Ginseng raw material distinguishing method based on UHPLC method and application

Abstract

The invention discloses a ginseng raw material discriminating method based on UHPLC (Ultra High Performance Liquid Chromatography) and application, a fingerprint spectrum containing 22 common peaks is established by optimizing chromatographic conditions and UPLC-QTOF-MS (Ultra Performance Liquid Chromatography-Quadrature Time of Flight-Mass Spectrometry) analysis, variables are screened on the basis of OPLS-DA (Orthogonal Partial Least Squares-Discriminant Analysis), a more efficient machine learning discriminating method is optimized, and a certain origin traceability effect is achieved. According to the method, the specifications of the ginseng powder, the ginseng extract and the ginseng unilateral deep-processed product can be quickly traced, a certain raw material producing area traceability effect is achieved, and the method has important significance on raw material judgment and quality control of ginseng series products.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of traditional Chinese medicine analysis, and particularly relates to a ginseng powder and ginseng extract raw material application specification raw material discrimination method based on ultra-high performance liquid chromatography (UHPLC), which can simultaneously realize accurate discrimination of ginseng powder, ginseng extract and other ginseng product raw material application specifications and origin analysis. BACKGROUND

[0002] Ginseng is the dried root and rhizome of Panax ginseng C. A. Mey. of the Araliaceae family, and has a long history of application in China. Ginseng mainly contains saponins, polysaccharides and other components, and has various effects such as anti-fatigue, anti-tumor, anti-inflammatory and neuroprotective effects. Ginseng is widely used in the food field in China. In 2024, the State Administration for Market Regulation issued the Technical Requirements for the Record Products of Health Food Raw Materials Ginseng, American Ginseng and Ganoderma lucidum, which clearly states that ginseng can be used as a health food raw material for record management. It is clearly stated that the raw material used for production record should be in line with the current Chinese Pharmacopoeia. However, it is difficult to distinguish the application specification of ginseng record health food after processing from the appearance, such as ginseng single product including tablets, hard capsules, powder and other dosage forms. It is difficult to directly identify the application specification of the raw material used for these processed products. Ginseng medicinal materials contain different parts such as rootlets and rhizome heads, which have certain differences in application specifications with medicinal slices. Due to the differences in components of different parts, it often causes differences in components. However, there is still a lack of effective discrimination method for these ginseng single processing products, and there is a lack of discrimination method for the application specification of the raw material of the record product. At present, although there are many studies on ginseng high-performance liquid chromatography and LC-MS, the saponin components in ginseng are very complex, and there is still a lack of method based on high-resolution chromatography conditions UHPLC with raw material specification discrimination and origin discrimination ability. SUMMARY

[0003] The purpose of the present application is to provide a ginseng powder and ginseng extract raw material application specification discrimination method based on ultra-high performance liquid chromatography (UHPLC), which can be used for origin analysis of ginseng. The present application establishes a fingerprint containing 22 common peaks by optimizing chromatographic conditions and UPLC-QTOF-MS analysis, filters the variables based on OPLS-DA, optimizes the more efficient machine learning discrimination method, and has a certain origin traceability effect. This method can realize the rapid traceability of ginseng powder, extract and ginseng single deep processing product specifications, and has a certain raw material origin traceability effect, which has important significance for the raw material discrimination and quality control of ginseng series products.

[0004] In order to achieve the above purpose, the technical scheme adopted by the present application is as follows: A ginseng raw material discrimination method based on UHPLC method, comprising the following steps: (1) Preparation of the control solution: accurately weigh a certain amount of ginsenoside R1 and ginsenoside Rg1, Re, Rf, Rb1, Rc, Ro, Rb2, Rd control substances, accurately weigh and add 80% methanol to prepare the control stock solution; (2) Preparation of the test solution: accurately weigh 0.5-2 g of ginseng powder sample, add 25 ml of 80% methanol and heat reflux for 3 h, cool down, make up the lost mass, filter, and take the filtrate, which is the test solution; (3) Chromatographic analysis conditions: use octadecylsilane-bonded silica gel as the filler of the chromatographic column, use acetonitrile as the mobile phase A and 0.02% phosphoric acid aqueous solution as the mobile phase B for gradient elution, and use the ultraviolet detector for sample injection and detection; (4) Establishment of the fingerprint spectrum: import the obtained test solution fingerprint spectrum into the traditional Chinese medicine chromatographic fingerprint similarity evaluation system (2012 edition) to generate the control fingerprint spectrum and calculate the similarity between the ginseng medicinal materials and pieces of different origins and the control fingerprint spectrum; (5) Qualitative analysis of the common peaks: use acetonitrile as the mobile phase A and 0.02% formic acid aqueous solution B for gradient elution, the solvent gradient is the same as that in step 3, use UPLC-QTOF-MS for qualitative analysis, and identify 17 common peaks: ginsenoside R1, ginsenoside Rg1, ginsenoside Re, ginsenoside Rf, ginsenoside R2, ginsenoside Rg2, ginsenoside Rb1, ginsenoside mRb1, ginsenoside Rc, ginsenoside Ro, malonyl ginsenoside Rc, ginsenoside Rb2, malonyl ginsenoside Rb2, panaxoside R1, ginsenoside Rd, malonyl ginsenoside Rd, and panaxoside III; (6) OPLS-DA analysis and differential component screening: collect the peak areas of the common peaks of the sample fingerprint spectrum as variables for OPLS-DA analysis, and perform discriminant analysis on samples of different specifications and different origins to screen out the differential components.

[0005] Further, the concentrations of ginsenoside R1 and ginsenoside Rg1, Re, Rf, Rb1, Rc, Ro, Rb2, and Rd in the control stock solution of step 1 are 33.46, 115.31, 126.09, 39.88, 247.15, 102.78, 52.68, 74.15, and 27.34 μg·mL -1 .

[0006] Further, the sample preparation method described in step 2 is the preparation method of ginseng medicinal materials and pieces of powder samples. The sample preparation method of ginseng extract, ginseng products (granules, capsules) includes but is not limited to ultrasonic extraction and reflux extraction, filtration, and taking the filtrate, which is the relevant test solution.

[0007] Further, the detection condition of the ultra-high performance liquid chromatography in step 3 is as follows: the chromatographic column is a C18 chromatographic column; the elution gradient of the mobile phase is 0~9 min, 18%~20% A; 9~13 min, 20%~28% A; 13~37 min, 28%~35% A; 37~40 min, 35%~90% A; the detection wavelength is 203 nm; the volume flow rate is 0.4 mL / min; the column temperature is 30 ℃; and the injection volume is 2 μL.

[0008] Further, the relative retention times of the 22 common peaks in step 4 are 4.30 (peak 1), 5.27 (peak 2), 7.64 (peak 3), 10.11 (peak 4), 10.69 (peak 5), 15.16 (peak 6), 16.59 (peak 7), 17.59 (peak 8), 19.05 (peak 9), 20.29 (peak 10), 20.72 (peak 11), 21.87 (peak 12), 22.21 (peak 13), 22.51 (peak 14), 22.75 (peak 15), 23.52 (peak 16), 24.21 (peak 17), 24.92 (peak 18), 25.55 (peak 19), 28.03 (peak 20), 29.61 (peak 21), 35.86 (peak 22), respectively. The similarity of each sample spectrum to the control spectrum is above 0.99. R

[0009] Further, the mapping relationship of the 17 qualitative chromatographic peaks and the common peaks in the fingerprint is as follows: notoginsenoside R1 (peak 3), ginsenoside Rg1 (peak 4), ginsenoside Re (peak 5), ginsenoside Rf (peak 7), notoginsenoside R2 (peak 8), ginsenoside Rg2 (peak 9), ginsenoside Rb1 (peak 11), ginsenoside mRb1 (peak 12), ginsenoside Rc (peak 13), ginsenoside Ro (peak 14), malonyl-ginsenoside Rc (peak 16), ginsenoside Rb2 (peak 17), malonyl-ginsenoside Rb2 (peak 18), ginsenoside R1 (peak 19), ginsenoside Rd (peak 20), malonyl-ginsenoside Rd (peak 21), ginsenoside III (peak 22).

[0010] ​Further, step 6 adopts OPLS-DA to perform sample application specification, raw material origin discriminant analysis and differential component screening, in the raw material specification (medicinal material / decoction piece source) discriminant model, the model AUC is all 1, 11 differential components ginsenoside III, ginsenoside Rd, ginsenoside Rb2, ginsenoside Rc, peak 15, ginsenoside Rb1, peak 6, malonyl ginsenoside Rc, ginsenoside mRb1, malonyl ginsenoside Rd, malonyl ginsenoside Rb2 in ginseng medicinal materials and decoction pieces are screened out; and in the origin model (Jilin / other origins), the model reaches a good origin discriminant effect at the same time, and the screened origin differential components are ginsenoside R1, ginsenoside Rg1, peak 15, peak 2, peak 1 and sanchinoside R2 in turn, which are judged as the representative differential components of the two groups of origins.

[0011] Further, step 6 can also adopt a machine learning method, takes the 11 differential common peaks screened by the whole component or specification discriminant model as variables for comparison, and screens the parameters of machine learning, to determine the best method and condition for discriminant analysis of medicinal materials and decoction pieces.

[0012] Further, the machine learning method in step 7 includes but is not limited to SVM, RF, RBF, ELM, CNN and BP, based on grid search parameter screening and optimization of differential component variables, the discriminant accuracy of the training set and the test set can all reach 100%.

[0013] Compared with the prior art, the present application has the following beneficial effects: (1) The health food products recorded products required according to the current Chinese Pharmacopoeia of Chinese herbal medicine decoction pieces, but after ginseng is pulverized, extracted or finely processed, the ginseng raw material selected by the product cannot be identified by appearance, and the present application can provide a rapid discriminant method based on the whole component.

[0014] (2) The present application establishes a stable discriminant model through parameter optimization, and further develops an efficient machine learning method for the 11 differential ginsenoside components screened, which can realize rapid discriminant analysis of sample raw materials, and is more convenient and efficient to operate.

[0015] (3) The present application can have certain origin discriminant ability while carrying out raw material specification discriminant analysis, can realize discriminant analysis of Jilin ginseng and ginseng from other origins, and has a certain raw material origin traceability application prospect. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the embodiments will be briefly introduced as follows. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as limiting the scope. For those skilled in the art, other related drawings can also be obtained without creative labor on the premise of the drawings.

[0017] Figure 1 R is a ginseng medicinal material (A) and piece (B) fingerprint and control map; Figure 2 R is a ginseng sample total ion flow chart; Figure 3 R is a ginseng powder different solvent extraction effect spectrum chart; Figure 4 R is a ginseng powder different extraction method spectrum chart; Figure 5 R is a ginseng powder different solvent concentration spectrum chart comparison; Figure 6 R is a ginseng powder different extraction time spectrum chart comparison; Figure 7 R is a ginseng sample specification OPLS-DA model score chart; Figure 8 R is a sample specification OPLS-DA model permutation validation chart; Figure 9 R is a sample specification OPLS-DA model VIP chart; Figure 10 R is a sample specification OPLS-DA model ROC curve; Figure 11 R is a sample specification machine learning model confusion matrix and ROC curve representative example; Figure 12 R is a production place OPLS-DA model score chart; Figure 13 R is a production place OPLS-DA model permutation validation chart; Figure 14 R is a production place OPLS-DA model ROC curve; Figure 15 R is a ginseng powder, extract and different dosage form product fingerprint comparison. DETAILED DESCRIPTION

[0018] The embodiments of the present application will be described in detail below with reference to the embodiments, but those skilled in the art will understand that the following embodiments are only used to illustrate the present application, and should not be regarded as limiting the scope of the present application. The specific conditions are not specified in the embodiments, which are carried out according to the conventional conditions or the conditions recommended by the manufacturer. The reagents or instruments used are conventional products that can be purchased on the market.

[0019] Determination method of ginseng powder fingerprint Accurately weigh about 1 g of ginseng medicinal material / powder (pass through No. 4 sieve), add 25 ml of 80% methanol, heat and reflux for 3 h, cool, supplement the lost mass, filter, and take the filtrate to obtain the ginseng powder test solution.

[0020] Accurately weigh a proper amount of notoginseng saponin R1 and ginsenoside Rg1, Re, Rf, Rb1, Rc, Ro, Rb2, Rd reference substances, add 80% methanol to prepare a mixed reference solution with a concentration of 33.46, 115.31, 126.09, 39.88, 247.15, 102.78, 52.68, 74.15, and 27.34 μg·mL -1 .

[0021] Use octadecylsilane-bonded silica gel as the filler; use acetonitrile (A)-0.02% phosphoric acid solution (B) as the mobile phase for gradient elution, 0~9 min, 18%~20% A; 9~13 min, 20%~28% A; 13~37 min, 28%~35% A; 37~40 min, 35%~90% A; the detection wavelength is 203 nm; the volume flow rate is 0.4 mL / min; the column temperature is 30 ℃; the injection amount is 2 μL. The theoretical plate number should not be less than 10,000 calculated by ginsenoside Rb1.

[0022] Accurately take 2 μl of the reference solution and the test solution, respectively, inject into the ultra-high liquid chromatograph, measure, and record the chromatogram.

[0023] Methodology investigation of ginseng powder fingerprint Precision: Take 1 g of the same batch of ginseng powder sample (S1), accurately weigh, prepare the test solution according to the preparation method of the test solution in Example 1, and measure 6 times continuously under the chromatographic condition method.

[0024] Repeatability: Take 1 g of the same batch of ginseng powder sample (S1), accurately weigh, prepare 6 test solutions in parallel according to the preparation method of the test solution in Example 1, and measure under the same chromatographic condition.

[0025] Stability: Take 1 g of the same batch of ginseng powder sample (S1), accurately weigh, prepare the test solution according to the preparation method of the test solution in Example 1, and measure under the chromatographic condition method after preparation for 0, 4, 8, 12, 16, and 24 h.

[0026] The above precision, repeatability, stability methodological test investigations were all with 22 common chromatographic peaks as the investigation objects, with peak 11 (ginsenoside Rb1) retention time and peak area as the controls, and with the RSD values of the relative retention times (RRTs) and relative peak areas (RPAs) of the other chromatographic peaks to evaluate the feasibility of the methodological experiments. The test results showed that the RSD values of the RRTs and RPAs of each chromatographic peak in the precision test, repeatability test and stability test were all less than 3%.

[0027] Example 3 Ginseng powder fingerprint spectrum establishment and common peak calibration The chromatogram data was imported into the “Traditional Chinese Medicine Chromatographic Fingerprint Spectrum Similarity Evaluation System (2012 Edition)”, and after the median method and multi-point correction, the chromatographic peak matching was performed, and 22 common peaks were calibrated; the superimposed spectrum and the control fingerprint spectrum R of the ginseng powder determined from different sources (medicinal materials and decoction pieces) were Figure 1 . With peak 11 as the reference peak S, the RSD values of the relative retention times (RRTs) of the common peaks between each batch of medicinal materials / decoction pieces were between 0.02% and 0.48%, and the similarity with each control spectrum R was above 0.99 (see Table 1).

[0028] Table 1 Ginseng medicinal materials (S1-25) and their decoction pieces (D1-25) sample and control spectrum R similarity evaluation results

[0029] Example 3 Ginseng powder fingerprint spectrum common peak UPLC-QTOF-MS analysis Octadecylsilane-bonded silica gel was used as the filler; acetonitrile (A)-0.02% formic acid aqueous solution (B) was used as the mobile phase for gradient elution, 0-9 min, 18%-20% A; 9-13 min, 20%-28% A; 13-37 min, 28%-35% A; 37-40 min, 35%-90% A. Mass spectrometry was performed in negative ion mode detection, full scan mode, spray voltage 3.0 kV, ion source temperature 120°C, sheath gas flow rate 800 L / h, collision gas 7 psi, collision energy 6-40 eV, desolvation temperature 500°C, mass range m / z 50-1200.

[0030] Because the mobile phase used formic acid aqueous solution, the common peak identification information was further compared with the control substance, especially ginsenoside Rc which was greatly affected by the acid condition. The qualitative analysis results of all common peaks are shown in Table 2, and the total ion current chromatogram of the ginseng sample is shown in Figure 2 .

[0031] Table 2 Common peak mass spectrometry qualitative analysis results

[0032] Note: * Indicates that the component has been verified by using a reference substance.

[0033] Example 4 Ginseng powder extraction solvent type and extraction method investigation First, in the case of consistent time, the extraction effects of methanol and ethanol were compared and analyzed by using reflux extraction method. The comparison chromatogram is shown in Figure 3 According to the peak area comparison results, the extraction effect of ethanol, especially for high-grade saponins, is not as good as that of methanol. Therefore, methanol is selected as the extraction solvent.

[0034] Further, the extraction effects of ginseng powder were compared by using reflux extraction and ultrasonic extraction. The results showed that the extraction effect of reflux extraction was slightly better than that of ultrasonic extraction for multiple saponins. The spectrum is shown in Figure 4 Therefore, reflux extraction is selected to further investigate the test sample preparation method.

[0035] Example 5 Ginseng powder solvent concentration investigation The extraction effects of 20% methanol, 40% methanol, 60% methanol, 80% methanol and methanol were compared by fixing the extraction time to 3h. The results showed that the extraction effects of chromatographic peaks before ginsenoside Rb1 (20.72 min) were similar, but the chromatographic peaks after 20.72 min were greatly affected by the solvent concentration. With the increase of solvent concentration, the peak area gradually increased, and the extraction effect of 80% ethanol reached the best, and the extraction effect of methanol was slightly lower than that of 80% ethanol (spectrum see Figure 5 Therefore, the extraction solvent is determined to be 80% ethanol. Example 6 Ginseng powder extraction time investigation 80% ethanol was used to extract 1, 2, 3 and 4h respectively, and the extraction effects were compared. The results showed that the best extraction effect could be achieved after 3h of extraction (spectrum see Figure 6 Therefore, the extraction time is determined to be 3h considering the test sample preparation time and efficiency.

[0036] Example 7 OPLS-DA analysis method of ginseng raw material specifications based on UHPLC fingerprint The OPLS-DA analysis of 25 batches of ginseng and its decoction pieces samples was carried out by using 22 common peaks as variables. The results showed that the ginseng medicinal materials and decoction pieces samples could be effectively distinguished by OPLS-DA (see Figure 7 ). The independent variable fitting index ( R 2 X ), the dependent variable fitting index ( R 2 Y ) and the model prediction index ( Q 20.814, 0.966 and 0.926, respectively, indicating that the fitting results were acceptable; 200 times of permutation test showed that, R 2 and Q 2 The intercepts of the regression lines with the longitudinal axis were less than 0.3 and 0.05, respectively, indicating that there was no overfitting of the model (see Figure 8 ), and the analysis results based on the method could be used for the source analysis of ginseng powder. Taking "VIP value > 1" and "P < 0.05" as thresholds, 11 components with larger contribution values were screened out, in order of Panax quinquefolium saponin III, ginsenoside Rd, ginsenoside Rb2, ginsenoside Rc, peak 15, ginsenoside Rb1, peak 6, malonyl ginsenoside Rc, ginsenoside mRb1, malonyl ginsenoside Rd, and malonyl ginsenoside Rb2, which were judged as representative difference components between medicinal materials and decoction pieces (see Figure 9 ). Meanwhile, the receiver operating characteristic (ROC) curve was used to test the classification ability of the model, and the area under the curve (AUC) was used for evaluation. When AUC was 0.9 or above, it was considered to have high accuracy. The AUC of the model for medicinal materials and decoction pieces was 1 (see Figure 10 ), indicating that the model had strong classification ability.

[0037] Example 8 Machine learning discrimination method based on different common peak selection modes Taking common peaks as variables, six kinds of machine learning algorithms, including SVM, RF, RBF, ELM, CNN, and BP, were used to discriminate the specifications of ginseng powder. All samples were divided into training set and test set (ratio 8:2) according to Kennard-Stone algorithm, and the optimal machine learning parameters were selected through hyperparameter screening. The model performance was evaluated by accuracy (Acc), precision (Pr), recall (Re), and F1 score (F1). The results showed that when all common peaks were used as variables, the parameters of the model were as shown in Table 3. The Acc, Pr, Re, and F1 of the six kinds of machine learning models were at a relatively optimal level. Most of the models (SVM, RF, CNN, and BP) showed perfect performance, but RBF and ELM slightly decreased on the test set and training set, respectively, but the performance indicators were all above 90%.

[0038] Table 7 Comparison of machine learning model parameters and performance for specification discrimination

[0039] Taking the difference components screened out in Example 4 as analysis variables, the analysis was carried out after optimization of the number of variables. The analysis results showed that the performance indicators of all models were 100%, and the model exhibited stronger robustness and generalization ability, as shown in the results of the confusion matrix and ROC curve of the model Figure 11 .

[0040] Example 9: Application of UHPLC fingerprinting in the identification of the origin of ginseng raw materials OPLS-DA analysis was performed on 25 batches of ginseng medicinal material samples and 22 common peak variables. The results showed that OPLS-DA could effectively distinguish samples from Jilin Province from samples from the other two provinces (Liaoning and Heilongjiang) (see the OPLS-DA score chart for place of origin). Figure 12 The fit index of the independent variable in the analysis (). R 2 X The value was 0.585, and the dependent variable fit index ( R 2 Y The model prediction index is 0.724. Q 2 The value is 0.610. R 2 and Q 2 All values ​​exceeded 0.5, indicating that the model's fitting results are acceptable; furthermore, through 200 permutation tests, R 2 and Q 2 The intercepts of the regression lines on the vertical axis are both less than 0.3 and 0.05 (see the permutation verification chart). Figure 13 This indicates that the model does not overfit. It can be used to differentiate ginseng from Jilin province and ginseng from other two provinces (Liaoning and Heilongjiang). The differentially expressed components between Jilin and other regions were, in descending order, ginsenoside R1, ginsenoside Rg1, peak 15, peak 2, peak 1, and notoginsenoside R2, to determine the representative differential components between the two groups. ROC analysis (see...) Figure 14 This also indicates that the model is stable and reliable, and can distinguish between samples from Jilin and other origins. Therefore, the method described in this patent has certain application potential in origin identification.

[0041] Example 10: Comparison of UHPLC chromatograms of ginseng powder, ginseng extract, ginseng granules, and ginseng capsules The differences in UHPLC analysis of ginseng powder, ginseng extract, ginseng granules, and ginseng capsules were mainly due to differences in sample preparation. Since the ginseng extract, ginseng granules, and ginseng capsules in this example had already undergone extraction, the sample solution was prepared using the method of "ultrasonication with 80% ethanol for 30 min, filtration, and collection of the filtrate." The preparation method for ginseng powder was the same as in Example 1, and the sample weight was calculated based on 1 g of raw drug. For injection and analysis, the chromatograms were imported into the "Traditional Chinese Medicine Chromatographic Fingerprint Similarity Evaluation System (2012 Edition)" for comparison. The similarity R between each chromatogram and the reference chromatogram was above 0.99, indicating high similarity and good peak separation (see...).Figure 15 ). It is indicated that this method can be used for the specification identification analysis of extract and single-ingredient preparation products of ginseng.

Claims

1. A method for discriminating ginseng raw materials based on a UHPLC method, characterized by, Comprising the following steps: (1) Preparation of reference solution: accurately weigh a certain amount of ginsenoside R1 and ginsenoside Rg1, Re, Rf, Rb1, Rc, Ro, Rb2, Rd reference substances, accurately weigh and add 80% methanol to prepare reference stock solution; (2) Preparation of test solution: accurately weigh 0.5-2 g of ginseng powder sample, add 25 ml of 80% methanol and heat reflux for 3 h, cool, supplement the lost mass, filter, and take the filtrate, which is obtained; (3) Chromatographic analysis conditions: use octadecylsilane bonded silica gel as the filler of the chromatographic column, use acetonitrile as the mobile phase A and 0.02% phosphoric acid aqueous solution as the mobile phase B for gradient elution, and use ultraviolet detector for sample detection; (4) Establishment of fingerprint spectrum: import the obtained test solution fingerprint spectrum into the Traditional Chinese Medicine Chromatographic Fingerprint Spectrum Similarity Evaluation System 2012 edition, generate the control fingerprint spectrum, and calculate the similarity between the ginseng medicinal materials fingerprint spectra of different origins and the control fingerprint spectrum; (5) Qualitative analysis of common peaks: use acetonitrile as the mobile phase A and 0.02% formic acid aqueous solution B for gradient elution, the solvent gradient is the same as step 3, use UPLC-QTOF-MS for qualitative analysis, and identify 17 common peaks: ginsenoside R1, ginsenoside Rg1, ginsenoside Re, ginsenoside Rf, ginsenoside R2, ginsenoside Rg2, ginsenoside Rb1, ginsenoside mRb1, ginsenoside Rc, ginsenoside Ro, malonyl ginsenoside Rc, ginsenoside Rb2, malonyl ginsenoside Rb2, panaxoside R1, ginsenoside Rd, malonyl ginsenoside Rd, and panaxoside III; (6) OPLS-DA analysis and differential component screening: collect the peak areas of the common peaks of the sample fingerprint spectra from medicinal materials and slices as variables for OPLS-DA analysis, perform discriminant analysis on samples of different specifications and different origins, and screen out differential components.

2. The method of claim 1, wherein, The concentrations of notoginsenoside R1 and ginsenosides Rg1, Re, Rf, Rb1, Rc, Ro, Rb2, Rd in the control stock solution of step (1) are 33.46, 115.31, 126.09, 39.88, 247.15, 102.78, 52.68, 74.15, and 27.34 μg·mL -1 , respectively.

3. The method of claim 1, wherein, The sample preparation method in step (2) is the preparation method of ginseng medicinal materials and slice powder samples. Ginseng extract and ginseng product samples include but are not limited to ultrasonic extraction and reflux extraction, filtration, and taking the filtrate, which is obtained.

4. The method of claim 1, wherein, The ultra-high liquid chromatography detection conditions in step (3) are as follows: the chromatographic column is a C18 chromatographic column; the mobile phase elution gradient is 18%-20% A for 0-9 min, 20%-28% A for 9-13 min, 28%-35% A for 13-37 min, 35%-90% A for 37-40 min; the detection wavelength is 203 nm; the volume flow rate is 0.4 mL / min; the column temperature is 30°C; and the sample injection amount is 2 μL.

5. The method of claim 1, wherein, The relative retention times of the 22 common peaks in step (4) are 4.30 (peak 1), 5.27 (peak 2), 7.64 (peak 3), 10.11 (peak 4), 10.69 (peak 5), 15.16 (peak 6), 16.59 (peak 7), 17.59 (peak 8), 19.05 (peak 9), 20.29 (peak 10), 20.72 (peak 11), 21.87 (peak 12), 22.21 (peak 13), 22.51 (peak 14), 22.75 (peak 15), 23.52 (peak 16), 24.21 (peak 17), 24.92 (peak 18), 25.55 (peak 19), 28.03 (peak 20), 29.61 (peak 21), and 35.86 (peak 22). The similarity of each sample spectrum to the control spectrum is above 0.

99. R ​ 6. The method of claim 1, wherein, The mapping relationship between the 17 qualitative chromatographic peaks in step (5) and the common peaks in the fingerprint is as follows: notoginsenoside R1 (peak 3), ginsenoside Rg1 (peak 4), ginsenoside Re (peak 5), ginsenoside Rf (peak 7), notoginsenoside R2 (peak 8), ginsenoside Rg2 (peak 9), ginsenoside Rb1 (peak 11), ginsenoside mRb1 (peak 12), ginsenoside Rc (peak 13), ginsenoside Ro (peak 14), malonyl-ginsenoside Rc (peak 16), ginsenoside Rb2 (peak 17), malonyl-ginsenoside Rb2 (peak 18), panaxoside R1 (peak 19), ginsenoside Rd (peak 20), malonyl-ginsenoside Rd (peak 21), panaxoside III (peak 22).

7. The method of claim 1, wherein, In step (6), OPLS-DA is used for sample application specification, raw material origin discrimination analysis and differential component screening. In the raw material specification discrimination model, the model AUC is 1, and 11 differential components, panaxoside III, ginsenoside Rd, ginsenoside Rb2, ginsenoside Rc, peak 15, ginsenoside Rb1, peak 6, malonyl-ginsenoside Rc, ginsenoside mRb1, malonyl-ginsenoside Rd and malonyl-ginsenoside Rb2, are screened out in the ginseng medicinal materials and decoction pieces. In the origin model, the model achieves good origin discrimination effect, and the screened origin differential components are panaxoside R1, ginsenoside Rg1, peak 15, peak 2, peak 1 and notoginsenoside R2 in turn, which are the representative differential components of the two groups of origins.

8. The method of claim 7, wherein, In step (6), machine learning can also be used, and all components or 11 differential common peaks screened by the specification discrimination model are used as variables for comparison, and the parameters of machine learning are screened to determine the best method and condition for discriminating and analyzing the medicinal materials and decoction pieces.

9. The method of claim 8, wherein, The machine learning method includes but is not limited to SVM, RF, RBF, ELM, CNN and BP. Based on grid search parameter screening and optimization of differential component variables, the discrimination accuracy of the training set and the test set can reach 100%.

Images