OBJECTIVE To propose a real-time release test for Dendrobii Officinalis Caulis powder using near-infrared spectroscopy and address the inefficiency of the current polysaccharide detection method(phenol-sulfuric acid assay), which suffers from prolonged analysis time and consequently reduces production efficiency.

METHODS  Initially, data on the short-wave and long-wave near-infrared spectrum and polysaccharide content of the powder were collected. Then, the spectral signals were processed using preprocessing, characteristic band, and a quantitative model. A pretreatment fusion algorithm enhanced the long-wave near-infrared spectral data quality. Finally, a rapid prediction model for polysaccharide content was chosen, with generated a predicted value distribution map, estimated probabilistic risk, and determined the release of the current batch of powder intermediates.

RESULTS  The root mean squared error of the preprocessing fusion algorithm in predicting the polysaccharide content on the test set was 1.297%, and the coefficient of determination was 0.924, indicating a small error between predicted and true values. The predicted value distribution effectively quantified the probabilistic risk of a sample.

CONCLUSION  The proposed method can be used for Dendrobii Officinalis Caulis powder release tests, reduces production time, and may be applicable to other medicinal and edible products.