WO2021056814A1 - 一种基于药效信息建立评价中药质量的化学模式识别方法 - Google Patents

一种基于药效信息建立评价中药质量的化学模式识别方法 Download PDF

Info

Publication number
WO2021056814A1
WO2021056814A1 PCT/CN2019/122425 CN2019122425W WO2021056814A1 WO 2021056814 A1 WO2021056814 A1 WO 2021056814A1 CN 2019122425 W CN2019122425 W CN 2019122425W WO 2021056814 A1 WO2021056814 A1 WO 2021056814A1
Authority
WO
WIPO (PCT)
Prior art keywords
samples
chemical
pattern recognition
traditional chinese
chinese medicine
Prior art date
Application number
PCT/CN2019/122425
Other languages
English (en)
French (fr)
Inventor
王铁杰
鲁艺
王丽君
江坤
王洋
王珏
殷果
黄洋
金一宝
Original Assignee
深圳市药品检验研究院（深圳市医疗器械检测中心）
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳市药品检验研究院（深圳市医疗器械检测中心） filed Critical 深圳市药品检验研究院（深圳市医疗器械检测中心）
Priority to US17/779,835 priority Critical patent/US11710541B2/en
Priority to EP19946314.2A priority patent/EP3907493A4/en
Publication of WO2021056814A1 publication Critical patent/WO2021056814A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
    • G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
    • G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
    • G01N21/33—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using ultraviolet light
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
    • G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
    • G16C20/20—Identification of molecular entities, parts thereof or of chemical compositions
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
    • G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
    • G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
    • G01N21/35—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light
    • G01N21/3563—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light for analysing solids; Preparation of samples therefor
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/62—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light
    • G01N21/63—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light optically excited
    • G01N21/64—Fluorescence; Phosphorescence
    • G01N21/6428—Measuring fluorescence of fluorescent products of reactions or of fluorochrome labelled reactive substances, e.g. measuring quenching effects, using measuring "optrodes"
    • G01N21/643—Measuring fluorescence of fluorescent products of reactions or of fluorochrome labelled reactive substances, e.g. measuring quenching effects, using measuring "optrodes" non-biological material
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N24/00—Investigating or analyzing materials by the use of nuclear magnetic resonance, electron paramagnetic resonance or other spin effects
    • G01N24/08—Investigating or analyzing materials by the use of nuclear magnetic resonance, electron paramagnetic resonance or other spin effects by using nuclear magnetic resonance
    • G01N24/088—Assessment or manipulation of a chemical or biochemical reaction, e.g. verification whether a chemical reaction occurred or whether a ligand binds to a receptor in drug screening or assessing reaction kinetics
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N27/00—Investigating or analysing materials by the use of electric, electrochemical, or magnetic means
    • G01N27/62—Investigating or analysing materials by the use of electric, electrochemical, or magnetic means by investigating the ionisation of gases, e.g. aerosols; by investigating electric discharges, e.g. emission of cathode
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N30/00—Investigating or analysing materials by separation into components using adsorption, absorption or similar phenomena or using ion-exchange, e.g. chromatography or field flow fractionation
    • G01N30/02—Column chromatography
    • G01N30/86—Signal analysis
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N30/00—Investigating or analysing materials by separation into components using adsorption, absorption or similar phenomena or using ion-exchange, e.g. chromatography or field flow fractionation
    • G01N30/02—Column chromatography
    • G01N30/86—Signal analysis
    • G01N30/8624—Detection of slopes or peaks; baseline correction
    • G01N30/8631—Peaks
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
    • G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
    • G16C20/30—Prediction of properties of chemical compounds, compositions or mixtures
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
    • G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
    • G16C20/70—Machine learning, data mining or chemometrics
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/62—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light
    • G01N21/63—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light optically excited
    • G01N21/64—Fluorescence; Phosphorescence
    • G01N2021/6417—Spectrofluorimetric devices
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N30/00—Investigating or analysing materials by separation into components using adsorption, absorption or similar phenomena or using ion-exchange, e.g. chromatography or field flow fractionation
    • G01N30/02—Column chromatography
    • G01N2030/022—Column chromatography characterised by the kind of separation mechanism
    • G01N2030/027—Liquid chromatography
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
    • G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
    • G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
    • G01N21/35—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light
    • G01N21/359—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light using near infrared light
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/62—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light
    • G01N21/63—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light optically excited
    • G01N21/65—Raman scattering
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N2201/00—Features of devices classified in G01N21/00
    • G01N2201/12—Circuits of general importance; Signal processing
    • G01N2201/129—Using chemometrical methods
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N24/00—Investigating or analyzing materials by the use of nuclear magnetic resonance, electron paramagnetic resonance or other spin effects
    • G01N24/08—Investigating or analyzing materials by the use of nuclear magnetic resonance, electron paramagnetic resonance or other spin effects by using nuclear magnetic resonance

Definitions

  • the invention belongs to the field of quality evaluation of traditional Chinese medicines, and relates to a chemical pattern recognition method for establishing and evaluating the quality of traditional Chinese medicines based on pharmacodynamic information.
  • Traditional Chinese medicine itself is a complex and huge mixed system, which has the characteristics of multi-component, multi-target, and multi-channel action, which increases the difficulty of quality evaluation of Chinese medicine to a certain extent.
  • the quality evaluation of traditional Chinese medicine at home and abroad is mainly to detect several chemical components in traditional Chinese medicine, and the established evaluation methods do not rely on the efficacy of the medicine.
  • the lack of an overall and comprehensive and reliable quality evaluation system for traditional Chinese medicine not only increases the health risks of users, but also affects the international reputation, competitiveness and influence of traditional Chinese medicine.
  • CN108509997A discloses a method for chemical pattern recognition of the authenticity of traditional Chinese medicine saponin thorn based on near-infrared spectroscopy technology.
  • the method uses near-infrared spectroscopy acquisition method, first-order derivative preprocessing method, continuous projection algorithm, and Kennard-Stone algorithm And the combination of step algorithm to chemical pattern recognition of the authenticity of the traditional Chinese medicine saponaria thorn, so that the result of the pattern recognition method is accurate and reliable, and can accurately distinguish the saponaria thorn and its fakes.
  • this method is only based on chemical information collection and chemical processing methods to obtain characteristic wavenumber points.
  • the characteristic wavenumber points are not necessarily related to the pharmacodynamics of the drug.
  • the extra irrelevant characteristic wavenumber points make the discrimination model more complicated.
  • the purpose of the present invention is to provide a chemical pattern recognition method for establishing and evaluating the quality of traditional Chinese medicines based on the efficacy information.
  • the method of the present invention does not require a chemical reference substance, can comprehensively reflect the chemical information of traditional Chinese medicines, uses pharmacodynamics information to establish a basis for chemical pattern recognition models, so that the discrimination model is more accurate, and the present invention overcomes the subjectivity of identification, and the identification results are accurate reliable.
  • the present invention provides a chemical pattern recognition method for establishing and evaluating the quality of traditional Chinese medicine based on pharmacodynamic information.
  • the method includes the following steps:
  • step (2) Divide the traditional Chinese medicine samples into training set and test set samples, and use the supervised pattern recognition method to take the characteristic chemical indicators described in step (1) as input variables, and extract the characteristic variables of the training set samples;
  • step (3) Use the feature variables extracted in step (2) to establish a pattern recognition model
  • the pattern recognition model is established.
  • These characteristic variables are significantly related to the drug effect, avoiding the interference of irrelevant variables and causing
  • the complexity of the pattern recognition model can obtain a more accurate pattern recognition model, and perform simpler and more direct identification of authenticity of traditional Chinese medicines, quality grading of traditional Chinese medicines, etc., and the results are accurate and reliable; and the method of the present invention is helpful for fine-grained traditional Chinese medicines. Looking for alternatives.
  • the traditional Chinese medicine includes Huajuhong, Salvia miltiorrhiza, Saponaria saponaria, Amomum villosum, Kungluomu or Panax notoginseng.
  • the collection of chemical information of the traditional Chinese medicine sample refers to the identification of the target of the traditional Chinese medicine to take its chemical characteristic information. For example, if the goal is to identify the authenticity of traditional Chinese medicines, then it is to collect the overall chemical information that can reflect the inherent quality of traditional Chinese medicines and their fake samples. If the goal is to classify the quality of traditional Chinese medicines, then it is to collect all grades of traditional Chinese medicines that can reflect the quality levels. The overall chemical information of the internal quality of the sample.
  • the acquisition of pharmacodynamic information reflecting the clinical efficacy of traditional Chinese medicine is carried out by conventional means in the research of traditional Chinese medicine pharmacodynamics.
  • the collected data is converted into an m ⁇ n matrix, where n is the number of traditional Chinese medicine samples, and m is each traditional Chinese medicine.
  • n is the number of traditional Chinese medicine samples
  • m is each traditional Chinese medicine.
  • the method for collecting chemical information of a traditional Chinese medicine sample is a spectral collection method, a chromatographic collection method, a mass spectrometry collection method, or a nuclear magnetic method;
  • the spectrum collection method is any one of ultraviolet spectrum, infrared spectrum, near-infrared spectrum, Raman spectrum or fluorescence spectrum;
  • the chromatographic collection method is high performance liquid chromatography or ultra high performance liquid chromatography.
  • the collection of chemical information is the collection of characteristic chemical signals that can reflect the inherent quality of traditional Chinese medicine. For example, if you use ultraviolet spectroscopy to collect, you are collecting the ultraviolet characteristic absorption peak of traditional Chinese medicine. If you use high-performance liquid to collect, you are collecting. All the Chinese medicines have obvious peaks in the high-performance liquid phase.
  • the correlation analysis of the drug effect on chemical information refers to the correlation analysis between the collected chemical information and the drug effect, the selection of the chemical information significantly related to the drug effect as the drug effect index, and the elimination of the chemical information related to the drug effect. Effectively irrelevant chemical information.
  • the method of analyzing the spectral effect relationship in step (1) may be a bivariate correlation analysis method, a regression analysis method, a gray level correlation analysis method, a partial least square method or a principal component analysis method.
  • the supervised pattern recognition method described in step (2) is a principal component discriminant analysis method, a stepwise discriminant analysis method, a partial least squares discriminant method, a support vector machine or an artificial neural network algorithm.
  • step (2) when extracting the characteristic variables in step (2), remove k irrelevant chemical information to obtain a matrix of (mk) ⁇ n, where n is the number of traditional Chinese medicine samples, and m is the chemical information collected for each traditional Chinese medicine sample quantity.
  • Figure 1 the process of establishing a chemical pattern recognition method for evaluating the quality of traditional Chinese medicine based on drug efficacy information is shown in Figure 1, which reflects the overall process of the method as a whole, and the model is completed under the guidance of drug efficacy (ie, pharmacological activity) Identification to be able to evaluate the quality of traditional Chinese medicine and predict and analyze unknown samples.
  • drug efficacy ie, pharmacological activity
  • the method for establishing a chemical pattern recognition evaluation method for the quality of traditional Chinese medicine based on the efficacy information includes chemical pattern recognition of the authenticity of the traditional Chinese medicine danshen, chemical pattern distinction between the orange red, the hair orange red and the light orange red, or the authenticity of the saponaria thorn Perform chemical pattern recognition.
  • the method for chemical pattern recognition of the authenticity of the traditional Chinese medicine Salvia miltiorrhiza, or chemical pattern distinction between the red orange, the hair orange red and the light orange red includes:
  • HPLC fingerprint data is used as a characteristic chemical reflecting the pharmacodynamics. index;
  • stepwise discriminant analysis method uses the characteristic chemical indicators described in step A as input variables to screen the characteristic chemistry of the training set samples to remove Related variables, filter out characteristic variables;
  • step B Use the characteristic variables obtained in step B to establish a pattern recognition model for Salvia miltiorrhiza and its counterfeit products, or establish a pattern recognition model for orange-red samples;
  • the selection principle of the specific absorption peaks of Salvia miltiorrhiza and its counterfeit products in step A is a peak that meets at least one of the following conditions: (1) a peak shared by Salvia miltiorrhiza, Sage ganxiensis and Sage yunnanensis; (II) The unique peaks of Salvia miltiorrhiza, Ganxi sage, and Yunnan sage; (III) The peaks with higher component content.
  • the selection principle of the specific absorption peaks of hair orange and light orange in step A is the peak shared by both hair orange and light orange.
  • the selected specific absorption peaks represent the main chemical information of the three medicinal materials: Salvia miltiorrhiza, Sage Ganxi and Sage Yunnan.
  • the method of randomly dividing into training set and test set in step B is to use a random algorithm to perform random division.
  • the training set of Salvia miltiorrhiza and its fakes in step B includes 20 batches of samples, including 12 batches of Salvia miltiorrhiza, 4 batches of Ganxi sage and 4 batches of Yunnan sage; the test set includes 29 batches of samples, Including 26 batches of Salvia miltiorrhiza, 2 batches of Ganxi sage and 1 batch of Yunnan sage.
  • the division into training set and test set is randomly divided, so it is not limited to the training set and test set including a specific number of batches of samples described herein.
  • the training set of orange red samples in step B includes 22 batches of samples, including 10 batches of hair orange samples and 12 batches of light orange samples, and the test set includes 9 batches of samples, including 5 batches of hair orange samples and 4 batches Light orange sample.
  • the characteristic variables screened out in step B are X 6 , X 7 and X 13 ; that is to say, although a lot of HPLC fingerprint data that is significantly correlated with the pharmacodynamic activity are obtained under the method of this application, it is through discriminant analysis There are only 3 characteristic variables related to classification obtained by stepwise screening, which greatly simplifies the model function.
  • the function of the pattern recognition model described in step C is as follows:
  • the characteristic variables selected for the orange-red sample in step B are X 7 , X 8 and X 20 .
  • the function of the pattern recognition model for the orange-red sample established in step C is as follows:
  • the method for chemical pattern recognition of the authenticity of saponins spines of the present invention includes the following steps:
  • step II Divide the saponaria spinosa and its fakes into training set and test set samples randomly, and use stepwise discriminant analysis method to use the characteristic chemical indicators described in step I as input variables to screen the characteristic chemical indicators of the training set samples to remove Related variables, filter out characteristic variables;
  • step III Use the characteristic variables obtained in step II to establish a pattern recognition model
  • step I after collecting chemical information of saponins and its fakes by using the near-infrared spectroscopy in step I, it further includes performing spectral data preprocessing on the chemical information: removing interference peaks and water peaks in the original spectrum to obtain 11800-7500cm ⁇ 1, 6500 ⁇ 5500cm -1 and 4200cm -1 spectral range 5000 to peak, select 5000 ⁇ 4200cm -1 as a peak spectral peak analysis model, and using the first derivative pretreatment spectral peak of 5000 to pre--1 4200cm Processing, using continuous projection algorithm (SPA) for feature peak extraction.
  • SPA continuous projection algorithm
  • the interference peak is 12000 ⁇ 11800cm -1, 4200 ⁇ 4000cm -1, 7500 6500cm -1 and a peak spectral ranges 5500 ⁇ ⁇ 5000cm -1; the water peak is 7500 ⁇ 6500cm -1, 5500 ⁇ 5000cm -1 peak of the spectrum.
  • the training set in step II includes 32 batches of samples, including 24 batches of saponins, 3 batches of saponins, 2 batches of wild saponins, and 3 batches of rubus, and the test set samples include 11 batches of samples , Including 8 batches of saponaria saponaria, 1 batch of saponaria saponaria, 1 batch of saponaria saponaria and 1 batch of raspberry.
  • the characteristic variables selected in step II are X 8 , X 10 , X 14 and X 21 ;
  • the function of the pattern recognition model in step III is:
  • the present invention has the following beneficial effects:
  • the identification method of the present invention does not require a chemical reference substance, can comprehensively reflect the chemical information of traditional Chinese medicine, uses pharmacodynamics information to establish a basis for chemical pattern recognition models, so that the relationship between the discriminant model and the drug effect is closer, and the obtained chemical pattern recognition model
  • the function is relatively simple, and it can ensure the accuracy of identification. It overcomes the one-sidedness and subjectivity of the current standards in evaluating the quality of traditional Chinese medicine with only one or a few ingredients, and finally forms a method that can reflect the clinical efficacy and take into account chemical components. Informational quality evaluation system of traditional Chinese medicine, the identification results are accurate and reliable.
  • the method of the present invention can be used for simpler and more direct identification of authenticity of traditional Chinese medicines, quality classification of traditional Chinese medicines, etc., and the results are accurate and reliable; and the method of the present invention is helpful to find substitutes for fine and expensive traditional Chinese medicines, and the method can realize the classification of unknown samples. Based on the prediction of the method of the present invention, a quality evaluation system of traditional Chinese medicine is formed.
  • Figure 1 is the overall flow chart of establishing a chemical pattern recognition method for evaluating the quality of traditional Chinese medicine based on the efficacy information
  • Figure 2 shows the high performance liquid chromatograms collected from Salvia miltiorrhiza, Ganxi sage and Yunnan sage, where S1 is the high performance liquid chromatogram collected from DS 3 Salvia samples and S2 is the GX 39 Salvia samples.
  • S1 is the high performance liquid chromatogram collected from DS 3 Salvia samples
  • S2 is the GX 39 Salvia samples.
  • S3 is the high-performance liquid chromatography collected from YN 45 Sage;
  • Figure 3 is a distribution diagram of the training set samples of Salvia miltiorrhiza and its fakes with the discriminant function values (that is, the values of F 1 and F 2 , that is, function 1 and function 2) as the horizontal and vertical coordinates;
  • Figure 4 is a distribution diagram of the training set and test set samples of Salvia miltiorrhiza and its fakes, with the discriminant function value (that is, the value of F 1 and F 2 , that is, function 1 and function 2) as the horizontal and vertical coordinates;
  • Figure 5 is a high-performance liquid chromatogram of Mao Juhong sample
  • Figure 6 is a high performance liquid chromatogram of a light orange sample
  • Fig. 7 is a distribution diagram of the samples of the orange-red training set with the sample number and the discriminant function value (that is, the value of F 1 , that is, the score value of function 1) as the horizontal and vertical coordinates;
  • Figure 8 is the distribution diagram of the orange-red training set and test set samples with sample serial numbers and discriminant function values (that is, the value of F 1 , that is, the score value of function 1) as the horizontal and vertical coordinates;
  • Fig. 9 is an original average near-infrared spectrum obtained by infrared spectroscopy collection of samples of saponaria saponaria and its counterfeit products according to the present invention.
  • Figure 10 is a near-infrared spectrogram obtained by preprocessing the original average near-infrared spectrum using the first derivative (First Derivative, 1st D) method;
  • Figure 11 is the distribution diagram of the training set samples of the saponaria spinosa and its fakes with the discriminant function values (that is, the values of F 1 and F 2 , that is, function 1 and function 2) as the horizontal and vertical coordinates;
  • Fig. 12 The training set and test set samples of saponin spines and its fakes are distributed with the value of the discriminant function (that is, the value of F 1 and F 2 , that is, function 1 and function 2) as the abscissa and ordinate.
  • the discriminant function that is, the value of F 1 and F 2 , that is, function 1 and function 2
  • the overall process of establishing a chemical pattern recognition method for evaluating the quality of traditional Chinese medicine based on drug efficacy information is shown in Figure 1.
  • Typical representative traditional Chinese medicines are collected, and overall chemical information that can reflect the inherent quality of traditional Chinese medicine samples is collected.
  • Obtain pharmacodynamic information that can reflect the clinical efficacy of traditional Chinese medicine extract the characteristics of chemical information under the guidance of pharmacodynamics, and obtain characteristic chemical indicators that can reflect the pharmacodynamics. Both the pharmacodynamics correlation analysis of the chemical information and the pharmacodynamics is obtained.
  • the chemical information indicators that are significantly related to the drug efficacy are used as the feature indicators; the traditional Chinese medicine samples are divided into training sets and test sets; the supervised pattern recognition method is used to reflect the clinical efficacy of the chemical indicators as input variables, and the training set samples Characteristic variable extraction; use the extracted characteristic variables to establish a pattern recognition model; substitute the characteristic variable values of the test set samples into the pattern recognition model, so as to complete the chemical pattern recognition of the quality of the traditional Chinese medicine sample under the guidance of the drug effect (ie pharmacological activity) Evaluation.
  • High performance liquid chromatography Chromatographic column: Zobax SB-aq (250mm ⁇ 4.6mm, 5 ⁇ m, Agilent); mobile phase: acetonitrile (A)-0.03% phosphoric acid solution (B), gradient elution, the elution program is shown in Table 1 ; Detection wavelength: 280nm, flow rate: 0.8mL ⁇ min-1, column temperature: 30°C, injection volume: 20 ⁇ L.
  • the operation of the random algorithm uses SPSS software (IBM Corporation, USA).
  • the samples used are as follows:
  • Salvia samples No. 1 to 38 are Salvia.
  • the dried roots and rhizomes of miltiorrhiza, the dry roots and rhizomes of the salvia sage samples No. 39-43 are the dried roots and rhizomes of the Salvia gravwalskii, and the sage samples Nos. 44-49 are the dried salvia yunnanensis
  • Table 2 the sources of all samples are shown in Table 2.
  • the method for chemical pattern recognition of the authenticity of traditional Chinese medicine danshen specifically includes the following steps:
  • the selected chromatographic peaks are shown in Figure 2, where S1 is the high performance liquid chromatogram collected from the DS 3 Salvia miltiorrhiza sample, S2 is the high performance liquid chromatogram collected from the GX 39 Ganxi sage sample, and S3 is the YN 45 sage
  • S1 is the high performance liquid chromatogram collected from the DS 3 Salvia miltiorrhiza sample
  • S2 is the high performance liquid chromatogram collected from the GX 39 Ganxi sage sample
  • S3 is the YN 45 sage
  • the high-performance liquid chromatograms collected from the samples, and the high-performance liquid chromatograms of the three samples are all marked with the corresponding peak numbers selected.
  • HPLC fingerprint data that are significantly correlated with the potency and activity of Salvia miltiorrhiza, Ganxi sage and Yunnan sage are A6, A7, A8, A10, A13, A14, A17, A18, A19, A20, and A21.
  • Training set samples DS2, DS3, DS4, DS6, DS7, DS13, DS15, DS16, DS18, DS20, DS22, DS35, GX39, GX42, GX43, GX44, YN46, YN47, YN48, YN49);
  • Test set samples DS1, DS5, DS8, DS9, DS10, DS11, DS12, DS14, DS17, DS19, DS21, DS23, DS24, DS25, DS26, DS27, DS28, DS29, DS30, DS31, DS32, DS33, DS34, DS36, DS37, DS38, GX40, GX41, YN45.
  • stepwise discriminant analysis the variables (A6, A7, A8, A10, A10, A13, A14, A17, A18, A19, A20, and A21) that are significantly related to the efficacy of the spectrum effect correlation analysis Screening is used to feature extraction variables, and the screening is carried out step by step through F-test. Each step selects the most significant variables that meet the specified level and eliminates the originally introduced variables that have become insignificant due to the introduction of new variables, until it can neither be introduced nor eliminated.
  • stepwise discriminant analysis the salvia miltiorrhiza, Ganxi sage and Yunnan sage were compared at the same time, and the representative peak variables of the characteristics were screened out. The dimensionality reduction results (that is, the selected characteristic variables) are shown in Table 7.
  • F1 and F2 are the horizontal and vertical coordinates of the sample in the distribution diagram, respectively, and the distribution diagram results are shown in Figure 3 (training set) and Figure 4 (training set and test set).
  • Figure 3 and Figure 4 Salvia miltiorrhiza (DS), Ganxi sage (GX) and Yunnan sage (YN) in the training set and test set samples can be effectively distinguished.
  • the method described above is used for feature extraction by stepwise discriminant analysis under the guidance of drug efficacy, obtaining 3 feature values, and establishing 2 discriminant functions, which can effectively distinguish Salvia miltiorrhiza, Ganxi sage and Yunnan sage.
  • Elution gradient using binary gradient elution system, solvent A (methanol)-solvent B (0.5% glacial acetic acid) detection wavelength: 320nm, flow rate: 1.0mL ⁇ min -1 , column temperature: 30°C, injection volume: 20 ⁇ L.
  • the samples used are as follows:
  • the specific method for pattern recognition of Huajuhong medicinal materials includes the following steps:
  • the 31 batches of medicinal materials were analyzed by HPLC chromatographic analysis, and all chromatographic peak data were obtained.
  • the results of the hair orange sample are shown in Figure 5, and the results of the light orange sample are shown in Figure 6.
  • a random algorithm is used to divide 31 batches of orange-red samples into training set and test set:
  • Training set samples 2, 3, 4, 7, 8, 10, 11, 13, 14, 15, 18, 20, 21, 23, 24, 26, 28, 29, 30, 31, 32, 33.
  • Test set samples 1, 5, 9, 12, 16, 17, 22, 25, 27.
  • stepwise discriminant analysis is used to screen out the peaks that contribute to the classification.
  • the method of stepwise discriminant analysis is adopted, Wilks' Lambda is used as the evaluation index, and the main peak is selected with the same probability within 0.05, and the peak is retained. The same probability is greater than 0.1 as the indifference peak. The peak is eliminated and the orange classification is judged.
  • Table 16 shows the result of feature extraction of discriminative variables step by step.
  • the characteristic variables for the classification of orange are X 7 , X 8 and X 20 .
  • the training set samples are used as the data set, and the characteristic variables X 7 , X 8 , and X 20 selected in the stepwise discriminant analysis are used as input variables, as shown in Table 17.
  • the characteristic variables X 7 , X 8 , and X 20 selected in the stepwise discriminant analysis are used as input variables, as shown in Table 17.
  • F 1 >0 is hairy orange
  • F 1 ⁇ 0 is light orange
  • the discriminant function value F 1 and the sample serial number are the vertical and horizontal coordinates of the sample in the distribution diagram.
  • the distribution diagram results are shown in Figure 7 (training set) and Figure 8 (training set and test set). In Figure 7 and Figure 8, the hair orange and light orange in the training set and test set samples can be effectively distinguished.
  • the method described above performs feature extraction by stepwise discriminant analysis under the guidance of drug efficacy, obtains 3 feature values, and establishes a discriminant function, which can effectively distinguish hair orange and light orange.
  • the near-infrared spectrum of the sample is collected with a fiber optic probe, the collection range is 12000 ⁇ 4000cm-1, the resolution of the instrument is 4cm -1 , and the number of scans is 32 times.
  • each batch of samples collects spectra from 3 different positions, and calculates the average spectra as the representative spectra.
  • Use OPUS 6.5 workstation German Bruker company to find the average spectrum.
  • the experimental temperature is balanced at 25°C, and the humidity is maintained at about 60%.
  • the original average near-infrared spectra of the saponaria spinosa and its fakes are shown in Figure 9.
  • the SPA algorithm is used to extract the characteristic wave numbers in the three spectral intervals under different pre-processing conditions. Using Matlab R2014a software to run SPA algorithm, after extracting feature variables, the complexity of modeling is greatly reduced.
  • LPS lipopolysaccharide
  • macrophages After being stimulated by lipopolysaccharide (LPS), macrophages can activate cell surface receptors to initiate a variety of signal cascade amplification effects, leading to the production of nitric oxide (NO), TNF- ⁇ , IL-6 and other pro-inflammatory factors , Causing damage. Measuring the level of NO in the cell supernatant can detect the level of inflammation.
  • NO nitric oxide
  • TNF- ⁇ TNF- ⁇
  • IL-6 pro-inflammatory factor-6
  • Causing damage Measuring the level of NO in the cell supernatant can detect the level of inflammation.
  • Nitrite can react with p-aminobenzenesulfonic acid and ⁇ -naphthylamine in Griess Reagent under acidic conditions to form a red couple
  • Nitrogen compounds have a maximum absorption peak at 540nm, and the product concentration has a linear relationship with the NO concentration. Therefore, this principle can be used to determine the NO content in the cell culture supernatant.
  • the specific measurement steps are as follows:
  • step II Take 50 ⁇ L/well Griess reagent and place it in a 96-well plate, add 50 ⁇ L/well of the supernatant or sodium nitrite standard solution of different concentrations in step I. After reacting for 30min at room temperature, remove the air bubbles in the well, under 540nm Determine the absorbance value;
  • ORAC method uses sodium fluorescein (sodium flourescein, FL) as a fluorescent probe to observe its interaction with the azo compound 2,2'-azo-bis-(2-amidinopropane) dihydrogen chloride [2,2'- azobis(2-amidinopropane)dihydrochloride, AAPH]
  • the fluorescence intensity declines (in the presence of antioxidants, its fluorescence declines slowly) and the antioxidant standard substance-water solubility is used
  • Vitamin E analog (6-hydro-2,5,7,8-tetramethylchroman-2-carboxylic acid, Trolox) equivalent is used to measure the ability of various antioxidants in the system to delay the decline of the fluorescence intensity of the probe to evaluate antioxidants The antioxidant capacity.
  • the NO inhibitory activity and ORAC antioxidant activity of the samples are shown in Table 22.
  • the training set samples include 32 batches of samples, including 24 batches of Saponaria saponaria, 3 batches of Saponaria saponaria, 2 batches of Saponaria saponaria and 3 batches of raspberry.
  • the test set samples include 11 batches of samples. Including 8 batches of saponins, 1 batch of saponins, 1 batch of wild saponins, and 1 batch of rubus.
  • Training set samples 2, 5, 6, 7, 8, 9, 10, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 28, 30, 34, 35, 36, 38, 39,, 41, 42, 43;
  • Test set samples 1, 3, 4, 11, 27, 29, 31, 32, 33, 37, 40.
  • the variables (X1, X7, X8, X9, X10, X12, X13, X14, X20, X21, X22, X23, X24, X25, X26, X27, X28) perform variable screening to feature extraction of variables.
  • the screening is carried out step by step through F-test. Each step selects the most significant variables that meet the specified level and eliminates the originally introduced variables that have become insignificant due to the introduction of new variables, until it can neither be introduced nor eliminated.
  • stepwise discriminant analysis the four types of Saponaria saponaria, Saponaria saponaria, and Saponaria saponaria were simultaneously compared, and representative peak variables of characteristics were selected. The dimensionality reduction results are shown in Table 24.
  • the discriminant function value Based on the discriminant function value, make a distribution diagram of the training set and test set samples. Taking F1 and F2 as the horizontal and vertical coordinates of the sample in the distribution diagram, the distribution diagram results are shown in Figure 11 (training set) and Figure 12 (training set and test set). In Fig. 11 and Fig. 12, the samples of the training set and the test set can be effectively distinguished from the saponaria japonicus (ZJC), the saponilla japonica (SZJ), the wild japonicus (YZJ) and the rubus (XGZ).
  • ZJC saponaria japonicus
  • SZJ saponilla japonica
  • YZJ wild japonicus
  • XGZ rubus
  • the method described above is used to extract features by stepwise discriminant analysis under the guidance of drug efficacy, obtain 4 eigenvalues, and establish 2 discriminant functions, which can perform saponaria saponaria, saponaria saponaria, wild saponaria and raspberry spurs. Effective distinction.
  • the present invention uses the above-mentioned embodiments to illustrate the method of the present invention, but the present invention is not limited to the above-mentioned process steps, which does not mean that the present invention must rely on the above-mentioned process steps to be implemented.
  • Those skilled in the art should understand that any improvement to the present invention, the equivalent replacement of the raw materials selected in the present invention, the addition of auxiliary components, the selection of specific methods, etc. fall within the scope of protection and disclosure of the present invention.

Landscapes

  • Physics & Mathematics (AREA)
  • Chemical & Material Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Immunology (AREA)
  • Pathology (AREA)
  • Analytical Chemistry (AREA)
  • General Physics & Mathematics (AREA)
  • Biochemistry (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Crystallography & Structural Chemistry (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computing Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Evolutionary Computation (AREA)
  • Databases & Information Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Artificial Intelligence (AREA)
  • Chemical Kinetics & Catalysis (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • High Energy & Nuclear Physics (AREA)
  • Electrochemistry (AREA)
  • Molecular Biology (AREA)
  • Optics & Photonics (AREA)
  • Medicinal Chemistry (AREA)
  • Medicines Containing Plant Substances (AREA)
  • Investigating Or Analysing Biological Materials (AREA)

Abstract

一种基于药效信息建立评价中药质量的化学模式识别方法，包括：采集中药样品的化学信息，获取反映其临床疗效的药效信息，对化学信息和药效信息进行谱效关系分析，得到与药效显著相关的指标作为特征化学指标；将中药样品划分为训练集和测试集；采用模式识别方法以特征化学指标为输入变量，对训练集样本提取特征变量；利用特征变量，建立模式识别模型；将测试集样本的特征变量值代入模型中，完成中药质量的化学模式识别评价。这一方法无需化学对照品，以反映药效的特征化学指标为基础建立化学模式识别模型，克服现行标准的片面性和主观性，最终形成既能反映临床疗效又能反映整体化学成分信息的中药质量评价体系。

Description

一种基于药效信息建立评价中药质量的化学模式识别方法 技术领域
本发明属于中药质量评价领域，涉及一种基于药效信息建立评价中药质量的化学模式识别方法。
背景技术
我国是世界上中药资源最丰富的国家，占据了近70％国际市场。随着经济全球化以及中药在临床应用中的杰出表现，使得中药获得了很大的发展机遇。也涌现出一些问题：对诸多药用价值及经济价值兼具的中药，存在以次充好及掺伪现象；中药受产地、气候、土壤、地势、采收季节等诸多因素影响，中药质量常出现参差不齐现象；有些珍贵野生中药由于过度的开采而处于濒危，亟需寻求新的药用部位和替代品种。中药本身是个复杂且庞大的混合体系，具有多成分、多靶点、多途径作用特点，这在一定程度上增加了中药质量评价的难度。前国内外对中药的质量评价主要是检测中药中的几个化学成分，并且所建立的评价方法均未依托药效。整体和全面可靠的中药质量评价体系的缺乏不但增加用药者的健康风险，甚至影响中药国际信誉度、竞争力和影响力。
CN108509997A公开了一种基于近红外光谱技术对中药皂角刺的真伪进行化学模式识别的方法，所述方法利用近红外光谱采集法、一阶导数预处理方法以及连续投影算法、Kennard-Stone算法以及步进算法的结合对中药皂角刺的真伪进行化学模式识别，使得模式识别方法的结果准确可靠，可以准确区分皂角刺及其伪品。然而，该方法仅仅是基于化学信息采集以及化学处理方法来获取特征波数点，该特征波数点与药品的药效学并不一定相关，多出的无关特征波数点导致判别模型较为复杂。
因此，为实现中药的现代化和国际化进程，迫切建立一种既尊重传统中医药理论，又能在现代科学药效实验的指导下，全面反映中药整体化学信息的中药质量评价方法。
发明内容
针对现有技术的不足，本发明的目的在于提供一种基于药效信息建立评价中药质量的化学模式识别方法。本发明的方法不需要化学对照品，能够全面反映中药的化学信息，以药效学信息进行化学模式识别模型的建立基础，使得判别模型更加准确，并且本发明克服鉴别的主观性，鉴别结果准确可靠。
为达到此发明目的，本发明采用以下技术方案：
本发明提供一种基于药效信息建立评价中药质量的化学模式识别方法，所述方法包括以下步骤：
(1)采集能反映中药样品内在质量的整体的化学信息，获取反映其临床疗效的药效信息，对化学信息与药效信息进行谱效关系分析，得到与药效显著相关的指标作为特征化学指标；
(2)将中药样品划分为训练集和测试集样本，采用有监督的模式识别方法以步骤(1)所述特征化学指标为输入变量，对训练集样本进行特征变量提取；
(3)利用步骤(2)提取的特征变量，建立模式识别模型；
(4)将测试集样本的特征变量值代入所述模式识别模型中，完成对所述中药质量的化学模式识别评价。
在本发明中，通过得到与药效显著相关的指标作为特征化学指标，并提取出有效的特征变量，进而建立模式识别模型，这些特征变量与药效显著相关，避免了无关变量的干扰以及造成的模式识别模型的复杂化，从而能够得到更为精准的模式识别模型，进行更为简单直接的中药真伪鉴别、中药质量分级等，结果准确可靠；并且本发明方法有助于为细贵中药寻找替代品。
在本发明中，所述中药包括化橘红、丹参、皂角刺、砂仁、功劳木或三七。在本发明中，所述采集中药样品的化学信息，是指针对中药的识别目标来采取其化学特征信息。例如如果目标是对中药真伪进行鉴别，那么就是采集能反映中药及其伪品样品内在质量的整体的化学 信息，如果目标是对中药进行质量分级，那么就是采集能反映质量等级的各等级中药样品内在质量的整体的化学信息。
在本发明中，获取反映中药临床疗效的药效信息采用中药药效研究中的常规手段进行。
优选地，步骤(1)所述采集能反映中药样品内在质量的整体的化学信息后，将采集到的数据转化为m×n的矩阵，其中n为中药样品的个数，m为每个中药样品采集到的化学信息的数量。
在本发明中，采集中药样品的化学信息的方法为光谱采集方法、色谱采集方法、质谱采集方法或核磁法；
优选地，所述光谱采集方法为紫外光谱、红外光谱、近红外光谱、拉曼光谱或荧光光谱中的任意一种；
优选地，所述色谱采集方法为高效液相色谱法或超高效液相色谱法。
在本发明中，所述采集化学信息即为采集能反映中药内在质量的整体的特征化学信号，例如如果利用紫外光谱进行采集就是采集中药的紫外特征吸收峰，如果利用高效液相进行采集就是采集中药在高效液相中的全部明显出峰。
在本发明中，所述对化学信息进行药效相关性分析是指将采集的化学信息与药效之间进行相关性分析，选择与药效显著相关的化学信息作为药效指标，剔除与药效不相关的化学信息。
在本发明中，步骤(1)所述谱效关系分析的方法可为双变量相关分析法、回归分析法、灰度关联分析法、偏最小二乘法或主成分分析法。
在本发明中，步骤(2)所述有监督的模式识别方法为主成分判别分析法、逐步判别分析法、偏最小二乘判别法、支持向量机或人工神经网络算法。
优选地，步骤(2)所述特征变量提取时，去除k个无关化学信息，得到(m-k)×n的矩阵，其中n为中药样品的个数，m为每个中药样品采集到的化学信息的数量。
在本发明中，所述基于药效信息建立评价中药质量的化学模式识别方法的流程如附图1所示，其反映了方法整体的整体流程，在药效(即药理活性)指导下来完成模式识别，以能够对中药进行质量评价以及对未知样品进行预测和分析。
优选地，所述基于药效信息建立中药质量的化学模式识别评价方法包括对中药丹参的真伪进行化学模式识别，对化橘红中毛橘红和光橘红进行化学模式区分或者对皂角刺的真伪进行化学模式识别。
优选地，所述对中药丹参的真伪进行化学模式识别或者对化橘红中毛橘红和光橘红进行化学模式区分的方法包括：
A、利用高效液相色谱法采集丹参及其伪品或者采集化橘红中毛橘红和光橘红的化学信息；对高效液相色谱中选取的特定吸收峰用Z标准化方法进行数据标准化，而后对标准化后的数据进行双变量谱效相关性分析，得到与及其伪品或者化橘红中毛橘红和光橘红的药效活性呈显著相关性的HPLC指纹数据，将该HPLC指纹数据作为反映药效的特征化学指标；
B、将丹参及其伪品或者化橘红样品随机划分为训练集和测试集样本，逐步判别分析法以步骤A所述特征化学指标为输入变量，对训练集样品的特征化学进行筛选，去除不相关的变量，筛选出特征变量；
C、利用步骤B得到的特征变量，针对丹参及其伪品的模式识别模型或者建立针对化橘红样品的模式识别模型；
D、将测试集样本的特征变量值代入所述模式识别模型中，以判别丹参及其伪品的判别准确率或者对化橘红中毛橘红和光橘红进行判别的准确率。
优选地，所述步骤A中丹参及其伪品的特定吸收峰的选取原则为满足如下至少一个条件的峰：(I)丹参、甘西鼠尾草和云南鼠尾草三者共有的峰；(II)丹参、甘西鼠尾草和云南鼠尾草各自特有的峰；(III)成分含量较大的峰。
优选地，步骤A中化橘红中毛橘红和光橘红的特定吸收峰的选取原则为毛橘红和光橘红 二者共有的峰。
在本发明中，选取的这些特定吸收峰代表了丹参、甘西鼠尾草和云南鼠尾草这3个药材的主要化学信息。
优选地，在步骤B中所述随机划分为训练集和测试集的方法为利用随机算法进行随机划分。
优选地，步骤B所述丹参及其伪品的训练集包括20批样品，其中包括12批丹参、4批甘西鼠尾草和4批云南鼠尾草；所述测试集包括29批样品，其中包括26批丹参、2批甘西鼠尾草和1批云南鼠尾草。在本发明中划分为训练集和测试集是随机划分的，因此并不局限于此处所述的包括特定数量批次的样品的训练集和测试集。
优选地，步骤B所述化橘红样品训练集包括22批样品，其中包括10批毛橘红样品和12批光橘红样品，所述测试集包括9批样品，其中包括5批毛橘红样品和4批光橘红样品。
优选地，步骤B筛选出的特征变量为X 6、X 7和X 13；也就是说在本申请的方法下虽然得到了很多与药效活性呈显著相关性的HPLC指纹数据，但是通过判别分析步进筛选得出的与分类相关的特征变量仅有3个，进而大大简化了模型函数。
优选地，步骤C所述的模式识别模型的函数如下：
F 1＝0.492X 6+8.762X 7-1.249X 13-1.869。
F 2＝-2.571X 6+4.521X 7+3.277X 13+1.288。
优选地，步骤B对化橘红样品筛选出的特征变量为X 7、X 8和X 20。
优选地，步骤C所述建立的针对化橘红样品的模式识别模型的函数如下：
F 1＝0.828X 7+0.767X 8-1.303X 20-0.099。
优选地，本发明所述对皂角刺的真伪进行化学模式识别的方法包括以下步骤：
I、利用近红外光谱采集皂角刺及其伪品的化学信息，获取反映中药临床疗效的药效信息；对化学信息与药效信息进行谱效关系分析，得到与药效显著相关的特征峰作为特征化学指标；
II、将皂角刺及其伪品随机划分为训练集和测试集样本，利用逐步判别分析法以步骤I所述特征化学指标为输入变量，对训练集样品的特征化学指标进行筛选，去除不相关的变量，筛选出特征变量；
III、利用步骤II得到的特征变量，建立模式识别模型；
IV、将测试集样品的特征变量值代入所述模式识别模型中，以判别皂角刺及其伪品的判别准确率。
优选地，步骤I所述利用近红外光谱采集皂角刺及其伪品的化学信息后还包括对化学信息进行光谱数据预处理：剔除原始光谱中的干扰峰以及水峰，得到11800～7500cm -1、6500～5500cm -1以及5000～4200cm -1谱段峰，选择5000～4200cm -1谱段峰作为模型分析峰，并采用一阶导数预处理方法对5000～4200cm -1谱段峰进行预处理，采用连续投影算法(SPA)进行特征峰提取。
优选地，所述干扰峰为12000～11800cm -1、4200～4000cm -1、7500～6500cm -1和5500～5000cm -1谱段的峰；所述水峰为7500～6500cm -1、5500～5000cm -1谱段的峰。
优选地，步骤II所述训练集包括32批样品，其中包括24批皂角刺、3批山皂角刺、2批野皂角刺和3批悬钩子，所述测试集样品包括11批样品，其中包括8批皂角刺、山1批皂角刺、1批野皂角刺和1批悬钩子。
优选地，步骤II筛选出的特征变量为X 8、X 10、X 14和X 21；
优选地，步骤III所述模式识别模型的函数为：
F 1＝49050.801X 8+8875.62X 10-2798.314X 14+21876.983X 21+2.356；
F 2＝-27730.331X 8+34288.661X 10-29368.865X 14+10924.346X 21+4.075。
相对于现有技术，本发明具有以下有益效果：
本发明的识别方法不需要化学对照品，能够全面反映中药的化学信息，以药效学信息进行化学模式识别模型的建立基础，使得判别模型与药效关系更加紧密，并且得到的化学模式 识别模型函数较为简单，又能够保证鉴别的准确性，克服了现行标准中仅以一种或几种成分含量进行中药质量评价的片面性和主观性，最终形成一种既能反映临床疗效又能兼顾化学成分信息的中药质量评价体系，鉴别结果准确可靠。可以利用本发明所述方法进行更为简单直接的中药真伪鉴别、中药质量分级等，结果准确可靠；并且本发明方法有助于为细贵中药寻找替代品，本方法可以实现对未知样品分类的预测，基于本发明所述方法形成一种中药质量评价体系。
附图说明
图1为基于药效信息建立评价中药质量的化学模式识别方法的总体流程图；
图2为对丹参、甘西鼠尾草和云南鼠尾草采集到的高效液相色谱图，其中S1为DS 3丹参样品采集到的高效液相色谱，S2为GX 39甘西鼠尾草样品采集到的高效液相色谱，S3为YN 45云南鼠尾草样品采集到的高效液相色谱；
图3为丹参及其伪品的训练集样品以判别函数值(即F 1和F 2的值，即函数1和函数2)为横纵坐标的分布图；
图4为丹参及其伪品的训练集及测试集样品以判别函数值(即F 1和F 2的值，即函数1和函数2)为横纵坐标的分布图；
图5为毛橘红样品的高效液相色谱图；
图6为光橘红样品的高效液相色谱图；
图7为化橘红训练集样品以样本序号及判别函数值(即F 1的值，即函数1得分值)为横纵坐标的分布图；
图8为化橘红训练集及测试集样品以样本序号及判别函数值(即F 1的值，即函数1得分值)为横纵坐标的分布图；
图9为本发明对皂角刺及其伪品样品进行红外光谱采集得到的原始平均近红外光谱图；
图10为采用一阶导数(First Derivative,1st D)方法对原始平均近红外光谱进行预处理后得到的近红外光谱图；
图11为皂角刺及其伪品的训练集样品以判别函数值(即F 1和F 2的值，即函数1和函数2)为横纵坐标的分布图；
图12皂角刺及其伪品的训练集及测试集样品以判别函数值(即F 1和F 2的值，即函数1和函数2)为横纵坐标的分布图。
具体实施方式
下面通过具体实施方式来进一步说明本发明的技术方案。本领域技术人员应该明了，所述实施例仅仅是帮助理解本发明，不应视为对本发明的具体限制。
在本发明中，所述基于药效信息建立评价中药质量的化学模式识别方法的总体流程如图1所示，收集典型具有代表性的中药，采集能反映中药样品内在质量的整体的化学信息，获取能反映中药临床疗效的药效信息，在药效指导下对化学信息进行的特征提取，获得能反映药效的特征化学指标，既对化学信息和药效信息进行药效相关性分析，得到与药效显著相关的化学信息指标作为特征指标；将中药样品划分为训练集和测试集；采用有监督的模式识别方法以能反映临床药效的特征化学指标为输入变量，对训练集样本进行特征变量提取；利用提取的特征变量，建立模式识别模型；将测试集样品的特征变量值代入所述模式识别模型中，从而药效(即药理活性)指导下完成对中药样品质量的化学模式识别评价。
实施例1
在本实施例中，使用的仪器与软件如下：
高效液相色谱：色谱柱：Zobax SB-aq(250mm×4.6mm，5μm，Agilent公司)；流动相：乙腈(A)-0.03％磷酸溶液(B)，梯度洗脱，洗脱程序见表1；检测波长：280nm，流速：0.8mL·min-1，柱温：30℃，进样量：20μL。
表1 梯度洗脱程序
Figure PCTCN2019122425-appb-000001
随机算法的运行采用SPSS软件(美国IBM公司)。
在本实施例中，使用的样品如下：
收集了不同产地丹参及同属植物甘西鼠尾草和云南鼠尾草样品共49批，所有样品均经北京中医药大学张继主任药师鉴定，1～38号丹参样品为鼠尾草属植物Salvia miltiorrhiza的干燥根和根茎，39～43号甘西鼠尾草样品为鼠尾草属植物Salvia przewalskii的干燥根和根茎，44～49号云南鼠尾草样品为鼠尾草属植物Salvia yunnanensis的干燥根和根茎，所有样品来源见表2。
表2 样品信息
Figure PCTCN2019122425-appb-000002
Figure PCTCN2019122425-appb-000003
对中药丹参的真伪进行化学模式识别的方法具体包括以下步骤：
1、化学信息采集
49批样品按照如上所述的高效液相色谱条件进样分析，记录色谱图，选取其中23个峰作为变量指标，选择原则为满足如下至少一个条件的峰均选为变量指标：1、丹参、甘西鼠尾草、云南鼠尾草三者共有的峰；2、丹参、甘西鼠尾草、云南鼠尾草各自特有的峰；3、成分含量较大的峰。因此这23个峰变量代表了3个药材的主要化学信息。选取色谱峰如图2所示，其中S1为DS 3丹参样品采集到的高效液相色谱，S2为GX 39甘西鼠尾草样品采集到的高效液相色谱，S3为YN 45云南鼠尾草样品采集到的高效液相色谱，3个样品的高效液相色谱中均标出了选取的相应峰编号。
49批样品的23个峰面积结果见表3-1和表3-2。
表3-1
Figure PCTCN2019122425-appb-000004
Figure PCTCN2019122425-appb-000005
表3-2
Figure PCTCN2019122425-appb-000006
Figure PCTCN2019122425-appb-000007
2、数据的标准化
进行多元统计分析时，往往要收集不同量纲的数据，这表现为变量在数量级和计量单位上的差别，从而使各个变量之间不具有综合性。而多元统计分析大多对变量有特殊的要求，比如符合正态分布或者变量之间具有可比性。这时就必须采用某种方法对各变量数值进行标准化处理。Z标准化方法是目前多变量综合分析中使用最多的一种方法，在原始数据呈正态 分布的情况下，需利用该方法进行数据无量纲处理。
本实验测定结果中不同峰面积的数值之间数量级差别较大，因此考虑用Z标准化方法计算，计算方法见如下公式，标准化后的新数据见表4-1和表4-2。
Figure PCTCN2019122425-appb-000008
表4-1
Figure PCTCN2019122425-appb-000009
Figure PCTCN2019122425-appb-000010
表4-2
Figure PCTCN2019122425-appb-000011
Figure PCTCN2019122425-appb-000012
3.丹参及其伪品抗心肌缺血药效测定
以大鼠心肌细胞缺氧复氧模型，通过测定大鼠心肌细胞存活率、乳酸脱氢酶(LDH)活性、活性氧(ROS)水平和细胞内钙离子浓度，比较丹参和两种易混淆品的75％甲醇提取物抗心肌缺血的作用，结果如表5所示。
表5 丹参及其伪品抗心肌缺血药效结果
Figure PCTCN2019122425-appb-000013
Figure PCTCN2019122425-appb-000014
4.谱效相关性分析
中药谱效关系研究是将化学成分即“谱”与药理作用即“效”相结合，从整体研究中药有效化学成分与其有效化学作用之间的关系。将49批中药的药效信息与HPLC指纹图谱数据进行了双变量相关谱效相关性分析。结果如表6所示。
表6 药效与指纹图谱数据相关分析结果
Figure PCTCN2019122425-appb-000015
**显著水平为0.01
*显著水平为0.05.
从表6中以看出与丹参、甘西鼠尾草和云南鼠尾草药效活性呈显著相关性的HPLC指纹数据为A6、A7、A8、A10、A13、A14、A17、A18、A19、A20、和A21。
5.训练集和测试集的划分
利用随机算法随机将49批样品划分为训练集和测试集，结果如下：
训练集样本：DS2,DS3,DS4,DS6,DS7,DS13,DS15,DS16,DS18,DS20,DS22,DS35,GX39,GX42,GX43,GX44,YN46,YN47,YN48,YN49)；
测试集样本：DS1,DS5,DS8,DS9,DS10,DS11,DS12,DS14,DS17,DS19,DS21,DS23,DS24,DS25,DS26,DS27,DS28,DS29,DS30,DS31,DS32,DS33,DS34,DS36,DS37,DS38,GX40,GX41,YN45。
6.药效指导下特征提取
用逐步判别分析法，对谱效相关分析结果中与药效具有显著相关的变量(A6、A7、A8、A10、A10、A13、A14、A17、A18、A19、A20、和A21)进行变量的筛选来特征提取变量，筛选是通过F检验逐步进行的。每一步选取满足指定水平最显著的变量并剔除因新变量的引入而变得不显著的原引入的变量，直到既不能引入也不能剔除为止。采用逐步判别分析，对丹参、甘西鼠尾草和云南鼠尾草三者同时进行对比，筛选出特征的有代表性的峰变量。降维结果(即筛选出的特征变量)见表7。
表7 组别和样品的特征峰
Figure PCTCN2019122425-appb-000016
7.模式识别模型的判别函数的建立
根据逐步判别选入的特征变量及判别系数，见表8，建立两个判别函数如下：
表8 典型区别函数系数
Figure PCTCN2019122425-appb-000017
F 1＝0.492X 6+8.762X 7-1.249X 13-1.869。
F 2＝-2.571X 6+4.521X 7+3.277X 13+1.288。
8、模型验证
(1)模型的内部验证。对模型进行留一法内部交叉验证，结果表明，在本如上所建模型中，留一法内部交叉验证判别准确率为100％。
(2)应用测试集对模型进行外部验证，将测试集样品特征峰代入判别函数，获得样品判别得分和判别分类结果。结果见表9，模型判别结果均与性状鉴别结果一致，判别正确率为100％。
表9 测试集样品判别结果表
Figure PCTCN2019122425-appb-000018
Figure PCTCN2019122425-appb-000019
8.结果可视化
以判别函数值为依据，做训练集和测试集样品的分布图。F1、F2分别为样品在分布图中的横纵坐标，分布图结果如图3(训练集)和图4(训练集及测试集)所示。在图3和图4中可以将训练集和测试集样品中丹参(DS)、甘西鼠尾草(GX)和云南鼠尾草(YN)进行有效区分。
因此，如上所述方法在药效指导下逐步判别分析进行特征提取，获取3个特征值，建立2个判别函数，可以对丹参、甘西鼠尾草和云南鼠尾草进行有效的区分。
实施例2
在本实施例中，使用的仪器如下：
高效液相色谱：色谱柱：Shiseido Capcell Pak C18(250mm×4.6mm，5μm，资生堂公司),
流动相：甲醇-0.5％冰醋酸梯度洗脱
洗脱梯度：采用二元梯度洗脱系统，溶剂A(甲醇)—溶剂B(0.5％冰醋酸)检测波长：320nm，流速：1.0mL·min -1，柱温：30℃，进样量：20μL。
梯度洗脱程序如下表10所示：
表10
Figure PCTCN2019122425-appb-000020
Figure PCTCN2019122425-appb-000021
在本实施例中，使用的样品如下：
实验共收集31批化橘红药材，其中15,7～16号为毛橘红，16～18,20-31号为光橘红，具体信息见表11(其中6号和19号为异常样本，进行了剔除)。
表11 化橘红样品信息
Figure PCTCN2019122425-appb-000022
对化橘红药材模式识别的具体方法包括以下步骤：
1、化学信息采集
将31批药材分别进行了HPLC色谱分析，获得了所有色谱峰数据，毛橘红样品的结果如图5所示，光橘红样品的结果如图6所示。
2、指纹图谱数据转化
获取橘红样品的共有峰数据，由于数据的个体间差异较大，甚至出现了不在同一数量级的问题，严重影响统计分析，因此有必要进行数据转化，使数值进行无量纲化，同时建立统一分析标准。采用标准化，得到结果如表12所示：
表12-1
Figure PCTCN2019122425-appb-000023
表12-2
Figure PCTCN2019122425-appb-000024
Figure PCTCN2019122425-appb-000025
3、药效信息获得
根据临床应用，将31批药材分别进行了止咳、祛痰和抗炎实验，药效学指标分别为潜伏期(越短越好)和咳嗽次数(越少越好)、酚红排出量(越多越好)、耳肿胀度(越小越好)，表13为得到的化橘红药效试验数据。
表13 化橘红药效实验数据
Figure PCTCN2019122425-appb-000026
Figure PCTCN2019122425-appb-000027
药效数据标准化
由于各种药效指标的数值测量单位及数量级不同，不能同时进行统计分析，根据数据标准化后，数据都进行无量纲化转换，进行对应分析，得到表14所示标准化数据。
表14
Figure PCTCN2019122425-appb-000028
Figure PCTCN2019122425-appb-000029
4、有效峰值与药效相关性分析
为判断有效峰值与药效关系，需要先判断各峰值与药效的相关性，获得能反映药效的特征化学指标，分析结果如表15所示。
表15
Figure PCTCN2019122425-appb-000030
**显著水平为0.01
*显著水平为0.05
由上表可知，各药效与各峰值的线性关系，与部分峰值有线性关系，但相关系数较小，大多数仅有0.7左右，与药效存在显著相关的峰有8个分别为X1、X7、X8、X10，X11、X14、X19和X20。
5、训练集和测试集的划分
采用随机算法将31批橘红样品划分训练集和测试集：
训练集样本：2、3、4、7、8、10、11、13、14、15、18、20、21、23、24、26、28、29、30、31、32、33。
测试集样本：1、5、9、12、16、17、22、25、27。
6、药效指导下特征提取
以与药效具有显著相关的指标峰组成的数据矩阵为依据(8*31的数据矩阵)采用逐步判别分析，筛选出对分类具有贡献的峰。采用逐步判别分析地方法，用Wilks'Lambda作为评价指标，采用相同概率在0.05以内选择为主要峰，保留该峰，相同概率大于0.1以上为无差异峰，剔除该峰，判别橘红的分类。
逐步判别变量特征提取结果如表16所示。
表16
Figure PCTCN2019122425-appb-000031
从上表中可以看出对橘红其分类作用的特征变量为X 7、X 8和X 20。
7.模式识别模型建立
以训练集样本为数据集，采用逐步判别分析中选出的特征变量X 7、X 8,X 20为输入变量，见表17。根据判别函数系数建立判别函数方程式。
表17 典型区别函数系数
Figure PCTCN2019122425-appb-000032
判别函数方程式为：F 1＝0.828X 7+0.767X 8-1.303X 20-0.099
F 1>0为毛橘红，F 1<0为光橘红
8、模型验证
(1)模型的内部验证。对模型进行留一法内部交叉验证，结果表明，在本如上所建模型中，留一法内部交叉验证判别准确率为100％。
(2)应用测试集对模型进行外部验证，将测试集样品特征峰代入判别函数，获得样品判别得分和判别分类结果。结果见表18，模型判别结果均与性状鉴别结果一致，判别正确率为100％。
表18 测试集样品判别结果表
Figure PCTCN2019122425-appb-000033
Figure PCTCN2019122425-appb-000034
9.结果可视化
以判别函数值和样品序号为依据，做训练集和测试集样品的分布图。判别函数值F 1、样本序列号为样品在分布图中的纵横坐标，分布图结果如图7(训练集)和图8(训练集及测试集)所示。在图7和图8中可以将训练集和测试集样品中毛橘红和光橘红进行有效区分。
因此，如上所述方法在药效指导下逐步判别分析进行特征提取，获取3个特征值，建立1个判别函数，可以对毛橘红和光橘红进行有效的区分。
实施例3
在本实施例中，使用的仪器和软件如表19所示：
表19 使用仪器及软件
Figure PCTCN2019122425-appb-000035
样品收集与预处理
样品收集
本实施例收集了典型代表性皂角刺及其伪品样品43批，其中正品皂角刺Gleditsia sinensis Lam.(G.sinensis)32批(1～32号)，伪品山皂角刺Gleditsia japonica Miq.(G.japonica)4批(33～36号)，伪品野皂角刺Gleditsia microphylla Gordon ex Y.T.Lee(G.microphylla)3批(37～39号)，伪品悬钩子Rubus cochinchinensis Tratt.(R.cochinchinensis)4批(40～42号)，所有样品均经北京中医药大学张继主任药师鉴定为中药皂角刺正品及其各类典型伪品，样品信息见表20。
表20
Figure PCTCN2019122425-appb-000036
Figure PCTCN2019122425-appb-000037
样品预处理
所有样品均洗净后去灰屑，干燥，粉碎过50目筛，25℃条件下密封备用。
1、近红外光谱采集
使用光纤探头采集样品的近红外光谱，采集范围为12000～4000cm-1，仪器分辨率为4cm -1，扫描次数为32次。扣除内置参比背景，每批样品采集3处不同位置的光谱，求平均光谱作为代表光谱。使用OPUS 6.5工作站(德国Bruker公司)求平均光谱。实验温度平衡在25℃，湿度保持在60％左右。皂角刺及其伪品的原始平均近红外光谱如图9所示。
光谱数据预处理方法
采用SG平滑、矢量归一化、最大最小归一化、一阶导数、二阶导数方法对样本光谱预处理，考察不同预处理方法对建模准确率的影响。使用OPUS 6.5工作站(德国Bruker公司)进行光谱数据预处理，如图10所示，为采用一阶导数(First Derivative,1st D)方法对原始平均近红外光谱进行预处理后得到的近红外光谱图。
谱段划分
剔除12000～11800cm -1、4200～4000cm -1两区间的噪声干扰峰，剔除7500～6500cm -1、 5500～5000cm -1两区间的水峰。剔除杂峰、水峰后的全谱段分成了三个区间，分别为11800～7500cm -1、6500～5500cm -1以及5000～4200cm -1。
提取特征波数
采用SPA算法提取了不同预处理条件下三个光谱区间内的特征波数。利用Matlab R2014a软件运行SPA算法，提取特征变量后大大降低了建模的复杂度。
通过前期研究发现采用5000～4200cm -1，一阶导数处理光谱建模分类识别正确率最优。因此本案例实施采用5000～4200cm -1，一阶导数处理光谱SPA特征提取数据(见表21-1、表21-2和表21-3)进行分析。
表21-1
Figure PCTCN2019122425-appb-000038
Figure PCTCN2019122425-appb-000039
表21-2
Figure PCTCN2019122425-appb-000040
Figure PCTCN2019122425-appb-000041
表21-3
Figure PCTCN2019122425-appb-000042
Figure PCTCN2019122425-appb-000043
2、皂角刺及其伪品药效数据的获取
(1)NO的测定-Griess法
巨噬细胞受到脂多糖(LPS)刺激后可激活细胞表面受体启动多种信号级联放大效应，导致一氧化氮(Nirtric Oxide,NO)、TNF-α、IL-6等促炎因子的产生，从而引起损伤。测定细胞上清液中NO水平可以检测炎症水平。
细胞培养上清液中NO极不稳定能很快被代谢生成相对稳定的亚硝酸根，亚硝酸根可以与Griess Reagent中对氨基苯磺酸和α-萘胺在酸性条件下反应，生成红色偶氮化合物在540nm处有最大吸收峰，并且产物浓度与NO浓度具有线性关系，因此可通过此原理测定细胞培养上清液中NO的含量。具体测定步骤如下：
I.配置亚硝酸钠标准品，准确配置10、20、40、60、80、100μM的亚硝酸钠溶液用于标准曲线测定；
II.取50μL/孔Griess reagent置96孔板中，加入I.步骤中的上清液或者不同浓度的亚硝酸钠标准品溶液50μL/孔，室温下反应30min后，除去孔中气泡，540nm下测定吸光度值；
III.根据亚硝酸钠标准溶液吸光度值制作标准曲线，将样品吸光度值代入标准曲线可得不用实验组中上清液中NO含量。
(2)抗氧化活性测定-ORAC法
ORAC法以荧光素钠(sodium flourescein，FL)为荧光探针，观察其与偶氮化合物2,2'-偶氮-双-(2-脒基丙烷)氯化二氢[2,2'-azobis(2-amidinopropane)dihydrochloride，AAPH]热分解产生的过氧化氢自由基作用后，荧光强度的衰退过程(在抗氧化物质存在时，其荧光衰退变缓)并且用抗氧化标准物质-水溶性维生素E类似物(6-hydro-2,5,7,8-tetramethylchroman-2-carboxylic acid，Trolox)当量来衡量体系中各种抗氧化物延缓探针荧光强度衰退的能力，以此评价抗氧化剂的抗氧化能力。
样品的NO抑制活性和ORAC抗氧化活性见表22。
表22.样本抗炎及抗氧化活性
Figure PCTCN2019122425-appb-000044
Figure PCTCN2019122425-appb-000045
“-”未检测
3、药效与近红外光谱的相关分析，寻找能反应药效的特征光谱
通过抗炎和抗氧化药效与SPA特征近红外进行皮尔逊双尾相关分析，从分析结果见表23，可以得出峰X1、X7、X8、X9、X10、X12、X13、X14、X20、X21、X22、X23、X24、X25、X26、X27、X28与皂角刺药效呈显著相关，为与药效相关的特征近红外光谱。
表23，皂角刺抗炎抗氧化与SPA特征近红外光谱相关分析结果
Figure PCTCN2019122425-appb-000046
*0.05水平下显著相关，**0.01水平下显著相关
4、训练集和测试集划分
Kennard-Stone算法，训练集样品包括32批样品，其中包括24批皂角刺、3批山皂角刺、2批野皂角刺和3批悬钩子，所述测试集样品包括11批样品，其中包括8批皂角刺、山1批皂角刺、1批野皂角刺和1批悬钩子。
训练集样本：2，5，6，7，8，9，10，12，13，14，15，16，17，18，19，20，21，22，23，24，25，26，28，30，34，35，36，38，39,，41，42，43；
测试集样本：1，3，4，11，27，29，31，32，33，37，40。
5、药效指导下特征提取
用逐步判别分析法，对谱效相关分析结果中与药效具有显著相关的变量(X1、X7、X8、X9、X10、X12、X13、X14、X20、X21、X22、X23、X24、X25、X26、X27、X28)进行变量的筛选来特征提取变量，筛选是通过F检验逐步进行的。每一步选取满足指定水平最显著的变量并剔除因新变量的引入而变得不显著的原引入的变量，直到既不能引入也不能剔除为止。采用逐步判别分析，对皂角刺、山皂角刺、野皂角刺四者同时进行对比，筛选出特征的有代表性的峰变量。降维结果见表24。
表24 组别和样品的特征峰
Figure PCTCN2019122425-appb-000047
6.模式识别模型的判别函数的建立
根据逐步判别选入的特征变量及判别系数，见表25，建立两个判别函数如下：
表25 典型判别函数系数
Figure PCTCN2019122425-appb-000048
F 1＝49050.801X 8+8875.62X 10-2798.314X 14+21876.983X 21+2.356。
F 2＝-27730.331X 8+34288.661X 10-29368.865X 14+10924.346X 21+4.075。
7、模型验证
(1)模型的内部验证。对模型进行留一法内部交叉验证，结果表明，在本如上所建模型中，留一法内部交叉验证判别准确率为100％。
(2)应用测试集对模型进行外部验证，将测试集样品特征峰代入判别函数，获得样品判别得分和判别分类结果。结果见表26，模型判别结果均与性状鉴别结果一致，判别正确率为100％。
表26 测试集样品判别结果表
Figure PCTCN2019122425-appb-000049
8.结果可视化
以判别函数值为依据，做训练集和测试集样品的分布图。以F1、F2分别为样品在分布图中的横纵坐标，分布图结果如图11(训练集)和图12(训练集及测试集)所示。在图11和图12中可以将训练集和测试集样品中皂角刺(ZJC)、山皂角刺(SZJ)、野皂角刺(YZJ)和悬钩子(XGZ)进行有效区分。
因此，如上所述方法在药效指导下逐步判别分析进行特征提取，获取4个特征值，建立 2个判别函数，可以对皂角刺、山皂角刺、野皂角刺和悬钩子刺进行有效的区分。
申请人声明，本发明通过上述实施例来说明本发明的方法，但本发明并不局限于上述工艺步骤，即不意味着本发明必须依赖上述工艺步骤才能实施。所属技术领域的技术人员应该明白，对本发明的任何改进，对本发明所选用原料的等效替换及辅助成分的添加、具体方式的选择等，均落在本发明的保护范围和公开范围之内。

Claims (13)

  1. 一种基于药效信息建立评价中药质量的化学模式识别方法，其中所述方法包括以下步骤：
    (1)采集能反映中药样品内在质量的整体的化学信息，获取反映其临床疗效的药效信息，对化学信息与药效信息进行谱效关系分析，得到与药效显著相关的指标作为特征化学指标；
    (2)将中药样品划分为训练集和测试集样本，采用有监督的模式识别方法以步骤(1)所述特征化学指标为输入变量，对训练集样本进行特征变量提取；
    (3)利用步骤(2)提取的特征变量，建立模式识别模型；以及
    (4)将测试集样本的特征变量值代入所述模式识别模型中，完成对所述中药质量的化学模式识别评价。
  2. 根据权利要求1所述的方法，其中所述中药包括化橘红、丹参、皂角刺、砂仁、功劳木或三七。
  3. 根据权利要求1所述的方法，其中步骤(1)所述采集能反映中药样品内在质量的整体的化学信息后，将采集到的数据转化为m×n的矩阵，其中n为中药样品的个数，m为每个中药样品采集到的化学信息的数量。
  4. 根据权利要求1至3任一项所述的方法，其中采集中药样品的化学信息的方法为光谱采集方法、色谱采集方法、质谱采集方法或核磁法。
  5. 根据权利要求4所述的方法，其中所述光谱采集方法为紫外光谱、红外光谱、近红外光谱、拉曼光谱或荧光光谱中的任意一种。
  6. 根据权利要求4所述的方法，其中所述色谱采集方法为高效液相色谱法或超高效液相色谱法。
  7. 根据权利要求1-6中任一项所述的方法，其中步骤(1)所述谱效关系分析的方法可为双变量相关分析法、回归分析法、灰度关联分析法、偏最小二乘法或主成分分析法；
    优选地，步骤(2)所述有监督的模式识别方法为主成分判别分析法、逐步判别分析法、偏最小二乘判别法、支持向量机或人工神经网络算法；
    优选地，步骤(2)所述特征变量提取时，去除k个无关化学信息，得到(m-k)×n的矩阵，其中n为中药样品的个数，m为每个中药样品采集到的化学信息的数量。
  8. 根据权利要求1-7中任一项所述的方法，其中所述基于药效信息建立评价中药质量的化学模式识别方法包括对中药丹参的真伪进行化学模式识别，对化橘红中毛橘红和光橘红进行化学模式识别或者对皂角刺的真伪进行化学模式识别。
  9. 根据权利要求1-8中任一项所述的方法，其中所述对中药丹参的真伪进行化学模式识别或者对化橘红中毛橘红和光橘红进行化学模式识别的方法包括：
    A、利用高效液相色谱法采集能反映丹参及其伪品或者采集能反映化橘红中毛橘红和光橘红样品内在质量的整体的化学信息，获取反映中药临床疗效的药效信息；对高效液相色谱中选取的特定吸收峰用Z标准化方法进行数据标准化，而后对标准化后的数据进行双变量谱效相关性分析，得到与丹参及其伪品或者化橘红中毛橘红和光橘红的药效活性呈显著相关性的HPLC指纹数据，将该HPLC指纹数据作为反映药效的特征化学指标；
    B、将丹参及其伪品或者化橘红样品随机划分为训练集和测试集样本，利用逐步判别分析法以步骤A所述特征化学指标为输入变量，对训练集样品的特征化学指标进行筛选，去除不相关的变量，筛选出特征变量；
    C、利用步骤B得到的特征变量，建立针对丹参及其伪品的模式识别模型或者建立针对化橘红样品的模式识别模型；
    D、将测试集样品的特征变量值代入所述模式识别模型中，以判别丹参及其伪品的判别准确率或者对化橘红中毛橘红和光橘红进行判别的准确率。
  10. 根据权利要求1-9中任一项所述的方法，其中所述步骤A中丹参及其伪品的特定吸收峰的选取原则为满足如下至少一个条件的峰：(I)丹参、甘西鼠尾草和云南鼠尾草三者共有的峰；(II)丹参、甘西鼠尾草和云南鼠尾草各自特有的峰；(III)成分含量较大的峰；
    优选地，步骤A中化橘红中毛橘红和光橘红的特定吸收峰的选取原则为毛橘红和光橘红二者共有的峰；
    优选地，在步骤B中所述随机划分为训练集和测试集的方法为利用随机算法进行随机划分；
    优选地，步骤B所述丹参及其伪品的训练集包括20批样品，其中包括12批丹参、4批甘西鼠尾草和4批云南鼠尾草；所述测试集包括29批样品，其中包括26批丹参、2批甘西鼠尾草和1批云南鼠尾草；
    优选地，步骤B所述化橘红样品训练集包括22批样品，其中包括10批毛橘红样品和12批光橘红样品，所述测试集包括9批样品，其中包括5批毛橘红样品和4批光橘红样品；
    优选地，步骤B对丹参及其伪品筛选出的特征变量为X 6、X 7和X 13；
    优选地，步骤B对化橘红样品筛选出的特征变量为X 7、X 8和X 20。
    优选地，步骤C所述建立的针对丹参及其伪品的模式识别模型的函数如下：
    F 1＝0.492X 6+8.762X 7-1.249X 13-1.869；
    F 2＝-2.571X 6+4.521X 7+3.277X 13+1.288；
    优选地，步骤C所述建立的针对化橘红样品的模式识别模型的函数如下：
    F 1＝0.828X 7+0.767X 8-1.303X 20-0.099。
  11. 根据权利要求1-8中任一项所述的方法，其中对皂角刺的真伪进行化学模式识别的方法包括以下步骤：
    I、利用近红外光谱采集能反映皂角刺及其伪品样品内在质量的整体的化学信息，获取反映中药临床疗效的药效信息；对化学信息与药效信息进行谱效关系分析，得到与药效显著相关的特征峰作为特征化学指标；
    II、将皂角刺及其伪品随机划分为训练集和测试集样本，利用逐步判别分析法以步骤I所述特征化学指标为输入变量，对训练集样品的特征化学指标进行筛选，去除不相关的变量，筛选出特征变量；
    III、利用步骤II得到的特征变量，建立模式识别模型；
    IV、将测试集样品的特征变量值代入所述模式识别模型中，以判别皂角刺及其伪品的判别准确率。
  12. 根据权利要求11所述的方法，其中步骤I所述利用近红外光谱采集皂角刺及其伪品的化学信息后还包括对化学信息进行光谱数据预处理：剔除原始光谱中的干扰峰以及水峰，得到11800～7500cm -1、6500～5500cm -1以及5000～4200cm -1谱段峰，选择5000～4200cm -1谱段峰作为模型分析峰，并采用一阶导数预处理方法对5000～4200cm -1谱段峰进行预处理，采用连续投影算法进行特征峰提取；
    优选地，所述干扰峰为12000～11800cm -1、4200～4000cm -1、7500～6500cm -1和5500～5000cm -1谱段的峰；所述水峰为7500～6500cm -1、5500～5000cm -1谱段的峰；
    优选地，步骤II所述训练集包括步骤II所述训练集包括32批样品，其中包括24批皂角刺、3批山皂角刺、2批野皂角刺和3批悬钩子，所述测试集样品包括11批样品，其中包括8批皂角刺、山1批皂角刺、1批野皂角刺和1批悬钩子。
  13. 根据权利要求11或12所述的方法，其中步骤II筛选出的特征变量为X 8、X 10、X 14和X 21；
    优选地，步骤III所述模式识别模型的函数为：
    F 1＝49050.801X 8+8875.62X 10-2798.314X 14+21876.983X 21+2.356；
    F 2＝-27730.331X 8+34288.661X 10-29368.865X 14+10924.346X 21+4.075。
PCT/CN2019/122425 2019-09-25 2019-12-02 一种基于药效信息建立评价中药质量的化学模式识别方法 WO2021056814A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
US17/779,835 US11710541B2 (en) 2019-09-25 2019-12-02 Chemical pattern recognition method for evaluating quality of traditional Chinese medicine based on medicine effect information
EP19946314.2A EP3907493A4 (en) 2019-09-25 2019-12-02 CHEMICAL PATTERN RECOGNITION PROCEDURE FOR ASSESSING THE QUALITY OF TRADITIONAL CHINESE MEDICINE BASED ON DRUG EFFECT INFORMATION

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910913203.2A CN110514611B (zh) 2019-09-25 2019-09-25 一种基于药效信息建立评价中药质量的化学模式识别方法
CN201910913203.2 2019-09-25

Publications (1)

Publication Number Publication Date
WO2021056814A1 true WO2021056814A1 (zh) 2021-04-01

Family

ID=68632430

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/122425 WO2021056814A1 (zh) 2019-09-25 2019-12-02 一种基于药效信息建立评价中药质量的化学模式识别方法

Country Status (4)

Country Link
US (1) US11710541B2 (zh)
EP (1) EP3907493A4 (zh)
CN (1) CN110514611B (zh)
WO (1) WO2021056814A1 (zh)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113239322A (zh) * 2021-06-25 2021-08-10 河北中烟工业有限责任公司 一种津巴布韦进口烟叶替代模块的构建方法
CN113358774A (zh) * 2021-05-25 2021-09-07 广西民生中检联检测有限公司 一种凌云白毫绿茶识别方法
CN114444946A (zh) * 2022-01-28 2022-05-06 荃豆数字科技有限公司 一种中药饮片精细化运营指导方法、装置及计算机设备
CN115239994A (zh) * 2022-07-12 2022-10-25 中关村科学城城市大脑股份有限公司 应用于甘草饮片的等级信息生成方法、装置和电子设备
WO2022264138A1 (en) * 2021-06-14 2022-12-22 Neswell Group Ltd. Scoring and classifying extracts from a non-animal natural eukaryotic source

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108509997A (zh) * 2018-04-03 2018-09-07 深圳市药品检验研究院（深圳市医疗器械检测中心） 一种基于近红外光谱技术对中药皂角刺的真伪进行化学模式识别的方法
CN110514611B (zh) * 2019-09-25 2023-01-20 深圳市药品检验研究院（深圳市医疗器械检测中心） 一种基于药效信息建立评价中药质量的化学模式识别方法
CN111122635B (zh) * 2019-12-10 2023-05-12 北京农业智能装备技术研究中心 一种食用菌干燥状态在线评估方法及系统
CN110967428B (zh) * 2019-12-23 2021-11-19 深圳市药品检验研究院（深圳市医疗器械检测中心） 一种基于高效液相指纹图谱信息建立评价中药化橘红质量的模式识别方法
CN112666111A (zh) * 2020-10-01 2021-04-16 华中农业大学 奶牛奶与马奶的快速鉴别方法
CN113567639B (zh) * 2021-07-13 2023-05-16 中国食品药品检定研究院 中药材质量综合评价方法
CN113740474A (zh) * 2021-08-27 2021-12-03 深圳市药品检验研究院（深圳市医疗器械检测中心） 一种结合抗新冠病毒效应的金银花和山银花鉴别模型及其构建方法和应用

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20090062323A (ko) * 2007-12-12 2009-06-17 대한민국 (식품의약품안전청장) 대사지문법을 이용한 간독성 예측방법
CN105138861A (zh) * 2015-05-31 2015-12-09 青岛市食品药品检验研究院 一种大黄药材真伪及基源种属类别预测模型的构建方法
CN106290643A (zh) * 2016-08-17 2017-01-04 辽宁中医药大学 一种中药荆芥抗肺癌活性成分的质量控制方法
CN108195989A (zh) * 2018-02-01 2018-06-22 山西省医药与生命科学研究院 一种基于抗血栓谱效关系的黄刺玫化学成分评价方法
CN108509997A (zh) 2018-04-03 2018-09-07 深圳市药品检验研究院（深圳市医疗器械检测中心） 一种基于近红外光谱技术对中药皂角刺的真伪进行化学模式识别的方法
CN109856265A (zh) * 2019-01-15 2019-06-07 广西中医药大学 基于抗氧化和降血糖谱效关系的龙眼叶化学成分评价方法
CN110514611A (zh) * 2019-09-25 2019-11-29 深圳市药品检验研究院（深圳市医疗器械检测中心） 一种基于药效信息建立评价中药质量的化学模式识别方法

Family Cites Families (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
UA78194C2 (en) * 2000-12-08 2007-03-15 Council Scient Ind Res Method for revealing and identification of active components in extracts by chromatographic fingerprinting of "fingerprints" and processor of program data processing chromatograms
CN1938586B (zh) * 2004-01-28 2015-05-13 科学与工业研究委员会 使用活动色谱指纹图谱的食品和药物的化学及治疗价值的标准化方法
US7710269B2 (en) * 2007-01-11 2010-05-04 University Of Southern California Systems and methods to prevent counterfeit, grey and black market proliferation of pharmaceutical, medical and other products
US11162892B2 (en) * 2016-01-27 2021-11-02 The Usa, As Represented By The Secretary, Dept. Of Health And Human Services Device and method for detection of counterfeit pharmaceuticals
WO2018136465A1 (en) * 2017-01-20 2018-07-26 Western New England University System and method of inferring the quality level of medicines
CN108732126A (zh) * 2017-04-25 2018-11-02 天士力医药集团股份有限公司 一种采用近红外光谱法测定丹参药材中多成分含量的方法
CN107132194A (zh) * 2017-07-10 2017-09-05 天津工业大学 一种基于紫外可见光谱及化学模式识别的三七及其伪品鉴别方法
CN109507123A (zh) * 2018-11-22 2019-03-22 天津工业大学 一种基于紫外可见漫反射光谱及化学模式识别的当归及其相似品的鉴别方法
CN110110789A (zh) * 2019-05-08 2019-08-09 杭州麦迪特检测技术服务有限公司 一种基于多谱图信息融合技术的中草药品质鉴别方法
CN110187020A (zh) * 2019-05-17 2019-08-30 山西大学 基于治疗慢性萎缩性胃炎谱效关系的黄芪成分评价方法

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20090062323A (ko) * 2007-12-12 2009-06-17 대한민국 (식품의약품안전청장) 대사지문법을 이용한 간독성 예측방법
CN105138861A (zh) * 2015-05-31 2015-12-09 青岛市食品药品检验研究院 一种大黄药材真伪及基源种属类别预测模型的构建方法
CN106290643A (zh) * 2016-08-17 2017-01-04 辽宁中医药大学 一种中药荆芥抗肺癌活性成分的质量控制方法
CN108195989A (zh) * 2018-02-01 2018-06-22 山西省医药与生命科学研究院 一种基于抗血栓谱效关系的黄刺玫化学成分评价方法
CN108509997A (zh) 2018-04-03 2018-09-07 深圳市药品检验研究院（深圳市医疗器械检测中心） 一种基于近红外光谱技术对中药皂角刺的真伪进行化学模式识别的方法
CN109856265A (zh) * 2019-01-15 2019-06-07 广西中医药大学 基于抗氧化和降血糖谱效关系的龙眼叶化学成分评价方法
CN110514611A (zh) * 2019-09-25 2019-11-29 深圳市药品检验研究院（深圳市医疗器械检测中心） 一种基于药效信息建立评价中药质量的化学模式识别方法

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP3907493A4

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113358774A (zh) * 2021-05-25 2021-09-07 广西民生中检联检测有限公司 一种凌云白毫绿茶识别方法
CN113358774B (zh) * 2021-05-25 2023-10-03 广西民生中检联检测有限公司 一种凌云白毫绿茶识别方法
WO2022264138A1 (en) * 2021-06-14 2022-12-22 Neswell Group Ltd. Scoring and classifying extracts from a non-animal natural eukaryotic source
CN113239322A (zh) * 2021-06-25 2021-08-10 河北中烟工业有限责任公司 一种津巴布韦进口烟叶替代模块的构建方法
CN114444946A (zh) * 2022-01-28 2022-05-06 荃豆数字科技有限公司 一种中药饮片精细化运营指导方法、装置及计算机设备
CN115239994A (zh) * 2022-07-12 2022-10-25 中关村科学城城市大脑股份有限公司 应用于甘草饮片的等级信息生成方法、装置和电子设备

Also Published As

Publication number Publication date
EP3907493A1 (en) 2021-11-10
US20230017825A1 (en) 2023-01-19
US11710541B2 (en) 2023-07-25
EP3907493A4 (en) 2023-02-15
CN110514611B (zh) 2023-01-20
CN110514611A (zh) 2019-11-29

Similar Documents

Publication Publication Date Title
WO2021056814A1 (zh) 一种基于药效信息建立评价中药质量的化学模式识别方法
US11656176B2 (en) Near-infrared spectroscopy-based method for chemical pattern recognition of authenticity of traditional Chinese medicine Gleditsiae spina
CN105138861B (zh) 一种大黄药材真伪及基源种属类别预测模型的构建方法
CN101961379B (zh) 丹参的近红外光谱鉴别方法
CN101961360B (zh) 三七的近红外光谱鉴别方法
CN111007032B (zh) 一种快速无损鉴别甘草及其伪品刺果甘草的近红外光谱法
CN104568822B (zh) 一种连翘药材多指标同时快速检测方法
CN102288551A (zh) 利用高光谱数据鉴别中药材的方法
CN104076012A (zh) 一种近红外光谱法快速检测冰片质量的模型建立方法
CN103364359A (zh) Simca模式识别法在近红外光谱识别大黄药材中的应用
Chen et al. Rapid and automatic chemical identification of the medicinal flower buds of Lonicera plants by the benchtop and hand-held Fourier transform infrared spectroscopy
CN104730030A (zh) 基于近红外分析技术的党参真伪鉴别和产地判定的方法
CN103674887B (zh) 冬虫夏草全草的红外光谱分析鉴定方法
CN102243170A (zh) 用近红外光谱技术鉴别麦冬药材产地的方法
CN113029999B (zh) 一种姜科中药配方颗粒的近红外光谱鉴别模型构建方法和鉴别方法
CN112414967B (zh) 一种快速实时检测蒲黄炭炮制的近红外质控方法
CN105138834A (zh) 基于近红外光谱波数k均值聚类的烟草化学值定量方法
CN104345045A (zh) 一种基于化学模式识别和近红外光谱的相似药材鉴别方法
Yang et al. Discrimination of Radix Astragali according to geographical regions by data fusion of laser induced breakdown spectroscopy (LIBS) and infrared spectroscopy (IR) combined with random forest (RF)
CN103760126B (zh) 冬虫夏草虫体的红外光谱分析鉴定方法
Zhang et al. Discrimination of geographical origin and adulteration of radix astragali using fourier transform infrared spectroscopy and chemometric methods
CN105334183A (zh) 一种基于近红外光谱鉴别法定麻黄药材的方法
AU2018101606A4 (en) A method for identifying meconopsis quintuplinervia regel from different geographical origins
CN103076300A (zh) 专属性模式识别模型判别分析中药材资源指纹信息的方法
CN116008245A (zh) 桑叶拉曼光谱指纹图谱的建立结合机器学习算法在桑叶属地来源鉴定中的应用

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19946314

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2019946314

Country of ref document: EP

Effective date: 20210805

NENP Non-entry into the national phase

Ref country code: DE