US 12,352,693 B2
Method for qualitative identification and quantitative determination of caffeine in drug
Yan Peng, Shanghai (CN); Linggao Zeng, Chongqing (CN); Yiming Zhu, Shanghai (CN); Qingrou Yang, Shanghai (CN); Chenjun Shi, Shanghai (CN); Xu Wu, Shanghai (CN); and Songlin Zhuang, Shanghai (CN)
Assigned to University of Shanghai for Science and Technology, Shanghai (CN); and CHONGQING INSTITUTE FORFOOD AND DRUG CONTROL, Chongqing (CN)
Filed by University of Shanghai for Science and Technology, Shanghai (CN); and CHONGQING INSTITUTE FOR FOOD AND DRUG CONTROL, Chongqing (CN)
Filed on Aug. 27, 2021, as Appl. No. 17/321,413.
Claims priority of application No. 202011030657.4 (CN), filed on Sep. 27, 2020.
Prior Publication US 2021/0381963 A1, Dec. 9, 2021
Int. Cl. G01N 21/35 (2014.01); G01N 21/3581 (2014.01)
CPC G01N 21/3581 (2013.01) [G01N 2021/3595 (2013.01)] 1 Claim
OG exemplary drawing
 
1. A method for qualitative identification and quantitative determination of caffeine in a drug, comprising:
(1) selecting n reference samples with different concentrations of caffeine; and scanning each of the n reference samples with a Fourier transform infrared spectrometer (FTIR) to obtain a terahertz absorption coefficient spectrum of each of the n reference samples;
(2) extracting frequency points, amplitudes and peak areas of at least one of characteristic peak of each of the concentrations of caffeine from the corresponding terahertz absorption coefficient spectrum obtained by step 1; and establishing concentration gradients between the concentrations and three characteristic quantities, respectively, wherein the three characteristic quantities are the frequency points, amplitudes and peak areas;
(3) grouping the frequency points, amplitudes and peak areas obtained by step 2 as characteristic vectors into two groups, wherein one group is used as training data, and another group is used as test data; and using the training data to establish training of a support vector regression (SVR) model to find out characteristic relationship between the concentrations of the caffeine and the characteristic vectors of the terahertz absorption coefficient spectra, thereby obtaining a trained SVR model; and
(4) scanning an analyte with the FTIR to obtain the terahertz absorption coefficient spectrum of the analyte, from which the frequency points, amplitudes and peak areas of the analyte are obtained; importing the frequency points, amplitudes and peak areas into the trained SVR model obtained by step 3; and performing a mathematical calculation to qualitatively identify and quantitatively determine the caffeine in the analyte;
wherein the characteristic peak is one or more of the characteristic peaks in the terahertz absorption coefficient spectrum of the caffeine.