Fekete Points: Uses weighted Fekete points derived from a randomized candidate set for near-optimal sampling and PC construction.
• Non-Intrusive Pipeline: Provides a non-intrusive pipeline for integration with existing software libraries and simulation suites.
• Demonstrated Biomedical Test Cases: Applied to modeling bioelectric potentials in the heart and electric stimulation in the brain to estimate variability, statistics, and sensitivities across parameters.

Scientific Applications:

• Cardiac bioelectric potential modeling: Estimates variability, statistics, and parameter sensitivities in models of cardiac bioelectric potentials.
• Brain electric stimulation modeling: Estimates variability, statistics, and parameter sensitivities in models of electric stimulation in the brain.
• Parametric uncertainty analysis in biomedicine and bioengineering: Quantifies propagation of parametric variability to model outputs in biomedicine and bioengineering simulations.

Methodology:

Employs sampling methods to build polynomial chaos (PC) emulators, models parameters as random variables, computes output statistics and sensitivities, selects weighted Fekete points from a randomized candidate set, and is implemented in Python with a non-intrusive integration approach.