Traditional Chinese medicine constitution classification method based on plasma metabolism profile

Abstract

The invention belongs to the technical field of traditional Chinese medicine constitution classification, and particularly relates to a traditional Chinese medicine constitution classification method based on plasma metabolism profiles. The method comprises the following steps: constructing a multiple linear regression prediction model of a physique by taking a plasma metabolism profile as an independent variable and an original score of one physique in traditional Chinese medicine physiques as a dependent variable; according to the same method, multiple linear regression prediction models of the other eight physiques are constructed respectively; respectively inputting plasma metabolism profiles of healthy people with unknown traditional Chinese medicine constitution types into the constructed multiple linear regression prediction model of the nine constitution types, and predicting metabolic constitution original scores of the nine constitution types; converting the original metabolic constitution score to obtain a conversion score, and judging the traditional Chinese medicine constitution classification according to the conversion score. The method has the advantages that the traditional Chinese medicine theory and the modern biological technology can be combined, an objective scientific basis is provided for traditional Chinese medicine constitution classification, the accuracy and credibility of traditional Chinese medicine constitution classification are improved, and the defects of traditional subjective evaluation are overcome.

Description

Technical Field

[0001] The present invention belongs to the technical field of TCM constitution classification, and in particular relates to a TCM constitution classification method based on plasma metabolic profile. Background Art

[0002] Constitution in traditional Chinese medicine refers to the physical characteristics and physiological state of an individual under the influence of genetics, environment, lifestyle and other factors. It reflects the individual's health status, physiological functions and ability to adapt to the environment. The difference in constitution is reflected in the reaction and adaptability to external stimuli, as well as the susceptibility to certain pathogenic factors. Studying constitution helps to analyze the relationship between different constitution types and diseases, and provides a reference for disease prevention and treatment.

[0003] Professor Wang Qi divides human constitution into nine types: peaceful constitution, qi deficiency constitution, yang deficiency constitution, yin deficiency constitution, phlegm-damp constitution, damp-heat constitution, blood stasis constitution, qi stagnation constitution and special constitution. Among them, the peaceful constitution is a relatively healthy constitution, and the others are biased constitutions. The "Standards for Classification and Determination of TCM Constitution" is currently recognized as the diagnostic standard for TCM constitution. In a nationwide epidemiological survey of 21,948 cases, it has shown good practicality and operability, and has also been promoted and applied internationally. The identification of TCM constitution is of great value in health management. By understanding their own constitution type, people can better understand their own health status, adjust their lifestyle in time, improve their physical condition, and prevent the occurrence of related diseases.

[0004] The Standard for Classification and Determination of TCM Constitution contains a total of 67 questions, each of which has a five-level score, namely, never, rarely, sometimes, often, and always, corresponding to 1 point, 2 points, 3 points, 4 points, and 5 points, respectively. In some questions related to the peaceful constitution, there is a *, indicating that reverse scoring is required, that is, level 1 corresponds to 5 points, level 2 corresponds to 4 points, level 3 corresponds to 3 points, level 4 corresponds to 2 points, and level 5 corresponds to 1 point. By answering these questions and combining the standard table for the determination of the peaceful constitution and the biased constitution (Table 1), you can determine your own constitution type. However, the fast pace of modern life and high-pressure work make it difficult for people to take the time to answer all the questions carefully. In addition, the questionnaire survey mainly relies on the subjective feelings and judgments of the subjects, which are easily affected by subjective factors such as personal emotions, cognitive levels, and understanding abilities, thus affecting the objectivity and accuracy of the evaluation results. In addition, the questionnaire survey mainly reflects the subjective feelings of individuals, and lacks consideration of objective indicators such as physiological status, so it cannot fully reflect the overall condition of individuals. Therefore, a more comprehensive, objective and quantifiable method is urgently needed to classify TCM constitutions. Summary of the invention

[0005] In view of the above-mentioned deficiencies in the prior art, the purpose of the present invention is to provide a TCM constitution classification method based on plasma metabolic profile, aiming to solve the problem that the existing self-assessed constitution type is easily affected by subjective factors, thereby affecting the objectivity and accuracy of the evaluation results.

[0006] The technical solution of the present invention is as follows:

[0007] The TCM constitution classification method based on plasma metabolic profile includes:

[0008] Step A, taking the plasma metabolic profile as the independent variable and the original score of one of the TCM constitutions as the dependent variable, constructing a multiple linear regression prediction model of the constitution; wherein the TCM constitution includes 9 constitutions;

[0009] Step B: Using the same method as step A, construct multiple linear regression prediction models for the other 8 constitutions in TCM constitution respectively;

[0010] Step C, inputting the plasma metabolic profiles of healthy people with unknown TCM constitution types into the constructed multiple linear regression prediction models of the nine constitutions, respectively, and predicting the metabolic constitution raw scores of the nine constitutions;

[0011] Step D: transforming the metabolic constitution original scores of the 9 constitutions respectively to obtain the transformation scores of the 9 constitutions, and determining the TCM constitution classification according to the transformation scores.

[0012] Optionally, the step A specifically includes:

[0013] Step A1, obtaining the plasma metabolic profile of healthy people with known TCM constitution types, wherein TCM constitutions include 9 constitutions, and the known TCM constitution types include the original scores of the known 9 constitutions;

[0014] Step A2, constructing a multiple linear regression model using the plasma metabolic profile of healthy people with known TCM constitution types as an independent variable and the original score of one of the TCM constitutions as a dependent variable;

[0015] Step A3, optimizing the multiple linear regression model to obtain a multiple linear regression prediction model for the constitution.

[0016] Optionally, a plasma metabolic profile is obtained by:

[0017] Collecting plasma from healthy people and pre-treating the plasma to obtain a test solution;

[0018] Using a chromatography-mass spectrometer to detect the test liquid to obtain detection data;

[0019] The detection data is processed, and then metabolites are identified to obtain the plasma metabolic profile.

[0020] Optionally, in step A1, the original scores of the 9 constitutions of healthy people are calculated by the following method: according to the constitution determination method of the "Standards for Classification and Determination of Traditional Chinese Medicine Constitutions", the self-test forms of the Traditional Chinese Medicine constitution survey of healthy people are collected, and corresponding scores are given according to the answers of the healthy people, and the total scores of the 9 constitutions are calculated respectively to obtain the original scores of the 9 constitutions.

[0021] Optionally, the 9 constitutions include balanced constitution, qi deficiency constitution, yang deficiency constitution, yin deficiency constitution, phlegm-damp constitution, damp-heat constitution, blood stasis constitution, qi stagnation constitution and special constitution.

[0022] Optionally, in step A3, the multiple linear regression model is optimized by a step function of stepwise regression analysis to obtain a multiple linear regression prediction model for the constitution.

[0023] Optionally, in step A3, a one-by-one elimination cross-validation method is used to evaluate the predictive ability of the multivariate linear regression prediction model.

[0024] Optionally, according to the following formula [1], the metabolic constitution raw scores of the 9 constitutions are converted respectively to obtain the conversion scores of the 9 constitutions;

[0025] Formula [1]: Conversion score = [(original score - number of items) / (number of items × 4)] × 100.

[0026] Optionally, the step of determining the TCM constitution classification according to the conversion score specifically includes: determining the TCM constitution classification according to the conversion score and according to the criteria for determining a balanced constitution and an unbalanced constitution.

[0027] The present invention has the following beneficial effects: the present invention uses metabolomics technology to analyze the metabolite composition of individual biological samples, obtains the individual's metabolic spectrum, and combines the TCM constitution classification system to classify the individual's constitution. The method of predicting the TCM constitution classification of individuals based on metabolic profiles has the advantage of combining TCM theory and modern biological technology to provide an objective scientific basis for TCM constitution classification, improve the accuracy and credibility of TCM constitution classification, and provide a scientific basis and guidance for the health management of individuals related to TCM constitution and the prevention and treatment of diseases. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 Schematic diagram of the Leave-one-out Cross-Validation method.

[0029] Figure 2This is a linear correlation analysis chart of the questionnaire original scores (abscissa, original scores) and metabolic prediction original scores (ordinate, predicted scores) of the 93 subject samples in the modeling group.

[0030] Figure 3 This is a radar chart of the biased constitution prediction results for subject No. 8 in the validation group (blue: metabolic prediction biased constitution classification, red: questionnaire survey biased constitution classification).

[0031] Figure 4 This is a radar chart of the biased constitution prediction results for subject No. 5 in the validation group (blue: metabolic prediction biased constitution classification, red: questionnaire survey biased constitution classification).

[0032] Figure 5 This is the confusion matrix diagram of the metabolic prediction biased constitution of subject No. 8 in the validation group. DETAILED DESCRIPTION

[0033] The present invention provides a TCM constitution classification method based on plasma metabolic profile. To make the purpose, technical solution and effect of the present invention clearer and more specific, the present invention is further described in detail below. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0034] Constitution refers to the comprehensive and relatively stable characteristics of a human individual formed by inheritance or acquisition during the life process, including morphological structure, physiological function and psychological state. In the physiological state, constitution manifests itself as differences in response and adaptability to external stimuli, as well as susceptibility to certain pathogenic factors and tendency to develop diseases. Studying constitution helps to analyze the relationship between different constitution types and diseases, and provides a basis for preventing and treating diseases.

[0035] Metabolomics is the technology that is closest to biological phenotypes, and the information feedback from metabolomics fluctuations is also closest to biological phenotypes. By extensively and holistically detecting and analyzing endogenous and exogenous small molecule compounds in organisms, it is possible to depict the metabolite spectrum at a certain time point and observe the biochemical response of organisms to environmental stimuli. Therefore, the detection and analysis of metabolites is the "molecular pulse diagnosis" of metabolic abnormalities in organisms by modern life science technology, and is an objective means of evaluation. Traditional questionnaire surveys are easily affected by subjective factors, while metabolomics data can reflect the metabolic state of the human body at the molecular level, providing a more objective biological basis for constitution classification. Therefore, the concept of "metabolic molecular pulse TCM constitution classification" is transformed from "taking the pulse" to "metabolic molecular pulse" as the connection point between "taking the pulse" and "metabolic molecular pulse". Metabolic molecular pulse TCM constitution classification is a health assessment method that combines modern biology and TCM theory. It applies modern metabolomics technology to the TCM constitution classification system, establishes a unified and repeatable biological standard for constitution classification by analyzing the composition and level of individual metabolites, and avoids the subjectivity and inconsistent standards of existing self-assessed constitution types.

[0036] Specifically, the embodiment of the present invention provides a TCM constitution classification method based on plasma metabolic profile, which includes:

[0037] Step A, taking the plasma metabolic profile as the independent variable and the original score of one of the TCM constitutions as the dependent variable, constructing a multiple linear regression prediction model of the constitution; wherein the TCM constitution includes 9 constitutions, and the 9 constitutions include peaceful constitution, qi deficiency constitution, yang deficiency constitution, yin deficiency constitution, phlegm-damp constitution, damp-heat constitution, blood stasis constitution, qi stagnation constitution and special constitution;

[0038] Step B: Using the same method as step A, construct multiple linear regression prediction models for the other 8 constitutions in TCM constitution respectively;

[0039] Step C, inputting the plasma metabolic profiles of healthy people with unknown TCM constitution types into the constructed multiple linear regression prediction models of the nine constitutions, respectively, and predicting the metabolic constitution raw scores (also referred to as prediction scores herein) of the nine constitutions;

[0040] Step D: transforming the metabolic constitution original scores of the 9 constitutions respectively to obtain the transformation scores of the 9 constitutions, and determining the TCM constitution classification according to the transformation scores.

[0041] It should be noted that metabolic profile refers to the overall profile of metabolite composition and relative content obtained by comprehensive analysis and identification of metabolites in biological samples through metabolomics technology. It reflects the metabolic activity and physiological state of the body under a specific state. Metabolic profile can include many different types of metabolites, such as carbohydrates, lipids, amino acids, nucleotides, etc. Metabolic profile analysis is an important means to study the metabolic state of organisms and discover disease biomarkers.

[0042] The holistic characteristics of metabolomics are consistent with the thinking mode of "holistic view", "differential diagnosis and treatment", and "studying the outside and understanding the inside" of traditional Chinese medicine. Therefore, the present invention uses metabolomics technology to analyze the metabolite composition of individual biological samples, obtain the individual's metabolic spectrum, and classify the individual's constitution in combination with the traditional Chinese medicine constitution classification system, and finally develop a method for predicting the individual's traditional Chinese medicine constitution classification based on metabolic profile. The advantage of this method is that it can combine traditional Chinese medicine theory and modern biological technology to provide an objective scientific basis for the classification of traditional Chinese medicine constitutions, improve the accuracy and credibility of the classification of traditional Chinese medicine constitutions, and provide a scientific basis and guidance for the health management of individuals related to traditional Chinese medicine constitutions and the prevention and treatment of diseases.

[0043] The present invention utilizes modern life science technology means of metabolomics to objectively reflect the metabolic state of the human body at the molecular level, overcome the defects of traditional subjective evaluation, and establish more objective and standardized TCM constitution classification standards and methods.

[0044] The present invention also has the following technical advantages: 1) It is a non-invasive method that uses a small amount of blood sample to comprehensively detect the metabolomics spectrum; 2) It can be applied to high-throughput screening (HTS) detection, which is convenient and fast and can produce results quickly; 3) Through the modeling group and the verification group, a high-specificity, high-sensitivity and high-accuracy TCM constitution discrimination model is constructed.

[0045] In one embodiment, the step A specifically comprises:

[0046] Step A1, obtaining the plasma metabolic profile of healthy people with known TCM constitution types, wherein TCM constitutions include 9 constitutions, and the known TCM constitution types include the original scores of the known 9 constitutions;

[0047] Step A2, constructing a multiple linear regression model using the plasma metabolic profile of healthy people with known TCM constitution types as an independent variable and the original score of one of the TCM constitutions as a dependent variable;

[0048] Step A3, optimizing the multiple linear regression model to obtain a multiple linear regression prediction model for the constitution.

[0049] Using the function (linear model) used to fit the linear regression model in R language, a multivariate linear regression model between the metabolic profile and the original score of TCM constitution was constructed for the metabolomics data of the modeling group samples. Among them, the independent variable is the metabolite characteristics of the metabolome spectrum, and the dependent variable is the original score of one of the 9 constitutions of TCM constitution (for example: peaceful constitution). That is, a multivariate linear regression model between the metabolic profile and the original score of TCM constitution was constructed. When fitting the regression model on the modeling group data, the model was trained with the minimization of the mean absolute error (MAE) as the evaluation condition, and the optimal regression coefficient and its prediction model were obtained to determine the linear relationship between the metabolic profile and the original score of TCM constitution.

[0050] In one embodiment, in step A1, the raw scores of the 9 constitutions of healthy people are calculated by the following method: according to the constitution determination method of the "Standards for Classification and Determination of Traditional Chinese Medicine Constitutions", the self-test forms of the Traditional Chinese Medicine constitution survey of healthy people are collected, and corresponding scores are given according to the answers of the healthy people, and the total scores of the 9 constitutions are calculated respectively to obtain the raw scores of the 9 constitutions.

[0051] In one embodiment, in step A3, the multiple linear regression model is optimized by using the step function of stepwise regression analysis to obtain a multiple linear regression prediction model for the constitution.

[0052] In one embodiment, in step A3, a one-by-one elimination cross-validation method is used to evaluate the predictive ability of the multivariate linear regression prediction model.

[0053] In one embodiment, the plasma metabolic profile is obtained by:

[0054] Collecting plasma from healthy people and pre-treating the plasma to obtain a test solution;

[0055] Using a chromatography-mass spectrometer to detect the test liquid to obtain detection data;

[0056] The detection data is processed, and then metabolites are identified to obtain the plasma metabolic profile.

[0057] In step D, in one embodiment, the metabolic constitution raw scores of the nine constitutions are respectively converted according to the following formula [1] to obtain the conversion scores of the nine constitutions;

[0058] Formula [1]: Conversion score = [(original score - number of items) / (number of items × 4)] × 100.

[0059] In one embodiment, the step of determining the TCM constitution classification according to the conversion score specifically includes: determining the TCM constitution classification according to the conversion score and according to the criteria for determining a balanced constitution and an unbalanced constitution.

[0060] The present invention is further described in detail below through specific examples.

[0061] Example 1

[0062] 1. Detection object and sample information

[0063] 1) Test subjects: Written informed consent was obtained from all subjects before participating in the study.

[0064] 2) Inclusion criteria for healthy subjects:

[0065] (1) Male or female aged ≥18 years;

[0066] (2) Read and fully understand the patient information and sign the informed consent form;

[0067] (3) According to the questionnaire, subjects with common chronic diseases (hypertension, diabetes, coronary heart disease), a history of tumor treatment, and a history of major surgery were excluded;

[0068] (4) self-reported no obvious clinical symptoms;

[0069] (5) The physical examination results are within the normal range and no obvious abnormalities are found;

[0070] (6) Physical examination results are outside the normal range or are abnormal, but the doctor determines that they are of no clinical significance (physical examination items include: low-dose spiral CT (abbreviated as LDCT), abdominal color Doppler ultrasound, 4 tumor markers, blood pressure, etc.).

[0071] Subjects who met the above six conditions at the same time were included as healthy subjects.

[0072] 3) Sample information: A total of 108 healthy subjects were enrolled, of which 93 plasma samples were collected from the modeling group and 15 plasma samples were collected from the validation group. The samples used in the modeling group and validation group were actual samples collected from two medical centers, that is, the samples in the modeling group and the validation group were different samples, as shown in Table 2. Plasma samples were collected in the early morning on an empty stomach, and all collected plasma samples were stored in a -80 degree refrigerator.

[0073] 2. Sample pretreatment and metabolite detection, questionnaire survey to obtain data

[0074] 1) Sample pretreatment: After taking out the plasma sample from the -80 degree refrigerator, thaw it on ice, and then vortex mix it for 10 seconds. Then, take 100 μL of the mixed plasma, place it in a pre-cooled 1000 μL mixture of methyl tert-butyl ether and methanol (the volume ratio of methyl tert-butyl ether and methanol is 3:1), and vortex mix it to obtain sample extract 1; add 500 μL of a mixture of methanol and water (the volume ratio of methanol and water is 3:1) to the sample extract 1, and after ultrasonication, standing, vortexing and centrifugation, obtain the layered sample extract 2.

[0075] Organic phase: From sample extract 2, take 350 μL of the upper organic phase into a centrifuge tube, spin dry it using a vacuum concentrator (Speed-Vac), add 200 μL of a mixture of acetonitrile and isopropanol (the volume ratio of acetonitrile and isopropanol is 3:1) to dissolve it, and incubate it at room temperature for 15 minutes; vortex the incubated solution to mix it, ultrasonically treat it for 5 minutes, and then centrifuge the above solution at room temperature for 5 minutes (12000rpm); after centrifugation, take 180 μL from the supernatant and put it into a 2 mL glass injection vial as the organic phase, which is detected by the machine (LC-MS).

[0076] Aqueous phase: 400 μL of the lower aqueous phase from sample extract 2 was taken to a centrifuge tube, and 1100 μL of ice methanol was added thereto to precipitate the protein; after protein precipitation, incubated at 4°C for 1 hour. Then, centrifuged at 12700 rpm for 10 minutes at 4°C, 1000 μL of the supernatant was transferred to a new centrifuge tube, and dried with a vacuum concentrator (Speed-Vac); 200 μL of water was added to the dried centrifuge tube, and incubated at room temperature for 15 minutes; the incubated solution was vortexed and ultrasonically treated for 5 minutes, and the above solution was centrifuged at room temperature for 5 minutes (12000 rpm); after centrifugation, 180 μL of the supernatant was taken to a 2 mL glass injection vial as the aqueous phase, and detected by LC-MS.

[0077] 2) Metabolite detection: Waters ACQUTTY was used for organic phase BEH C8 (1.7μm 2.1*100mm) chromatographic column, water phase using Waters ACQUTTY HSS T3 (1.8 μm 2.1*100 mm) chromatographic column was used for small molecule separation; ACQUITY UPLC I-Class liquid chromatography system (Waters) and Q-Exactive mass spectrometry system (Thermo Fisher Scientific) were used for liquid chromatography and mass spectrometry.

[0078] The mobile phase parameters are as follows:

[0079] Organic phase: mobile phase A is an aqueous solution containing 0.1% acetic acid and 1% ammonium acetate; mobile phase B is a mixture of acetonitrile and isopropanol containing 0.1% acetic acid and 1% ammonium acetate (the volume ratio of acetonitrile and isopropanol is 7:3), and the flow rate is 0.4mL / min; the separation elution gradient is as follows: 0-1 minute is 55% mobile phase B, 1-4 minutes is 55%-75% mobile phase B, 4-12 minutes is 75%-89% mobile phase B, 12-15 minutes is 89%-100% mobile phase B, 15-19.5 minutes is 100% mobile phase B, 19.5-24 minutes is 55% mobile phase B, and % are all volume percentages.

[0080] Aqueous phase: Mobile phase A is an aqueous solution containing 0.1% formic acid, mobile phase B is an acetonitrile solution containing 0.1% formic acid, and the flow rate is 0.4 mL / min; the separation elution gradient is as follows: 0-1 minute is 1% mobile phase B, 1-11 minutes is 1%-40% mobile phase B, 11-13 minutes is 40%-70% mobile phase B, 13-15 minutes is 70%-99% mobile phase B, 15-19 minutes is 99% mobile phase B, 19-22 minutes is 1% mobile phase B, wherein % are all volume percentages.

[0081] The mass spectrometry parameters are as follows: the mass spectrometry data were collected in Full MS and Full MS / dd-MS2 modes (each including positive and negative modes). The parameters used for Q Exactive are as follows: the resolution of the Full MS mode is 70,000, the scanning ranges of the primary and secondary mass spectrometers are 100-1500 m / z and 50-1500 m / z, respectively, the AGC is 3E+6, and the Maximum IT is 200 milliseconds; in the FullMS / dd-MS2 mode, the resolution of the secondary mass spectrometer is 17,500, the quadrupole window is 1.5 m / z, the AGC is 1E+5, the maximum ion injection time is 50 ms, and the HCD relative collision energy is 30 eV.

[0082] 3) TCM Questionnaire Collection:

[0083] (1) Questionnaire collection: According to the constitution determination method of the "Standards for Classification and Determination of TCM Constitution", the subjects carefully answered the questions based on their experiences and feelings in the past three months, and collected the self-test form of the TCM constitution survey. The TCM constitution theory describes that human constitution is divided into 9 types, including peaceful constitution, qi deficiency constitution, yang deficiency constitution, yin deficiency constitution, phlegm-damp constitution, damp-heat constitution, blood stasis constitution, qi stagnation constitution, and special constitution. Each constitution has 7-8 questions, and each question has five levels of answers to choose from, namely, never, rarely, sometimes, often, and always. According to the corresponding score, record the score value.

[0084] (2) Classification of TCM constitutions:

[0085] ① Calculation of conversion scores based on questionnaire survey: According to the answers of the subjects on the TCM constitution questionnaire and the corresponding scores based on the five-level answers, scores are assigned, and then the total score is calculated as the original score for each constitution, that is, the original score = the sum of the scores of each item, and finally 9 original scores for 9 constitutions are obtained, where each item refers to each question; then, the calculated original score is converted to obtain the corresponding conversion score, the formula is as follows:

[0086] Formula [1]: Conversion score = [(original score - number of items) / (number of items × 4)] × 100;

[0087] ② Constitution judgment standard: According to the conversion score in the questionnaire, combined with the standard table for judging the peaceful constitution and the biased constitution, the constitution category of the subjects was judged (as shown in Table 1). Among them, the peaceful constitution is the normal constitution, and the other 8 constitutions are biased constitutions (Qi deficiency constitution, Yang deficiency constitution, Yin deficiency constitution, phlegm dampness constitution, damp heat constitution, blood stasis constitution, Qi stagnation constitution, and special constitution).

[0088] Table 1. Standards for Classifying and Determining Constitutions in Traditional Chinese Medicine - Standards for Determining Balanced and Unbalanced Constitutions

[0089]

[0090] 3. Metabolomics data processing and metabolite identification

[0091] 1) Metabolomics data processing: The steps of metabolomics data processing include peak extraction, peak alignment, baseline correction, peak filtering, isotope peak removal, missing value filling, etc. Specifically, firstly, the detection peaks are extracted and aligned from the mass spectrometry raw data, then the baseline correction is applied to remove the noise, the original signal peak is retained, the raw data is converted into central discrete data, and the isotope peaks in the data set are removed; based on machine learning algorithms, such as the nearest neighbor algorithm (KNN), random forest (RF), singular value decomposition (SVD), etc., missing values ​​are filled to obtain the final mass spectrometry matrix data; in order to reduce the difference in metabolite concentrations between samples and make the data distribution more symmetrical, the Normalization Autoencoder (NormAE) is used for homogenization.

[0092] 2) Identification of metabolites: The original data is analyzed by software to obtain the spectral information of the compound parent ion and secondary fragment ion, such as the mass-to-charge ratio (m / z) of the primary mass spectrum (MS1), the fragments of the secondary ion (MS2), and the retention time. The metabolites are qualitatively identified by matching the spectral information of the primary and secondary metabolites in the database. Commonly used metabolite, small molecule compound and mass spectrum databases include HMDB (www.hmdb.ca), PubChem

[0093] (https: / / pubchem.ncbi.nlm.nih.gov), MassBank (http: / / www.massbank.jp), MassBank of North America (https: / / massbank.us), and lipid databases Lipidblast (https: / / fiehnlab.ucdavis.edu / projects / LipidBlast) and Lipidmap

[0094] (https: / / www.lipidmaps.org); Based on the metabolites identified in the relevant database, the metabolites were finally verified according to the retention time, MS1, and MS2 mass spectrometry information of the standard when separated under the same chromatographic column and mass spectrometry conditions. The criteria for metabolite identification were retention time within 0.1min difference and metabolite molecular weight error less than 10ppm.

[0095] 4. TCM constitution classification method based on metabolic profile

[0096] 1) Collection of TCM questionnaires and constitution determination: Based on the constitution determination method in the “Standards for Classification and Determination of TCM Constitutions”, the self-test forms of the TCM constitution survey were collected from the subjects, and corresponding scores were given according to the subjects’ answers. The total score was then calculated to obtain the original score (original score = the sum of the scores of each item). The original score was then converted into a transformed score according to formula [1]. Finally, the TCM constitution was determined using the standards for the classification of mild and partial constitutions as the TCM constitution classification based on the questionnaire survey.

[0097] 2) Multiple linear regression analysis: Use the function (linear model) in R language for fitting linear regression models to construct a multiple linear regression model between the metabolic profile and the original score of TCM constitution for the metabolomics data of the subjects in the modeling group. Among them, the independent variable is the metabolite characteristics of the metabolome spectrum, and the dependent variable is the original score of one of the 9 constitutions of TCM constitution (for example: peaceful constitution). That is, a multiple linear regression model between the metabolic profile and the original score of TCM constitution is constructed. When fitting the regression model on the modeling group data, the model is trained with the minimization of the mean absolute error (MAE) as the evaluation condition to obtain the optimal regression coefficient and its prediction model, and determine the linear relationship between the metabolic profile and the original score of TCM constitution. For the metabolic profile of subjects with unknown TCM constitution type, this prediction model will be used to predict their metabolic constitution original score based on the metabolic profile.

[0098] 3) Multiple linear regression is a statistical method used to build a prediction model to explain the relationship between a dependent variable and multiple independent variables. Its mathematical formula is as follows:

[0099] Formula [2]:

[0100] Y i =β0+β1X 1i +β2X 2i +…+β k X ki +ε i

[0101] in:

[0102] Y i is the observed value of the dependent variable;

[0103] X 1i , X 2i ,...,X ki is the observed value of the independent variable;

[0104] β0, β1, β2, ..., β k is the regression coefficient, which indicates the effect of the independent variable on the dependent variable;

[0105] ε i is the error term, which represents the random part that cannot be explained by the model.

[0106] 4) The goal of multiple linear regression is to estimate the values ​​of the regression coefficients by minimizing the mean absolute error (MAE, which is the difference between the observed values ​​and the values ​​predicted by the regression equation). Specifically, the least squares method is usually used to estimate the regression coefficients so that the mean absolute error is minimized. In other words, multiple linear regression can be used to find a straight line (or a hyperplane in the case of multiple dimensions) that fits the observed data as closely as possible in order to predict the value of the dependent variable.

[0107] Formula [3]:

[0108]

[0109] in:

[0110] MAE is the mean absolute error, which represents the average absolute difference between the observed and predicted values;

[0111] n is the sample size, which indicates how many total observations there are in the data set;

[0112] y t is the ith observation;

[0113] is the predicted value of the ith observation given by the model.

[0114] 5. Prediction of TCM constitution classification based on metabolic profile

[0115] Based on the multivariate linear regression model established by fitting linear regression, the step function of stepwise regression analysis is applied to feature screening and optimization in the process of building the multivariate linear regression model to establish the optimal multivariate linear regression prediction model; based on the multivariate linear regression prediction model, the original metabolic constitution score predicted based on the metabolic profile under the constitution (for example, peaceful constitution) is finally obtained; similarly, the original score of another constitution (for example, Qi deficiency constitution) among the 9 constitutions is used as the dependent variable, and the independent variable is still the feature of the metabolome spectrum, and the multivariate linear regression prediction model is reconstructed; based on the multivariate linear regression prediction model, the original metabolic constitution score predicted based on the metabolic profile under the constitution (for example, Qi deficiency constitution) is obtained, and the same method is repeated 9 times to obtain the original metabolic constitution scores predicted based on the metabolic profile of all constitution types; the original metabolic constitution score is converted by the conversion formula [2], and the TCM constitution classification is predicted according to the criteria for judging peaceful constitution and biased constitution.

[0116] 6. External Validation

[0117] The metabolomic spectra of the validation group subjects, i.e. metabolic profiles, were used as the unknown sample data set and input into the multivariate linear regression prediction model of the nine constitutions established above. The metabolic constitution raw scores of the nine constitutions predicted based on the metabolic profiles were obtained respectively. The converted scores were converted using formula [1], and the TCM constitution category was determined based on the converted scores. Finally, they were compared with the TCM constitution classification based on the questionnaire to evaluate the prediction effect of the model.

[0118] 7. Results Analysis

[0119] 1) Subjects: The samples used in the modeling group and the validation group were actual samples collected from two medical centers, that is, the samples in the modeling group and the validation group were different samples. After the prediction model was constructed using the samples in the modeling group, the samples in the validation group were used for the final validation (Table 2).

[0120] Table 2. Sample information

[0121]

[0122] 2) TCM Questionnaire and Constitution Identification: A total of 108 subjects' TCM constitution questionnaires were collected, and the original scores and conversion scores in the questionnaires were calculated according to the corresponding formulas. Finally, the TCM constitution was determined by the criteria for the balanced constitution and the biased constitution, as the TCM constitution classification based on the questionnaire survey. TCM constitutions are divided into 9 types, and the specific criteria for the balanced constitution and the biased constitution are as follows:

[0123] -Peace quality judgment criteria:

[0124] When the conversion score of the peaceful constitution is ≥60 points, and the conversion scores of the other 8 biased constitutions (Qi deficiency constitution, Yang deficiency constitution, Yin deficiency constitution, Phlegm-damp constitution, Damp-heat constitution, Blood stasis constitution, Qi stagnation constitution, and Special constitution) are all less than 30 points, it indicates that the constitution is a peaceful constitution;

[0125] When the conversion score of the balanced constitution is ≥60 points, and the conversion scores of the other 8 biased constitutions are all between 30 and 40 points, but not 40 points, it indicates that the constitution is basically balanced.

[0126] When all the above conditions are not met, it indicates that the constitution is not of the peaceful nature;

[0127] -Criteria for judging biased constitution:

[0128] When the conversion scores of the eight biased constitutions are ≥40 points, it indicates that the constitution is biased;

[0129] When the conversion scores of the eight biased constitutions are between 30 and 40 points, but not 40 points, it indicates that the constitution tends to be biased constitution;

[0130] When the conversion scores of the 8 biased constitutions are <30 points, it indicates that the constitution is not a biased constitution.

[0131] 3) Multiple linear regression analysis:

[0132] Using the data of 93 subjects in the modeling group, a multiple linear regression model was established using the function of fitting the linear regression model. Among them, the independent variable is the metabolite characteristics of the metabolome spectrum, and the dependent variable is the original score of one of the 9 types of TCM constitutions (for example, peaceful constitution), that is, the multiple linear regression model between the metabolic profile and the original score of TCM constitution. The model was optimized by the step function of the stepwise recursive algorithm to obtain the optimal multiple linear regression prediction model, and the original score of the metabolic constitution predicted under this constitution type (for example, peaceful constitution) was obtained. Similarly, the independent variable is kept as the metabolite characteristics of the metabolome spectrum, and then the original score of another constitution type (for example, qi deficiency) is used as the dependent variable to reconstruct the multiple linear regression prediction model between the metabolic profile and the original score of TCM constitution, and the original score of the metabolic constitution predicted under this constitution type (for example, qi deficiency) is obtained. The same method was repeated 9 times, and the original scores of metabolic constitutions predicted for all constitution types were finally obtained. The original metabolic constitution score is converted into a score using formula [1], and the TCM constitution is identified according to the criteria for determining the balanced constitution and the biased constitution, thereby obtaining a TCM constitution classification based on the metabolic profile.

[0133] In order to verify the multiple linear regression prediction model between metabolic profile and TCM constitution raw score, and the discrimination ability in the reconstruction and optimization process, the Leave-one-out Cross-Validation method was used. Figure 1 As shown in the figure, cross-validation refers to the 93 sample data of the modeling group (SP1, SP2, SP3, SP4, SP5, SP6, SP7, SP8, ..., SP 93 ), randomly select one sample as the test set, and the remaining 92 samples as the training set. Use the 92 sample data of the training set to build a multivariate linear regression prediction model, and then test the samples of the test set to predict the metabolic fitness raw scores of the test set samples. Similarly, repeat this process 93 times, so that each sample in the data set is modeled 92 times as a training set and verified once as a test set. The advantage of this method is that it uses as much data as possible as a training set, and each training set and test set are independent of each other and completely cover the entire data set.

[0134] Based on the established multivariate linear regression prediction model between metabolic profile and TCM constitution original score, the metabolic constitution original score is predicted, and the predicted metabolic constitution original score is transformed using formula [1] to obtain the corresponding transformation score, and the TCM constitution classification is determined based on the transformation score. Figure 2 Linear correlation analysis between the predicted metabolic constitution raw scores (vertical axis, predicted scores) and the questionnaire raw scores (horizontal axis, raw scores) of 93 subjects with qi stagnation constitution. The results showed that R 2 They are all greater than 0.96 and close to 1, indicating that the model fits the data very well, the linear correlation between the two variables is strong, and the prediction results are more accurate;

[0135] External validation of the multivariate linear regression prediction model: The metabolic profile data of the 15 subjects in the validation group were used as unknown data sets and were placed in the multivariate linear regression prediction model between the 9 metabolic profiles and the original TCM constitution scores established above to evaluate the prediction effect of the model on the independent validation group data outside the modeling group. Based on the metabolic profile, the metabolic constitution original scores of the subjects in the validation group were predicted according to the multivariate linear regression prediction model. The predicted metabolic constitution original scores were transformed using formula [1] to obtain the corresponding transformation scores, and the TCM constitution classification was determined according to the criteria for the balanced constitution and biased constitution. The results show (Table 3) that in the prediction of balanced constitution, among the 15 subjects, 2 subjects were "yes", 2 subjects were "basically yes", and 11 subjects were "no". The metabolic prediction results were completely consistent with the questionnaire survey results, with an accuracy rate of 100%. Table 4 and Figure 3-Figure 4The prediction results of partial constitution of the subjects in the validation group are shown (please refer to the subject number in Table 3 for the serial number). The results show that the metabolic prediction results are consistent with the questionnaire survey results. For subject No. 8, the metabolic prediction results and the questionnaire survey results are completely consistent in the prediction of yang deficiency constitution, yin deficiency constitution, phlegm dampness constitution, damp heat constitution, special constitution, and qi stagnation constitution; for subject No. 5, the metabolic prediction results and the questionnaire survey results are completely consistent in the prediction of damp heat constitution and blood stasis constitution. For the case of incomplete consistency, it was found that the metabolic prediction results did not predict the new constitution type that did not appear in the questionnaire survey constitution, but "was a partial constitution" was judged as "tends to be a partial constitution", or "tends to be a partial constitution" was judged as "was a partial constitution". In the prediction of partial constitution of 15 subjects in the validation group, the mean absolute error (MAE) of the metabolic prediction and the questionnaire survey of the original score of the traditional Chinese medicine constitution was less than 28.4%-35.7% of the mean standard deviation (MAD) (performance indicators of 9 regression models). Using formula [1] and the Standard for Determining TCM Constitution Classification, we judged and compared the TCM constitution classification results of metabolic prediction and questionnaire survey, and generated a three-classification confusion matrix. Figure 5 The confusion matrix of the metabolic prediction biased constitution of subject No. 8 is shown. When B "tends to be" is judged as wrong, the accuracy and precision of metabolic prediction are 0.78 and 0.75 respectively; if B "tends to be" is judged as correct, the accuracy and precision of metabolic prediction are both 1.0. This further shows that the established multivariate regression linear prediction model based on metabolic profile and TCM constitution raw score has a 100% prediction accuracy for balanced constitution and a high accuracy for biased constitution. It has good prediction ability and can effectively classify TCM constitution.

[0136] Table 3. Prediction results of the peace quality of the subjects in the validation group

[0137]

[0138]

[0139] Table 4. Prediction results of biased constitution of some subjects in the validation group

[0140]

[0141] In summary, the present invention provides a method for TCM constitution classification based on plasma metabolic profile. The present invention uses metabolomics technology to analyze the metabolite composition of individual biological samples, obtain the individual's metabolic spectrum, and classify the individual's constitution in combination with the TCM constitution classification system. The method of the present invention for predicting individual TCM constitution classification based on metabolic profile has the advantage that it can combine TCM theory and modern biological technology to provide an objective scientific basis for TCM constitution classification, improve the accuracy and credibility of TCM constitution classification, and provide a scientific basis and guidance for health management and disease prevention and treatment of individuals related to TCM constitution.

[0142] It should be understood that the application of the present invention is not limited to the above examples. For ordinary technicians in this field, improvements or changes can be made based on the above description. All these improvements and changes should fall within the scope of protection of the claims attached to the present invention.

Claims

1. A TCM constitution classification method based on plasma metabolic profile, characterized in that: include: Step A, taking the plasma metabolic profile as the independent variable and the original score of one of the TCM constitutions as the dependent variable, constructing a multiple linear regression prediction model of the constitution; wherein the TCM constitution includes 9 constitutions; Step B: Using the same method as step A, construct multiple linear regression prediction models for the other 8 constitutions in TCM constitution respectively; Step C, inputting the plasma metabolic profiles of healthy people with unknown TCM constitution types into the constructed multiple linear regression prediction models of the nine constitutions, respectively, and predicting the metabolic constitution raw scores of the nine constitutions; Step D: transforming the metabolic constitution original scores of the 9 constitutions respectively to obtain the transformation scores of the 9 constitutions, and determining the TCM constitution classification according to the transformation scores.

2. The TCM constitution classification method based on plasma metabolic profile according to claim 1, characterized in that: The step A specifically comprises: Step A1, obtaining the plasma metabolic profile of healthy people with known TCM constitution types, wherein TCM constitutions include 9 constitutions, and the known TCM constitution types include the original scores of the known 9 constitutions; Step A2, constructing a multiple linear regression model using the plasma metabolic profile of healthy people with known TCM constitution types as an independent variable and the original score of one of the TCM constitutions as a dependent variable; Step A3, optimizing the multiple linear regression model to obtain a multiple linear regression prediction model for the constitution.

3. The TCM constitution classification method based on plasma metabolic profile according to claim 2, characterized in that: In step A1, the original scores of the 9 constitutions of healthy people are calculated by the following method: according to the constitution determination method of the "Standards for Classification and Determination of Traditional Chinese Medicine Constitutions", the self-test forms of the Traditional Chinese Medicine constitution survey of healthy people are collected, and corresponding scores are given according to the answers of the healthy people, and the total scores of the 9 constitutions are calculated respectively to obtain the original scores of the 9 constitutions.

4. The TCM constitution classification method based on plasma metabolic profile according to claim 1 or 2, characterized in that: The nine constitutions include balanced constitution, qi deficiency constitution, yang deficiency constitution, yin deficiency constitution, phlegm-damp constitution, damp-heat constitution, blood stasis constitution, qi stagnation constitution and special constitution.

5. The TCM constitution classification method based on plasma metabolic profile according to claim 2, characterized in that: In step A3, the multiple linear regression model is optimized by the step function of stepwise regression analysis to obtain the multiple linear regression prediction model of the constitution.

6. The TCM constitution classification method based on plasma metabolic profile according to claim 2, characterized in that: In step A3, a one-by-one elimination cross-validation method is used to evaluate the predictive ability of the multivariate linear regression prediction model.

7. The TCM constitution classification method based on plasma metabolic profile according to claim 1 or 2, characterized in that: Plasma metabolic profiles were obtained by: Collecting plasma from healthy people and pre-treating the plasma to obtain a test solution; Using a chromatography-mass spectrometer to detect the test liquid to obtain detection data; The detection data is processed, and then metabolites are identified to obtain the plasma metabolic profile.

8. The TCM constitution classification method based on plasma metabolic profile according to claim 1, characterized in that: According to the following formula [1], the metabolic constitution raw scores of the 9 constitutions are converted respectively to obtain the conversion scores of the 9 constitutions; Formula [1]: Conversion score = [(original score - number of items) / (number of items × 4)] × 100.

9. The TCM constitution classification method based on plasma metabolic profile according to claim 1, characterized in that: The step of determining the TCM constitution classification according to the conversion score specifically includes: determining the TCM constitution classification according to the conversion score and the criteria for determining a balanced constitution and an unbalanced constitution.

Images