RAPID AND RELIABLE SPECTRAL VARIABLE SELECTION FOR STATISTICAL CALIBRATIONS BASED ON PLS-REGRESSION VECTOR CHOICES

The optimum number of spectral variables necessary for analytical spec troscopy is still a subject of debate. For sensor applications using m iniaturized instrumentation, a small set of significant wavelengths wi th robust predictive performance is especially appreciated. A fast pro cedure is proposed, based on pairwise selection of spectral variables suggested by the weights of the optimum PLS-regression vector. The per formance of multiple linear regression models based on such choices wa s similar to, and sometimes improved on full spectrum based modeling. Several examples from clinical studies of blood substrate assays using attenuated total reflection infrared spectroscopy of biofluids are pr esented.