Multivariate calibration applied to near-infrared spectroscopy for the quantitative analysis of dilute aqueous solutions

The penetration depths possible with near-infrared spectroscopy make it well suited for reagentless monitoring of analytes in body fluids or noninvasive monitoring of human tissue. As an initial step in achieving these goals, we have conducted near-infrared in-vitro experiments of dilute aqueous solutions containing analytes of physiological importance. By combining partial least squares (PLS) multivariate calibration methods with Latin Hypercube statistical designs, we have obtained precise near-infrared spectral determinations of urea, creatinine, and NaCl in dilute aqueous solutions. Cross-validated PLS calibrations for the three analytes and temperature were very precise and resulted in R2 values greater than 0.997.