Measurement of triglycerides concentration in human serum using near-infrared transmission spectroscopy and interval PLS

In order to measurement of Triglycerides in human serum with reagent-less using near-infrared (NIR) spectroscopy. Interval partial least square (iPLS) was proposed as an effective variable selection approach for multivariate calibration. For this purpose, an independent sample set was employed to evaluate the prediction ability of the resulting model. The spectrum was split into different interval. Then, the informative region of Triglycerides (1654-1746nm), in which the PLS model has a low RMSEP with 0.157mmol/L and a high R with 0.967, is selected with 18 intervals. The results show that the informative region of Triglycerides can be obtained by iPLS and applied to design the simpler reagent-less NIR instruments for inexpensive Triglycerides measurement in future.