Uncertainty quantification in modeling of microfluidic T-sensor based diffusion immunoassay.

Area of Science:

• Biomedical Engineering
• Chemical Engineering
• Analytical Chemistry

Background:

• Quantitative measurements using T-sensors require comparing experimental data with model predictions.
• Accurate comparisons necessitate accounting for model prediction uncertainty arising from uncertain parameter values.

Purpose of the Study:

• To analyze uncertainty in T-sensor competitive diffusion immunoassay predictions.
• To quantify the impact of diffusion constants, binding kinetics, and flow speed on species concentrations.

Main Methods:

• Employed a stochastic uncertainty quantification method using polynomial chaos expansions.
• Represented the dependence of species concentrations on uncertain model parameters.

Main Results:

• Model parameter uncertainties induce significant, spatially varying uncertainty in predicted concentrations.
• Diffusivity of fluorescently labeled probe antigen is the dominant source of uncertainty.
• Uncertainty in fluorescence intensity is lowest near the T-sensor centerline.

Conclusions:

• Using centerline fluorescence intensity, instead of its first derivative, reduces uncertainty.
• This approach enhances detection sensitivity for measuring sample antigen concentration in T-sensor immunoassays.