An efficient method for signal improvement in near-infrared spectroscopy measurements during calibration and validation processes

► Hybrid algorithm cope with noise sources during spectroscopic measurements. ► Combination of spectral correlation, binning, and “Lucky Imaging” concept are made. ► Algorithm tested on blood samples at different hematocrit and glucose levels. ► Samples were exposed to different perturbations. ► A decrease in both standard error of prediction and bias was achieved.