Analysis of surface plasmon resonance data using a partial least square regression method for glucose concentration estimation

A wavelength-based surface plasmon resonance (SPR) technique has been used for the measurement of glucose concentration in aqueous solution. Adoption of partial least square (PLS) regression modeling on SPR data with the proposed simple data-pretreatment method provides a much better model than using traditional minima-hunting with curve-fitting method. PLS gives the prediction error of 27.63 mg/dL with using unscrambler PLS-toolbox while the traditional method gives an error of 72.15 mg/dL.