A near infrared spectroscopy(NIRS) was applied to the rapid and quantitative analysis of active components in the extraction process of Bupleuri Radix.Total of 126 extraction samples were selected

of which the contents of total flavonoids and polysaccharide were determined by ultraviolet-visible spectrophotometry

while the saikosaponin A and saikosaponin D were determined by high performance liquid chromatography(HPLC)

and their spectra were collected in transmission mode.The quantitative calibration models for above four compounds were established by partial least squares(PLS)

which were optimized by using different pretreatment methods

spectral bands and the number of principle components.Results indicated that there existed good fitting of the four quantitative calibration models between predicted values and the reference values

which exhibited high prediction accuracy with correlation coefficients in prediction sets(RP) greater than 0.9.The root mean square errors of prediction(RMSEP) were 3.46 μg/mL

0.743 mg/mL

1.53 μg/mL

0.406 μg/mL for total flavonoids

polysaccharide

saikosaponin A and saikosaponin D

respectively

and the relative standard errors of prediction(RSEP) were 1.65%

8.28%

5.74% and 7.52%

respectively.All the results confirmed that the combination of NIRS and PLS could be successfully applied to the monitoring of active components in the extraction of Bupleuri Radix with the advantages of rapidness

accuracy

non-destruction and environmental friendly.