Quantitative modeling and rapid detection method for total saponins in processed panax notoginseng and products thereof

Abstract

The application discloses a kind of dried panax notoginseng and its product total saponin content near infrared quantitative modeling and fast detection method, belong to the quality rapid analysis technical field of traditional Chinese medicine.For the distortion of the determination result of the current total saponin content determination method of the product, and the industry pain point that traditional HPLC detection is time-consuming and cannot be monitored on production line, the application first uses 70%~90% methanol aqueous solution reflux extraction as true value calibration means;Using near infrared spectrum and combining specific modeling technology, a specific quantitative model of dried panax notoginseng and its product is constructed, which is used for on-site quality monitoring of the production process of the product.The constructed detection method only needs to crush and sieve the sample, then scan its near infrared spectrum, and combine the pre-designed calculation software to output the total saponin result within 60s without organic solvent and complex pretreatment.The model determination coefficient R²=0.999 can be adapted to the whole process of on-site quality control of dried panax notoginseng, such as steaming, slicing, crushing and subpackaging, and can fully reflect the real saponin content level of medicinal materials.

Description

Technical Field

[0001] This invention belongs to the field of rapid detection and analysis technology of effective components in traditional Chinese medicine, specifically involving a method for accurately determining the total saponin content of three types of processed Panax notoginseng products: processed dried Panax notoginseng, processed dried Panax notoginseng slices, and processed Panax notoginseng powder, as well as a complete set of rapid quantitative detection methods based on near-infrared spectroscopy. This invention can be integrated with near-infrared spectroscopy cloud computing and enterprise quality cloud platforms to achieve real-time uploading of production line data and batch traceability and control of medicinal material quality. Background Technology

[0002] Panax notoginseng saponins are the core active substances in processed Panax notoginseng, mainly comprising components such as notoginseng saponin R1, ginsenoside Rg1, and ginsenoside Rb1. Their content is a core indicator for evaluating the quality of processed Panax notoginseng raw materials, slices, and powders. Current standards uniformly use pure methanol refluxed in an 80℃ water bath for 2 hours as the extraction process for processed Panax notoginseng samples, combined with high-performance liquid chromatography (HPLC) to determine the total saponin content. Existing technologies suffer from three major inherent limitations:

[0003] Defect 1: The pure methanol extraction system cannot efficiently overcome the gelatinization barrier of cooked Panax notoginseng starch, resulting in a low saponin extraction rate and severely distorted test results.

[0004] After processing such as steaming, boiling, and simmering, Panax notoginseng undergoes complete starch gelatinization, forming a dense colloidal film that encapsulates saponin molecules. Pure methanol, with its singular polarity, cannot fully wet and swell the gelatinized starch colloid. Consequently, a large amount of saponins remain trapped within the gelatinized starch, making them difficult to dissolve. HPLC readings are significantly lower than the actual total saponin content of the sample, failing to accurately reflect the quality of the processing. Existing literature and near-infrared spectroscopy patents only optimize extraction processes for raw Panax notoginseng, without addressing the issue of starch gelatinization and saponin encapsulation in processed Panax notoginseng by designing or improving extraction solvent systems.

[0005] Defect 2: There is no dedicated rapid assay model for monitoring total saponins during the processing of processed Panax notoginseng, making it difficult to monitor saponin quality levels in a timely manner. Traditional offline HPLC detection procedures are cumbersome, consuming large amounts of toxic organic solvents such as methanol and acetonitrile. Pretreatment and detection of a single batch of samples takes more than 4 hours, which cannot meet the real-time online quality control requirements of traditional Chinese medicine production lines.

[0006] Defect 3: Existing near-infrared modeling schemes for Panax notoginseng only use powder of a single particle size to establish a calibration set, without considering the baseline offset caused by the difference in particle size of cooked Panax notoginseng. The model is only applicable to powder with a fixed sieve size, has poor versatility, and cannot be adapted to Panax notoginseng powder produced by different grinding equipment in the factory.

[0007] In summary, existing technologies suffer from three major drawbacks: distortion in true value determination, low detection efficiency, and poor versatility of models due to granularity interference. Currently, there is no complete near-infrared quantitative detection solution specifically for Panax notoginseng that can solve all of the above problems at the same time. Summary of the Invention

[0008] The purpose of this invention is as follows:

[0009] The primary objective of this invention is to provide a high-extraction-rate HPLC method for the true value detection of total saponins from processed Panax notoginseng, thereby solving the problem of distorted results in existing methods for detecting the content of total saponins from processed Panax notoginseng.

[0010] The second objective of this invention is to provide a modeling method for constructing a near-infrared quantitative model of Panax notoginseng and its products with granular variable factors;

[0011] The third objective of this invention is to provide a non-destructive rapid detection process that is solvent-free, easy to operate, and produces results within 1 minute, thereby enabling on-site quality monitoring during the production of Panax notoginseng and its products.

[0012] This invention is implemented as follows:

[0013] The implementation process of a near-infrared quantitative modeling method for the total saponin content of processed Panax notoginseng and its products mainly includes the following steps: preparation of calibration set samples, detection of the true value of total saponin in the samples, acquisition of raw near-infrared spectra, spectral preprocessing, principal component dimensionality reduction, and construction of quantitative models.

[0014] The method for preparing the calibration set samples is as follows:

[0015] More than 100 processed Panax notoginseng products from different origins, with different cooking processes and processing forms were selected as calibration set samples. These products included dried Panax notoginseng, dried Panax notoginseng slices, and dried Panax notoginseng powder. For dried Panax notoginseng and dried Panax notoginseng slices, each batch of samples was divided into 5 portions and crushed separately, and then passed through sieves No. 3, No. 4, No. 5, No. 6, and No. 7 to obtain 5 different particle size powders. For dried Panax notoginseng powder, no secondary crushing was required, and it was directly used as the particle size sample to be tested.

[0016] The method for determining the true value of total saponins in a sample is as follows:

[0017] Z1. Take 0.6g of each particle size powder, weigh accurately, add 50mL of 70%~90% methanol aqueous solution, weigh and record; reflux gently in a water bath at 80℃ for 2h, cool to room temperature, make up the weight loss with methanol aqueous solution of corresponding concentration, shake well, filter through a filter membrane, and take the filtrate as the test sample.

[0018] Z2. Liquid Chromatography Conditions: An octadecylsilane-bonded silica C18 column was used; mobile phase A was acetonitrile, and mobile phase B was purified water; gradient elution program: 0–20 min, phase A 20%; 20–45 min, phase A 20% linearly increased to 46%; 45–55 min, phase A 46% linearly increased to 55%; 55–60 min, phase A maintained at 55%; detection wavelength 203 nm; theoretical plate number calculated based on the R1 peak of Panax notoginseng saponin was not less than 4000.

[0019] Z3. Preparation of reference solution: Dissolve in methanol to prepare a mixed reference solution, each 1 mL containing ginsenoside Rg1 0.4 mg, ginsenoside Rb1 0.4 mg, and notoginsenoside R1 0.1 mg;

[0020] Z4. Detection: Accurately inject 10 μL each of the reference solution and the test solution into the liquid chromatograph, and calculate the total saponin content using the external standard method;

[0021] The method for acquiring near-infrared raw spectra is as follows:

[0022] Powders of various particle sizes were spread evenly on a near-infrared sample cup, and the original diffuse reflectance spectra of the entire wavelength range of 900–2500 nm were collected.

[0023] The spectral preprocessing methods are as follows:

[0024] The first step is to perform standard normal transformation (SNV) on all raw spectra to eliminate intraday and interday instrument baseline fluctuation errors; the second step is to perform first derivative processing on the SNV-corrected spectra to amplify the characteristic absorption signals of saponins; both processing steps are performed synchronously across the entire band without segmentation.

[0025] The principal component dimensionality reduction method is as follows:

[0026] Principal component analysis was performed on all wavelength data in the characteristic band of 950–2250 nm of the preprocessed spectrum. The first 18 principal components were extracted as independent variables of the model, and the calculated coefficients of the principal components were stored simultaneously.

[0027] The quantitative model construction method is as follows:

[0028] The 18 principal component independent variables obtained by dimensionality reduction were fitted with the total saponin true value dependent variable of the corresponding sample in S2 by multiple linear regression to obtain the near-infrared quantitative detection model of total saponins of Panax notoginseng and its products.

[0029] A rapid detection method for total saponins of Panax notoginseng and its products includes the following steps: sample pretreatment, near-infrared spectral acquisition, spectral preprocessing, characteristic band dimensionality reduction calculation, and content quantification calculation. Among these steps, spectral preprocessing, characteristic band dimensionality reduction calculation, and content quantification calculation can be automatically calculated by pre-programmed software.

[0030] Sample pretreatment: Pulverize and sieve cooked Panax notoginseng slices, or mix them thoroughly; mix cooked Panax notoginseng powder directly and set aside to obtain the powder to be tested.

[0031] Near-infrared spectroscopy acquisition: The powder to be tested is evenly spread on the sample cup, and the diffuse reflectance raw spectrum of 900-2500 nm is acquired;

[0032] Spectral preprocessing: The original spectrum is subjected to standard normal transformation and first derivative processing in sequence;

[0033] Feature band dimensionality reduction calculation: Extract the 950-2250nm band data from the preprocessed spectrum, and calculate the first 18 principal components using the principal component conversion coefficients stored in claim 1;

[0034] Content quantification: Substitute the 18 principal components into the constructed multiple linear regression model to directly output the total saponin content detection results of the sample.

[0035] The beneficial effects of this invention are:

[0036] (1) The accuracy of sample true value determination is significantly improved.

[0037] The total saponin content of the test sample obtained by 80% methanol water extraction was 8.3% to 22.7% higher than that of the test sample obtained by pure methanol extraction in the pharmacopoeia, eliminating the detection distortion caused by starch gelatinization.

[0038] (2) Particle size correction significantly improves accuracy

[0039] Using a multi-particle-size cooked Panax notoginseng powder sample joint modeling method, the RMSEP of the prediction set was only 0.18%, while the RMSEP of the prediction set of the model without particle size correction was as high as 0.88%.

[0040] (3) Green and efficient, suitable for on-site quality control of production line

[0041] Rapid on-site testing eliminates the need for toxic organic solvents such as methanol and acetonitrile, as well as complex pretreatment processes like heating, reflux, filtration, and volume adjustment. The total time for single-sample spectral acquisition and model calculation is ≤60s. It can be equipped with online near-infrared devices to monitor saponin content fluctuations in the cooking, pulverizing, and packaging processes in real time, significantly reducing laboratory manpower, reagent, and time costs.

[0042] (4) The modeling process is standardized and easy to replicate and promote in the industry.

[0043] The spectral preprocessing range, principal component extraction quantity, and regression algorithm all have fixed parameters, eliminating the need for complex machine learning parameter tuning. This facilitates automated software computation and can be integrated into a spectral cloud computing system. Attached Figure Description

[0045] Figure 1 This is a flowchart of the near-infrared rapid detection modeling process for total saponins of Panax notoginseng and its products according to the present invention.

[0046] Figure 2 This is a flowchart of the near-infrared rapid detection process for total saponins of Panax notoginseng and its products according to the present invention;

[0047] Figure 3 This is a logic diagram for total saponins provided in an embodiment of the present invention;

[0048] Figure 4 This is a comparison table of the total saponin content detected by 80% methanol and pure methanol in the extraction of processed Panax notoginseng powder samples provided in this embodiment of the invention;

[0049] Figure 5 This is a schematic diagram illustrating the changes in starch granules before and after the ripening of Panax notoginseng starch, provided in an embodiment of the present invention.

[0050] Figure 6 These are near-infrared spectra of cooked Panax notoginseng powder of different particle sizes from the same batch, provided in an embodiment of the present invention.

[0051] Icons: a-Near-infrared spectrum of powder passing through a No. 3 sieve; b-Near-infrared spectrum of powder passing through a No. 3 sieve; c-Near-infrared spectrum of powder passing through a No. 3 sieve; d-Near-infrared spectrum of powder passing through a No. 3 sieve; e-Near-infrared spectrum of powder passing through a No. 3 sieve; f-Starch granules; g-Saponin components. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0054] Therefore, the following detailed description of the claims and embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0055] Example 1: Screening of extraction solvent concentrations to determine the optimal 80% methanol aqueous solution.

[0056] Experimental samples: Take 4 portions of cooked Panax notoginseng powder, each 0.6g, and weigh accurately;

[0057] Extraction solvents for each group: Group 1: pure methanol; Group 2: 70% methanol-water (v / v); Group 3: 80% methanol-water (v / v); Group 4: 90% methanol-water (v / v).

[0058] Extraction conditions: Add 50 mL of the corresponding solvent to each group, reflux in an 80℃ water bath for 2 h, cool and add weight, and determine the total saponins (sum of Panax notoginseng saponin R1, Ginsenoside Rg1, and Ginsenoside Rb1) by HPLC.

[0059] Test results: Total saponins in pure methanol group: 6.12%; 70% methanol-water group: 7.37%; 80% methanol-water group: 7.43%; 90% methanol-water group: 7.39%

[0060] Conclusion: The total saponin content measured in the pure methanol group was significantly lower than that in the other three groups. Specifically, the total saponin content measured in the 70%, 80%, and 90% methanol groups was basically consistent, indicating that methanol concentrations of 70%–90% can effectively extract saponins from the test samples. This study preferentially selected 80% methanol as the extraction solvent for preparing the test samples.

[0061] Example 2: Comparison of total saponins measured in samples using the optimized extraction solvent method and the original method.

[0062] Ten batches of processed Panax notoginseng powder were taken, and test samples were prepared using two different solvents. The total saponin content was determined by HPLC, and the differences between the two methods were compared. The results are shown in the figure. Figure 4 .

[0063] Conclusion: Comparing the total saponin results of 10 batches of processed Panax notoginseng prepared with two extraction solvents, the content of saponins in the 80% methanol-water extraction solvent group was significantly higher than that in the pure methanol extraction solvent group; the difference between the two was unstable, with a difference rate ranging from 8.3% to 22.7%, indicating the state of saponin encapsulation by gelatinized starch (see schematic diagram of starch gelatinization and saponin encapsulation during the cooking process). Figure 5 The presence of methanol as the solvent will directly and significantly affect the test results of the pure methanol extraction solvent group. Only by using 80% methanol-water as the extraction solvent can the true saponin quality of Panax notoginseng be reflected.

[0064] Example 3: Modeling and Validation with Granularity Variable Correction

[0065] Calibration set samples and spectral acquisition: 100 batches of cooked Panax notoginseng / cooked Panax notoginseng slices (each batch was divided into 5 particle sizes), totaling 500 calibration samples; all samples were acquired in the 900-2500 nm near-infrared spectrum;

[0066] True value calibration: The total saponin content was determined by reflux HPLC using 80% (v / v) methanol-water solution.

[0067] Spectral processing: The original spectrum was preprocessed in two stages: SNV and first derivative. The first 18 principal components were extracted from the 950-2250nm band, and multiple linear regression modeling was performed.

[0068] External validation set samples: 50 batches of processed Panax notoginseng powder; all samples were collected in the 900–2500 nm near-infrared spectrum;

[0069] Validation results: Model determination coefficient R² = 0.999, prediction root mean square error RMSEP = 0.18%;

[0070] Control experiment (without introducing granularity variables):

[0071] The calibration set samples that passed through sieve No. 4 and their corresponding total saponin contents were used for modeling, and the total saponin contents of 50 batches of processed Panax notoginseng powder were predicted. The results showed that the model's coefficient of determination R² = 0.980 and the root mean square error of prediction RMSEP = 0.88%.

[0072] Verification experiments demonstrate that granular variable correction can significantly improve prediction accuracy.

[0073] Example 4: Spectral Comparison of Cooked Panax notoginseng Powder of Different Particle Sizes from the Same Batch

[0074] Comparing the near-infrared spectra of the same batch of processed Panax notoginseng powder obtained by sieving with different sieve sizes, see... Figure 6 The samples corresponding to spectra a, b, c, d, and e in the figure are sieves No. 3, No. 4, No. 5, No. 6, and No. 7, respectively. It is easy to see that the spectral absorbance value decreases as the sieve number increases, indicating that particle size is an important factor affecting spectral quality.

[0075] Example 5: Modeling and Validation of Sequence Reversal Correction for Spectral Preprocessing Methods

[0076] For the spectra with particle size variable correction in Example 4, first-order differentiation was performed, followed by standard normal transformation preprocessing. Other steps were consistent with the particle size variable modeling in Example 4. The resulting model predicted the total saponin content of 50 batches of processed Panax notoginseng powder. The model determination coefficient R² = 0.991, and the root mean square error of prediction RMSEP = 0.59%. This indicates that the prediction ability of the calibrated model after spectral preprocessing is worse. The reason for this may be that the first derivative amplifies the baseline drift signal of the intraday and interday near-infrared spectra, and the subsequent standard normal transformation cannot effectively remove the baseline drift signal amplified by the first derivative.

[0077] Example 6: On-site rapid testing and practical verification

[0078] Sample to be tested: Commercially available dried Panax notoginseng slices, pulverized and passed through a No. 4 sieve, and mixed evenly;

[0079] Detection procedure: Powder is spread throughout the near-infrared sample cup, spectral scanning is performed for 40 seconds, and the system automatically completes SNV, first derivative preprocessing, wavelength selection, principal component dimensionality reduction, and inputs the results into the model for calculation;

[0080] The model output a total saponin content of 6.41%; simultaneous sampling and retesting using the HPLC calibration method of this invention yielded a measured value of 6.32%, with a relative error of 0.71%, which meets the error standard for the detection of traditional Chinese medicine content.

[0081] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A near-infrared quantitative modeling method for the total saponin content of processed Panax notoginseng and its products, characterized in that, Includes the following steps: S1. Preparation of Calibration Set Samples: More than 100 processed Panax notoginseng products from different origins, with different cooking processes and processing forms were selected as calibration set samples. These products include dried Panax notoginseng, dried Panax notoginseng slices, and dried Panax notoginseng powder. For dried Panax notoginseng and dried Panax notoginseng slices: each batch of samples was divided into 5 portions and pulverized separately, passing through sieves No. 3, No. 4, No. 5, No. 6, and No. 7 to obtain 5 different particle size powders. For dried Panax notoginseng powder: no secondary pulverization was required, and it was directly used as the particle size sample to be tested. All particle size powders were simultaneously included in the calibration set. S2, the true value of total saponins was determined by HPLC and used as the dependent variable in the modeling: Accurately weigh 0.6 g of each particle size powder, accurately add 50 mL of 70%–90% (v / v) methanol aqueous solution, weigh and record; reflux gently in an 80℃ water bath for 2 h, cool to room temperature, replenish the lost weight with the corresponding concentration of methanol aqueous solution, shake well, filter through a filter membrane, and use the filtrate as the test sample; Liquid chromatography conditions: use an octadecylsilane-bonded silica gel C18 column; mobile phase A is acetonitrile, mobile phase B is purified water; gradient elution program: 0–20 min, phase A 20%; From 20 to 45 min, phase A linearly increased from 20% to 46%; from 45 to 55 min, phase A linearly increased from 46% to 55%; from 55 to 60 min, phase A remained at 55%; detection wavelength was 203 nm; the theoretical plate number calculated based on the chromatographic peak of Panax notoginseng saponin R1 was not less than 4000. Preparation of reference solution: Dissolve in methanol to prepare a mixed reference solution, each 1 mL containing 0.4 mg of ginsenoside Rg1, 0.4 mg of ginsenoside Rb1, and 0.1 mg of notoginsenoside R1; Accurately inject 10 μL each of the reference solution and the test solution into the liquid chromatograph, and calculate the total saponin content using the external standard method based on peak area. S3. Near-infrared raw spectrum acquisition: Powders of various particle sizes after S1 treatment are spread evenly on a near-infrared sample cup, and raw diffuse reflectance spectra of the entire wavelength range of 900-2500nm are acquired; S4. Two-stage spectral preprocessing: The first step is to perform standard normal transformation (SNV) on all raw spectra to eliminate intraday and interday instrument baseline fluctuation errors; the second step is to perform first derivative processing on the SNV-corrected spectra to amplify the characteristic absorption signals of saponins; both stages of processing are performed using full-band synchronous operation without segmentation. S5. Principal component dimensionality reduction: Extract all wavelength data in the characteristic band of the preprocessed spectrum from 950 to 2250 nm for principal component analysis, extract the first 18 principal components as independent variables of the model, and store the principal component calculation coefficients simultaneously. S6. Quantitative Model Construction: The 18 principal component independent variables obtained in S5 were fitted with the true values ​​of total saponins in the corresponding samples in S2 using multiple linear regression to obtain a near-infrared quantitative detection model for total saponins of Panax notoginseng and its products.

2. The near-infrared quantitative modeling method for the total saponin content of processed Panax notoginseng and its products according to claim 1, characterized in that, The solvent used in step S2 is an 80% (v / v) methanol aqueous solution.

3. The near-infrared quantitative modeling method for the total saponin content of processed Panax notoginseng and its products according to claim 1, characterized in that, In step S1, all five particle sizes of powders prepared from the same batch of samples were included in the calibration set for modeling. Powder particle size was used as an interference variable in the regression calculation to eliminate the spectral prediction error caused by the difference in particle size of different powders.

4. The near-infrared quantitative modeling method for the total saponin content of processed Panax notoginseng and its products according to claim 1, characterized in that, Spectral preprocessing must be performed strictly in the order of standard normal transformation followed by first-order differentiation; the order cannot be changed.

5. A rapid method for detecting total saponins in processed Panax notoginseng and its products, characterized in that, Includes the following steps: T1. Sample pretreatment: Pulverize and sieve cooked Panax notoginseng slices, or mix them thoroughly; or mix cooked Panax notoginseng powder directly for later use to obtain the powder to be tested. T2. Near-infrared spectral acquisition: The powder to be tested is evenly spread on the sample cup, and the original diffuse reflectance spectrum of 900-2500 nm is acquired. T3. Spectral preprocessing: Perform standard normal transformation and first derivative processing on the original spectrum in sequence; T4. Characteristic band dimensionality reduction calculation: Extract the 950-2250nm band data from the preprocessed spectrum, and calculate the first 18 principal components using the principal component conversion coefficients stored in claim 1. T5. Content Quantitative Calculation: Substitute the 18 principal components into the multiple linear regression model constructed in claim 1 to directly output the detection results of the total saponin content of the sample.

6. The rapid detection method for total saponins of processed Panax notoginseng and its products according to claim 5, characterized in that, The entire detection process requires no organic solvents, no heating reflux, filtration or other complex pretreatments, and the total time for spectral acquisition and model calculation of a single sample is ≤60s.

7. The rapid detection method for total saponins of processed Panax notoginseng and its products according to claim 6, characterized in that, This testing method can be applied to on-site quality monitoring of the entire production line for processed Panax notoginseng, processed Panax notoginseng slices, crushing, powdering, and finished product packaging.

Images