Calibration of Surface Plasmon Resonance Refractometers Using Locally Weighted Parametric Regression

Abstract

A locally weighted parametric regression (LWPR) method has been developed and applied to the calibration of wavelength-modulated surface plasmon resonance (SPR) refractometers. This calibration method combines locally weighted regression (LWR) and nonlinear principal components regression (NPCR). It is shown that LWPR provides a calibration model that is more robust to random errors than traditional minima hunt calibration methods applied to SPR sensors. It is further shown that LWPR permits accurate and reliable calibration with low-resolution (32 nm) SPR data when the traditional methods fail.