Modelling of the immunoassay element of an optical immunosensor

Abstract

Several factors affect the viability of biosensor design. A computer-based model is being developed to enable the sources and effect of noise and variability within the sensor to be analysed. The work now presented details the modelling of the biochemical aspect of the biosensor model-the immunoassay. The equilibrium equations that describe the chemical reactions that occur when a sample containing the analyte is added to the immunosensor are cast as a sum of squares function that can be minimized using an optimization procedure. The optimization returns the concentrations of each species at equilibrium and the procedure is incorporated within a Monte Carlo simulation, which allows the variations in the resulting concentrations to be determined. Three classes of optimization technique are considered, classical regression techniques and two intelligent optimization techniques: simulated annealing and genetic algorithms. Several methods of imposing constraints are implemented and the issue of local minima is discussed. Classical regression procedures were found to be superior to the intelligent optimizations examined.