Multi-modal parameter analysis-based traditional Chinese medicine qi stagnation syndrome auxiliary diagnosis method, medium and equipment

Abstract

The invention discloses a multi-modal parameter analysis-based traditional Chinese medicine qi stagnation syndrome auxiliary diagnosis method, medium and equipment, and the method comprises the steps: collecting four diagnosis information of a user, carrying out the weighted calculation of qi stagnation syndrome element points, and dividing a qi stagnation syndrome group and a non-qi stagnation syndrome group; an expiration sample is collected and analyzed to obtain an expiration smell spectrum; the method comprises the following steps: detecting a condensate metabolite by adopting gas chromatography-time-of-flight mass spectrometry, and carrying out data preprocessing to obtain standard metabolite characteristic peak intensity; discriminant analysis is performed through an orthogonal partial least square method to screen a differential metabolite group, and a target metabolite group is determined in combination with odor spectrum characteristics; standardizing and splicing odor characteristics and metabolite characteristics into a combined characteristic matrix, and constructing a machine learning model; and finally, the qi stagnation syndrome probability and confidence of the patient are output through the model. The model is trained in combination with multi-modal data, the probability that the patient suffers from the qi stagnation syndrome is diagnosed in an auxiliary mode by analyzing the differential metabolites, and the method is suitable for traditional Chinese medicine clinical auxiliary diagnosis.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of smart medical treatment, and in particular to a TCM qi stagnation syndrome auxiliary diagnosis method based on multi-modal parameter analysis, a medium and an equipment. BACKGROUND

[0002] At present, the detection technology of Western medicine diseases mainly relies on technologies such as mass spectrometry (GC-MS) and gas chromatography (GC), and a disease risk assessment model is constructed by analyzing volatile organic compounds (VOCs) or metabolites in exhaled breath. However, these technologies have the following limitations:

[0003] (1) Disconnected from TCM theory: Existing research mainly focuses on biomarker screening for Western medicine diseases (such as lung cancer and diabetes), and lacks correlation analysis of TCM syndromes (such as qi stagnation syndrome), which cannot meet the needs of TCM syndrome differentiation and treatment.

[0004] (2) Single detection dimension: Existing breath analysis technology mainly optimizes detection efficiency (such as improving sensor sensitivity or shortening sampling time), but relies only on single modal data (such as VOCs or metabolite concentration), resulting in insufficient diagnostic specificity. SUMMARY

[0005] In view of the above problems, the present application provides a TCM qi stagnation syndrome auxiliary diagnosis method based on multi-modal parameter analysis to solve the technical problem that the biomarker screening method involved in the prior art cannot meet the needs of TCM syndrome differentiation and treatment, resulting in poor accuracy of qi stagnation syndrome auxiliary diagnosis.

[0006] To solve the above problems, in a first aspect, a TCM qi stagnation syndrome auxiliary diagnosis method based on multi-modal parameter analysis is provided, which comprises:

[0007] S1: Collecting four diagnostic information of a plurality of users, performing weighted calculation on the four diagnostic information to obtain qi stagnation syndrome element integral, and dividing the users into qi stagnation syndrome group and non qi stagnation syndrome group according to the qi stagnation syndrome element integral;

[0008] S2: Collecting the oral exhaled breath of the user using a single valve odorless aluminum foil composite film gas sampling bag, and continuously collecting the oral exhaled breath condensate sample of the user for several minutes through an exhaled breath condensate collector;

[0009] S3: Analyzing the oral exhaled breath by a gas molecule recognition device comprising a metal oxide semiconductor sensor array to obtain odor spectrum data, extracting odor spectrum features of each sensor channel from the odor spectrum data, and the odor spectrum features include amplitude feature T and slope feature K;

[0010] S4: performing gas chromatography-time of flight mass spectrometry detection on the condensate sample, and obtaining standardized metabolite characteristic peak intensity data after peak alignment and normalization processing on the raw data;

[0011] S5: screening a first differential metabolite group related to the qi stagnation syndrome from the standardized metabolite characteristic peak intensity data by using orthogonal partial least squares discriminant analysis;

[0012] S6: determining a target metabolite group significantly related to the odor spectrum characteristics from the first differential metabolite group by correlation analysis;

[0013] S7: performing Z-score standardization on the odor spectrum characteristics and the characteristic peak intensity of the target metabolite group, and splicing the two characteristics into a joint feature matrix;

[0014] S8: dividing the training set and the test set at a predetermined ratio based on the joint feature matrix, training the model based on the training set and the test set, and obtaining the trained model;

[0015] S9: obtaining the breath sample data of the patient to be tested, and inputting the breath sample data of the patient to be tested into the trained model;

[0016] S10: outputting a diagnosis report containing the probability of the patient suffering from the qi stagnation syndrome of traditional Chinese medicine and the confidence, wherein the confidence is calculated based on the classification probability and the distance from the decision boundary output by the model.

[0017] Further, step S1 comprises:

[0018] converting the four diagnostic information into structured data and eliminating outliers from the structured data;

[0019] obtaining an initial list of symptom weights, dynamically adjusting the initial weight values of each symptom according to the symptom type combination in the current structured data, and obtaining a weight correction value;

[0020] performing weighted calculation on the structured data according to the weight correction value to obtain qi stagnation syndrome element scores;

[0021] dividing the qi stagnation syndrome element scores greater than a first threshold value into a qi stagnation syndrome group, and dividing the qi stagnation syndrome element scores less than the first threshold value into a non qi stagnation syndrome group.

[0022] Further, continuously collecting the oral breath condensate samples of the user for several minutes by the exhaled breath condensate collector comprises:

[0023] maintaining the temperature of the exhaled breath condensate collector at [-5, -2] degrees Celsius, continuously collecting the condensate samples of the oral breath of the user for [3-8] minutes, and storing the collected condensate samples in a refrigerated environment.

[0024] Further, step S5 comprises:

[0025] Input the normalized metabolite feature peak intensity data, with the qi stagnation syndrome grouping as the classification variable;

[0026] Calculate the variable importance projection value VIP of each metabolite, and screen out metabolites with a VIP value greater than 1 as potential markers;

[0027] In combination with the VIP value and statistical test, screen out metabolites that are significantly up-regulated or down-regulated in the qi stagnation syndrome group compared with the non-qi stagnation syndrome group, to obtain a first group of differential metabolites.

[0028] Further, step S6 comprises:

[0029] Calculate the Spearman correlation coefficient for each differential metabolite in the first group of differential metabolites of the qi stagnation syndrome group and the odor spectrum feature, respectively;

[0030] According to the Spearman correlation coefficient, at least one metabolite that is significantly correlated with the amplitude feature T or the slope feature K is included in the target metabolite group;

[0031] The target metabolite group contains at least 8 of pyruvic acid, isobutylamine, maleimide, glycinamide, putrescine, beta-methyl aspartic acid, carvone, behenic acid, and N-acetylneuraminic acid.

[0032] Further, step S7 comprises:

[0033] Respectively, the odor spectrum feature and the feature peak intensity of the target metabolite group are subjected to Z-score standardization to obtain the standard odor spectrum feature and the feature peak intensity of the standard target metabolite group;

[0034] According to the sample ID, the standard odor spectrum feature and the feature peak intensity of the standard target metabolite group are subjected to feature alignment and horizontal splicing to obtain the joint feature matrix.

[0035] Further, step S8 comprises:

[0036] Divide the joint feature matrix into a training set and a test set in a ratio of 7:3;

[0037] Use the support vector machine algorithm to construct a classification model, and select a linear kernel or a radial basis function kernel for training according to the data characteristics;

[0038] Optimize the model parameters by maximizing the distance of the classification boundary, and use correlation analysis and principal component analysis to reduce the dimension of the high-dimensional features, and eliminate low-correlation features to improve the performance of the model, to obtain the trained model.

[0039] Further, the target metabolite group is determined by a pathway enrichment analysis method, specifically comprising:

[0040] Performing pathway enrichment analysis on the first differential metabolite group, and screening metabolites significantly enriched in energy metabolism, amino acid metabolism or fatty acid metabolism pathway;

[0041] Retaining metabolites with a difference in expression greater than a preset multiple between the qi stagnation syndrome group and the non qi stagnation syndrome group;

[0042] Based on KEGG pathway topology analysis, selecting metabolites involved in more than 2 related pathways or located at key nodes of the pathways;

[0043] Combining the pathogenesis characteristics of qi stagnation syndrome in traditional Chinese medicine, determine the final target metabolite group.

[0044] In a second aspect, the present application also provides a computer readable storage medium, which stores computer program instructions, the computer program instructions are executed by a processor to implement the method of the first aspect.

[0045] In a third aspect, the present application also provides an electronic device, comprising a memory and a processor, the memory is used to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the processor to implement the method of the first aspect.

[0046] Distinguish from prior art, the above scheme provides a traditional Chinese medicine qi stagnation syndrome auxiliary diagnosis method, medium and equipment of multi-modal parameter analysis, by fusing traditional Chinese medicine four diagnostic quantitative data, breath odor spectrum characteristics and metabolic markers, realizing the auxiliary diagnosis of qi stagnation syndrome. The method specifically comprises: collecting four diagnostic information of users and calculating qi stagnation syndrome element integral by weighting, dividing qi stagnation syndrome group and non qi stagnation syndrome group; collecting breath samples, analyzing breath odor spectrum by using metal oxide semiconductor sensor array, extracting amplitude feature T and slope feature K; adopting gas chromatography-time of flight mass spectrometry to detect condensate metabolites, obtaining standardized metabolite characteristic peak intensity after data preprocessing; screening differential metabolite group by orthogonal partial least squares discriminant analysis, and determining target metabolite group combined with odor spectrum characteristics; standardizing and splicing odor characteristics and metabolite characteristics into joint feature matrix, and constructing machine learning model; finally outputting qi stagnation syndrome probability and confidence of patients through the model. The present application trains the model combined with multi-modal data, assists in diagnosing the probability of patients suffering from qi stagnation syndrome by analyzing differential metabolites, and is suitable for traditional Chinese medicine clinical auxiliary diagnosis.

[0047] The above summary related to the invention is only a summary of the technical solutions of the present application, in order to enable those skilled in the art to more clearly understand the technical solutions of the present application, and then can be implemented according to the content of the description and the drawings, and in order to let the above-mentioned purpose and other purposes, characteristics and advantages of the present application can be more easily understood, the following is described in combination with the specific embodiments of the present application and the drawings. BRIEF DESCRIPTION OF DRAWINGS

[0048] The drawings are only used to show the principles, implementation modes, applications, characteristics and effects of the specific embodiments of the present application and other related contents, and cannot be considered as a limitation of the present application.

[0049] In the drawings of the specification:

[0050] Figure 1 The flow chart of the method for assisting in diagnosing the qi stagnation syndrome of traditional Chinese medicine by multi-modal parameter analysis related to the first exemplary embodiment of the present application;

[0051] Figure 2 The flow chart of the method for assisting in diagnosing the qi stagnation syndrome of traditional Chinese medicine by multi-modal parameter analysis related to the second exemplary embodiment of the present application;

[0052] Figure 3 The flow chart of the method for assisting in diagnosing the qi stagnation syndrome of traditional Chinese medicine by multi-modal parameter analysis related to the third exemplary embodiment of the present application;

[0053] Figure 4 The flow chart of the method for assisting in diagnosing the qi stagnation syndrome of traditional Chinese medicine by multi-modal parameter analysis related to the fourth exemplary embodiment of the present application;

[0054] Figure 5 The flow chart of the method for assisting in diagnosing the qi stagnation syndrome of traditional Chinese medicine by multi-modal parameter analysis related to the fifth exemplary embodiment of the present application;

[0055] Figure 6 The flow chart of the method for assisting in diagnosing the qi stagnation syndrome of traditional Chinese medicine by multi-modal parameter analysis related to the sixth exemplary embodiment of the present application;

[0056] Figure 7 The flow chart of the method for assisting in diagnosing the qi stagnation syndrome of traditional Chinese medicine by multi-modal parameter analysis related to the sixth exemplary embodiment of the present application;

[0057] Figure 8 The schematic diagram of the oral breath odor spectrum related to an exemplary embodiment of the present application;

[0058] Figure 9 The data schematic diagram of the amplitude feature and the slope feature of the oral breath odor spectrum related to an exemplary embodiment of the present application;

[0059] Figure 10A schematic diagram of the result of the orthogonal partial least squares discriminant analysis of the qi stagnation syndrome group and the non qi stagnation syndrome group involved in an exemplary embodiment of the present application;

[0060] Figure 11 A schematic diagram of the condensate sample and the odor spectrum data of the qi stagnation group involved in an exemplary embodiment of the present application;

[0061] Figure 12 A schematic diagram of the correlation analysis of the slope feature in the condensate sample and the odor spectrum feature of the qi stagnation group involved in an exemplary embodiment of the present application;

[0062] Figure 13 A schematic diagram of the correlation analysis of the amplitude feature in the condensate sample and the odor spectrum feature of the qi stagnation group involved in an exemplary embodiment of the present application;

[0063] Figure 14 A correlation heat map of the slope feature in the condensate sample and the odor spectrum feature of the qi stagnation group involved in an exemplary embodiment of the present application;

[0064] Figure 15 A correlation heat map of the amplitude feature in the condensate sample and the odor spectrum feature of the qi stagnation group involved in an exemplary embodiment of the present application;

[0065] Figure 16 A schematic diagram of the target metabolite group significantly correlated with the odor spectrum feature of the qi stagnation group involved in an exemplary embodiment of the present application;

[0066] Figure 17 A schematic diagram of the ROC curve of the odor spectrum identification model and the joint identification model involved in an exemplary embodiment of the present application;

[0067] Figure 18 A schematic diagram of the module of the electronic device involved in the present application;

[0068] The reference signs involved in the above-mentioned figures are explained as follows:

[0069] 10, electronic device;

[0070] 101, processor;

[0071] 102, storage medium. DETAILED DESCRIPTION

[0072] In order to explain the possible application scenarios, technical principles, specific implementable schemes, and the purposes and effects that can be achieved of the present application in detail, the following will be explained in detail in combination with the specific embodiments listed and the accompanying drawings. The embodiments described in this paper are only used to more clearly explain the technical solutions of the present application, and therefore only serve as examples, and cannot limit the protection scope of the present application.

[0073] In this document, the term "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The term "embodiment" appearing in various places throughout the specification does not necessarily refer to the same embodiment, nor does it specifically limit its independence or connection with other embodiments. In principle, in this invention, as long as there are no technical contradictions or conflicts, the technical features mentioned in each embodiment can be combined in any way to form corresponding implementable technical solutions.

[0074] Unless otherwise defined, the technical terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains; the use of related terms herein is merely for the purpose of describing particular embodiments and is not intended to limit the invention.

[0075] In the description of this invention, the term "and / or" is used to describe the logical relationship between objects, indicating that three relationships can exist. For example, A and / or B means: A exists, B exists, and A and B exist simultaneously. Additionally, the character " / " generally indicates that the preceding and following objects have an "or" logical relationship.

[0076] In this invention, terms such as “first” and “second” are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any actual quantity, hierarchy, or order between these entities or operations.

[0077] Without further limitations, the use of terms such as “comprising,” “including,” “having,” or other similar open-ended expressions in this invention is intended to cover non-exclusive inclusion, which does not exclude the presence of additional elements in a process, method, or product that includes the stated elements, such that a process, method, or product that includes a list of elements may include not only those defined elements but also other elements not expressly listed, or elements inherent to such a process, method, or product.

[0078] In this invention, expressions such as "greater than", "less than", and "exceeding" are understood to exclude the stated number; expressions such as "above", "below", and "within" are understood to include the stated number. Furthermore, in the description of the embodiments of this invention, "multiple" means two or more (including two), and similar expressions related to "multiple" are also understood in this way, such as "multiple groups" and "multiple times", unless otherwise explicitly specified.

[0079] In the first aspect, such as Figure 1 As shown, this application provides a multimodal parameter analysis-based auxiliary diagnostic method for Qi stagnation syndrome in Traditional Chinese Medicine, the method comprising:

[0080] S1: Collecting four diagnostic information of multiple users, performing weighted calculation on the four diagnostic information to obtain qi stagnation syndrome element integral, and dividing the users into qi stagnation syndrome group and non qi stagnation syndrome group according to the qi stagnation syndrome element integral.

[0081] In step S1, the TCM collects the patient's condition information through four means of inspection, auscultation, interrogation and palpation. Inspection observes the patient's spirit, color, shape, posture and tongue appearance; auscultation includes listening to sound and smell; interrogation inquires about symptoms, medical history, living habits, etc.; and palpation mainly includes pulse and palpation. Therefore, the four diagnostic information can be presented in the form of tongue coating image, audio data, pulse signal and symptom text data.

[0082] Qi stagnation syndrome element integral refers to the numerical value reflecting the degree of qi stagnation syndrome of the patient obtained by quantitatively processing the four diagnostic information according to the TCM syndrome element differentiation theory.

[0083] S2: Collecting the user's oral exhalation using a single valve odorless aluminum foil composite film gas sampling bag, and continuously collecting the user's oral exhalation condensate sample for several minutes through an exhaled gas condensate collector.

[0084] In step S2, the user's oral exhalation is collected using a single valve odorless aluminum foil composite film gas sampling bag to ensure that the gas is pure and free of interference; the oral exhalation condensate sample is continuously collected for several minutes at a specific temperature through an exhaled gas condensate collector, and is stored at low temperature to maintain sample stability, providing a sample basis for subsequent analysis.

[0085] S3: Analyzing the oral exhalation by a gas molecule recognition device comprising a metal oxide semiconductor sensor array to obtain odor spectrum data, extracting odor spectrum features of each sensor channel from the odor spectrum data, the odor spectrum features including amplitude feature T and slope feature K.

[0086] In step S3, the oral exhalation is analyzed by a gas molecule recognition device comprising a metal oxide semiconductor sensor array. Different gases interact with the sensitive film on the surface of the sensor, producing changes in electrical signals to form odor spectrum data. The amplitude feature T and the slope feature K of each sensor channel are extracted from the data. The calculation formulas of the two features are as follows:

[0087] T = y(max) - y(min);

[0088] K = (y(max) - y(min)) / (t(max) - t(min));

[0089] Where y(max) and y(min) are the maximum and minimum values of the electrical signals converted into odor spectrum data, and t(max) - t(min) represents the time span.

[0090] The amplitude characteristic T and the slope characteristic K reflect changes in gas composition and concentration, and are used to characterize the odor properties of breath.

[0091] S4: Gas chromatography-time-of-flight mass spectrometry was performed on the condensate sample. After peak alignment and normalization of the raw data, standardized metabolite characteristic peak intensity data were obtained.

[0092] In step S4, the condensate sample is analyzed by gas chromatography-time-of-flight mass spectrometry (GC-TOF-MS). GC separates the complex mixture into individual components, while TOF-MS determines the mass-to-charge ratio based on the ion's time of flight. The raw data is then processed through peak alignment and normalization to eliminate instrument errors and differences between samples, obtaining standardized peak intensity data for the metabolites. This provides accurate data for subsequent screening of differentially expressed metabolites.

[0093] S5: Orthogonal partial least squares discriminant analysis was used to screen the first differential metabolite group related to Qi stagnation syndrome from the standardized metabolite characteristic peak intensity data.

[0094] S6: Identify the target metabolite group that is significantly correlated with the odor spectrum characteristics of the first differential metabolite group through correlation analysis.

[0095] Preferably, the target metabolite group comprises at least eight of the following: pyruvate, isobutylamine, maleimide, glycine, putrescine, β-methylaspartic acid, carvone, behenic acid, and N-acetylneuraminic acid.

[0096] S7: The odor spectrum features and the characteristic peak intensities of the target metabolite group are Z-score normalized, and the two features are concatenated into a joint feature matrix;

[0097] S8: Based on the joint feature matrix, divide the training set and the test set by a predetermined ratio, train the model based on the training set and the test set, and obtain the trained model.

[0098] S9: Obtain the exhalation sample data of the patient to be tested, and input the exhalation sample data of the patient to be tested into the trained model;

[0099] S10: Output a diagnostic report containing the probability and confidence level of the patient having the TCM Qi stagnation syndrome, wherein the confidence level is calculated based on the classification probability output by the model and the distance to the decision boundary.

[0100] The above scheme integrates traditional Chinese medicine theory with modern technology, mines the characteristics of qi stagnation syndrome from multimodal data, improves the accuracy and objectivity of diagnosis, provides a new approach for the auxiliary diagnosis of qi stagnation syndrome in traditional Chinese medicine, and helps to improve the accuracy of the judgment of qi stagnation syndrome in traditional Chinese medicine.

[0101] like Figure 2 As shown, in some embodiments, step S1 includes:

[0102] Step S201: converting four diagnostic information into structured data, and eliminating outliers from the structured data.

[0103] Step S202: obtaining a symptom weight initial list, dynamically adjusting the initial weight of each symptom according to the symptom type combination in the current structured data, and obtaining a weight correction value;

[0104] Step S203: weighting calculation on the structured data according to the weight correction value to obtain the qi stagnation syndrome element score;

[0105] Step S204: dividing the qi stagnation syndrome group into the qi stagnation syndrome group if the qi stagnation syndrome element score is greater than the first threshold value, and dividing the qi stagnation syndrome group into the non qi stagnation syndrome group if the qi stagnation syndrome element score is less than the first threshold value.

[0106] In step S201, the structured data refers to the four diagnostic information which is arranged according to a specific format and standard, so that it has a clear data structure and meaning, which is convenient for subsequent processing and analysis. The outliers refer to the data points in the four diagnostic information data which are greatly different from the overall data characteristics, which may be caused by measurement error, recording error or special individual situation.

[0107] In step S202, the symptom weight initial list refers to the weight value list preset for various symptoms according to the theory of traditional Chinese medicine and clinical experience, which reflects the relative importance of different symptoms to the diagnosis of qi stagnation syndrome.

[0108] The above scheme converts the four diagnostic information into structured data, such as the complexion and tongue appearance of inspection, the respiratory sound and smell of auscultation, the symptom description and medical history of interrogation, and the pulse of palpation. The information is recorded in a unified format. The outliers are identified and eliminated by data statistical methods (such as box plot method and Z-score method) to ensure data reliability. According to the symptom type combination in the current structured data, such as the simultaneous occurrence of chest and hypochondrium distending pain, belching and emotional depression, the dynamic adjustment of the initial weight of each symptom is considered in view of the synergistic effect of these symptoms on the diagnosis of qi stagnation syndrome. For example, if a typical liver stagnation and qi stagnation symptom combination occurs, the weight of the related symptoms can be appropriately increased to obtain the weight correction value. Finally, the structured data is weighted calculated according to the weight correction value, such as multiplying the scores of each symptom by the corresponding weight and then accumulating, to obtain the qi stagnation syndrome element score. By dynamically adjusting the symptom weight, the influence of different symptom combinations on the diagnosis of qi stagnation syndrome is more flexible and accurate, which improves the accuracy of the score calculation and provides a more reliable basis for subsequent sample grouping.

[0109] In some embodiments, continuously collecting the oral exhaled breath condensate sample of the user for several minutes by the exhaled breath condensate collector includes: maintaining the temperature of the exhaled breath condensate collector at [-5, -2] degrees Celsius, continuously collecting the condensate sample of the user's oral exhaled breath for [3-8] minutes, and storing the collected condensate sample in a refrigerated environment.

[0110] In this embodiment, the exhaled breath condensate collector refers to a device specially used for collecting the condensate in human exhaled breath, which can condense the water vapor in the exhaled breath into liquid for collection.

[0111] By maintaining the temperature of the exhaled breath condensate collector at [-5, -2] degrees Celsius, the water vapor in the exhaled breath is more likely to condense in this low-temperature environment. The condensate sample of the user's mouth exhalation is collected for [3-8] minutes to ensure that a sufficient amount of sample is collected for subsequent analysis. After collection, the condensate sample is stored in a refrigerated environment (such as a -80°C refrigerator) to prevent degradation or chemical changes of metabolites in the sample and maintain sample stability.

[0112] By precisely controlling the collection temperature and time, the quality and quantity of the sample collection are ensured; low-temperature preservation effectively protects the metabolites in the sample, providing reliable samples for subsequent gas chromatography-time-of-flight mass spectrometry detection and improving the accuracy of the detection results.

[0113] In some embodiments, as shown in FIG. 5, step S5 includes: Figure 3

[0114] S301: Input the standardized metabolite characteristic peak intensity data, with the qi stagnation syndrome grouping as the classification variable;

[0115] S302: Calculate the variable importance projection value VIP of each metabolite, and screen out the metabolites with a VIP value greater than 1 as potential markers;

[0116] S303: Combine the VIP value and statistical test to screen out the metabolites that are significantly up-regulated or down-regulated in the qi stagnation syndrome group compared with the non-qi stagnation syndrome group, to obtain the first differential metabolite group.

[0117] In step S302, the variable importance projection value (VIP) refers to an index in the orthogonal partial least squares discriminant analysis that measures the contribution of each variable to the classification of the model. The larger the VIP value, the higher the importance of the variable in distinguishing different groups. The potential marker refers to a metabolite that has the potential to distinguish different physiological or pathological states in metabolite analysis.

[0118] In step S5, the standardized metabolite characteristic peak intensity data is input into the analysis model, with the qi stagnation syndrome grouping (qi stagnation syndrome group and non-qi stagnation syndrome group) as the classification variable. By calculating the VIP value of each metabolite, the metabolites with a VIP value greater than 1 are screened out as potential markers that play an important role in distinguishing qi stagnation syndrome from non-qi stagnation syndrome. Further combined with statistical test (such as t-test, analysis of variance), it is judged whether the expression difference of these potential markers between the qi stagnation syndrome group and the non-qi stagnation syndrome group is significant, and the metabolites that are significantly up-regulated or down-regulated in the qi stagnation syndrome group compared with the non-qi stagnation syndrome group are screened out, to obtain the first differential metabolite group.​

[0119] The above scheme effectively screens differential metabolites closely related to Qi stagnation syndrome by VIP value combined with statistical test, reduces the number of metabolites for subsequent analysis, focuses on key metabolites, and improves research efficiency and accuracy.

[0120] In some embodiments, as shown in Figure 4 Step S6 includes:

[0121] S401: Calculate the Spearman correlation coefficient for each differential metabolite in the first differential metabolite group of the Qi stagnation syndrome group and the odor profile feature respectively.

[0122] S402: According to the Spearman correlation coefficient, at least one metabolite significantly related to the amplitude feature T or the slope feature K is included in the target metabolite group.

[0123] In this embodiment, the Spearman correlation coefficient is a statistical index for measuring the strength and direction of monotonic relationship between two variables, with a value range of [-1, 1]. The closer the absolute value is to 1, the stronger the correlation.

[0124] In step S6, for each differential metabolite in the first differential metabolite group of the Qi stagnation syndrome group, the Spearman correlation coefficient is calculated with the odor profile feature (amplitude feature T and slope feature K) respectively. If the absolute value of the correlation coefficient is greater than a certain threshold (such as 0.5) and the corresponding P value is less than the set significance level (such as 0.05), it is considered that the metabolite is significantly related to the corresponding odor profile feature, and it is included in the target metabolite group.

[0125] The above scheme establishes the relationship between metabolites and odor profile features through correlation analysis, provides a basis for multi-modal data fusion, and helps to reveal the biological characteristics of Qi stagnation syndrome from different angles and improve the performance of the diagnostic model.

[0126] In some embodiments, as shown in Figure 5 Step S7 includes:

[0127] S501: Perform Z-score standardization on the odor profile feature and the feature peak intensity of the target metabolite group respectively to obtain the standard odor profile feature and the feature peak intensity of the standard target metabolite group.

[0128] S502: Perform feature alignment and horizontal splicing on the standard odor profile feature and the feature peak intensity of the standard target metabolite group according to the sample ID to obtain the joint feature matrix.

[0129] In the embodiment, Z-score standardization is a data standardization method that makes data have the characteristics of standard normal distribution with a mean of 0 and a standard deviation of 1 by subtracting the mean and dividing by the standard deviation of the original data, eliminating the differences in the orders of magnitude of different characteristic quantities.

[0130] Feature alignment refers to arranging different source feature data in the same sample order according to the sample ID to ensure the accuracy of subsequent feature splicing. Horizontal splicing refers to merging different feature data that have been standardized and feature-aligned in the column direction to form a joint feature matrix containing multiple feature information.

[0131] In step S7, the odor profile features and the feature peak intensities of the target metabolite group are respectively subjected to Z-score standardization. For example, for a feature value x in the odor profile feature, the standardized value x = (x - μ) / σ, where μ is the mean of all sample values of the feature, and σ is the standard deviation. The feature peak intensity data of the target metabolite group are processed in the same way to obtain the standard odor profile features and the feature peak intensity features of the standard target metabolite group. Then, the two sets of standardized features are feature-aligned according to the sample ID, so that different feature data of the same patient are matched and matched, and finally horizontal splicing is performed to form a joint feature matrix, which integrates multi-modal data in one data structure.

[0132] The above scheme makes different features comparable through Z-score standardization, and realizes multi-modal data fusion through feature alignment and horizontal splicing, providing rich and unified format data for subsequent model training, which helps to improve the model learning effect and diagnostic accuracy.

[0133] In some embodiments, as shown in FIG. 7, step S7 includes: Figure 6

[0134] S601: dividing the joint feature matrix into a training set and a test set in a ratio of 7:3;

[0135] S602: constructing a classification model using a support vector machine algorithm, and selecting a linear kernel or a radial basis function kernel for training according to the data characteristics;

[0136] S603: optimizing the model parameters by maximizing the distance of the classification boundary, and performing dimensionality reduction on the high-dimensional features using correlation analysis and principal component analysis to remove low-correlation features to improve the performance of the model, to obtain a trained model.

[0137] ​In step S602, the support vector machine (SVM) algorithm is a classification algorithm based on statistical learning theory, which separates different classes of samples as much as possible by finding an optimal classification hyperplane. Linear kernel is a simple kernel function in SVM, which assumes that the data is linearly separable in the original feature space and directly classifies in the original feature space. Radial basis function kernel (RBF kernel) is a non-linear kernel function, which maps the data to a high-dimensional feature space, so that the data which is linearly inseparable in low-dimensional space becomes linearly separable in high-dimensional space.

[0138] In step S603, principal component analysis (PCA) is a data dimensionality reduction technique, which converts the original high-dimensional data into a set of new orthogonal features (principal components) by linear transformation, retains the main information of the data while reducing the dimension, and reduces the data redundancy and computational complexity.

[0139] In step S8, the joint feature matrix is randomly divided into training set and test set in the ratio of 7:3. Support vector machine algorithm is used to construct classification model, and linear kernel or radial basis function kernel is selected for training according to data characteristics (such as whether the data distribution is linearly separable). In the training process, the model parameters are optimized by maximizing the distance of the classification boundary (i.e. finding the optimal classification hyperplane), so that the model has good discrimination ability for different classes of samples. At the same time, correlation analysis is used to remove features with low correlation with the target class, reducing noise interference; principal component analysis is used to reduce the dimension of high-dimensional features, retaining the main feature information and reducing the complexity of the model, and finally obtaining the trained model.

[0140] The above scheme can ensure the generalization ability of the model by reasonably dividing the training set and the test set, improve the classification accuracy of the model by selecting appropriate kernel function and optimization method, reduce the calculation amount and overfitting risk by dimensionality reduction processing, and improve the performance and efficiency of the model.

[0141] In some embodiments, the target metabolite group is determined by pathway enrichment analysis method, such as Figure 7 As shown, it specifically includes:

[0142] S701: performing pathway enrichment analysis on the first differential metabolite group, and screening metabolites significantly enriched in energy metabolism, amino acid metabolism or fatty acid metabolism pathway;

[0143] S702: retaining metabolites with expression difference fold greater than a preset fold between the qi stagnation syndrome group and the non qi stagnation syndrome group;

[0144] S703: based on KEGG pathway topology analysis, selecting metabolites involved in more than 2 related pathways or located at key nodes of the pathways;

[0145] S704: combining the pathogenesis characteristics of qi stagnation syndrome in traditional Chinese medicine, and determining the final target metabolite group.

[0146] In step S701, pathway enrichment analysis is a bioinformatics analysis method used to determine the degree of enrichment of a group of genes or metabolites in known biological pathways, revealing the main biological processes and signaling pathways in which they participate.

[0147] In step S703, KEGG pathway topology analysis is based on the Kyoto Encyclopedia of Genes and Genomes (KEGG) database to analyze the structure and function of biological pathways, including the interrelationships between nodes (genes, metabolites, etc.) in the pathway and the direction of metabolic flow.

[0148] In this embodiment, when determining the target metabolite group, pathway enrichment analysis is first performed on the first differentially expressed metabolite group. Using relevant software (such as MetaboAnalyst) and databases (such as KEGG and Reactome), metabolites significantly enriched in energy metabolism, amino acid metabolism, or fatty acid metabolism pathways are screened, as these pathways are closely related to the pathogenesis of Qi stagnation syndrome in Traditional Chinese Medicine (TCM). Then, metabolites whose expression levels differ from the non-Qi stagnation syndrome group by a fold greater than a preset fold (e.g., 2-fold) are retained, highlighting the significantly different metabolites. Based on KEGG pathway topology analysis, metabolites involved in more than two related pathways or located at key nodes in these pathways are selected, as these metabolites play an important regulatory role in pathway function. Finally, combined with the characteristics of the TCM Qi stagnation syndrome pathogenesis, such as organ dysfunction caused by Qi stagnation, the final target metabolite group is further screened and determined.

[0149] The above scheme uses pathway enrichment analysis and topological analysis to gain a deeper understanding of the biological mechanism of Qi stagnation syndrome at the metabolic pathway level. It combines the characteristics of TCM pathogenesis to screen metabolites, making the target metabolite group more biologically significant and clinically relevant, and providing more accurate biomarkers for the diagnosis of Qi stagnation syndrome.

[0150] like Figures 8-17 As shown, the auxiliary diagnostic method for qi stagnation syndrome involved in this application will be further explained below with reference to specific experimental cases:

[0151] As shown in Table 1 below, after weighted calculation based on the collected user information from the four diagnostic methods, the user group was divided into a qi stagnation group and a non-qi stagnation group based on the comparison results of the obtained qi stagnation syndrome points and thresholds. A total of 90 cases were included in the qi stagnation group and 34 cases in the non-qi stagnation group.

[0152] Table 1. Gender and age composition of the Qi stagnation group and the non-Qi stagnation group

[0153]

[0154] In practical application, the gas molecule recognition device selected can be an electronic nose (model GISXM-MQWA01), which can quickly obtain the breath odor spectrum of the oral cavity and complete detection within 20 minutes. The sensor array of the electronic nose contains 9 sensors, which can output 9 different color response curves (A, B, C, D, E, G, H, I, J) to form the odor spectrum data.

[0155] The working environment temperature of the electronic nose is controlled at 10-35℃, and the relative humidity is less than 80%. Before the experiment, the parameters of the electronic nose are adjusted to be stable, the gas bag is connected to the gas chamber interface, the gas flows into the sensor at a flow rate of 500mL / min, and the operation time is 120 seconds, and the sampling interval is 600 seconds. The electronic nose outputs 9 response curves for representing the odor spectrum data, and the collected odor spectrum data are as shown in Figure 8

[0156] The odor spectrum has 9 response channels and a "T" channel, and a total of 9 response curves. The amplitude characteristics and slope characteristics of the response curves of each channel in the odor spectrum are extracted as the characteristic parameters of the odor spectrum in this study, the horizontal axis is the detection time point, and the vertical axis is the detection value y(1)…y(120). The schematic diagram of the characteristic parameters is as shown in Figure 9

[0157] The condensate samples are detected and analyzed by gas chromatography-time of flight mass spectrometry (GC-TOF-MS), and a series of analysis steps such as peak extraction, baseline correction, deconvolution, peak integration, and peak alignment are performed on the mass spectrum data using ChromaTOF software. Peaks with a detection rate of less than 50% or a relative standard deviation of more than 30% in the quality control sample are removed. From the original data, 1023 features are extracted, and after pretreatment, 1022 features are retained. Due to the large number of samples, the top 10 reference values are selected and some samples are displayed as shown in the following Table 2:

[0158] Table 2: Sample original data statistics

[0159]

[0160] Then, the mass spectrum data are logarithmically converted and formatted using SIMCA software, and the first principal component is analyzed by orthogonal partial least squares discriminant analysis (hereinafter referred to as "OPLS-DA analysis") modeling analysis. Based on the t-test P<0.05 and the variable projection importance VIP of the first principal component of the OPLS-DA model, the differential metabolites (VIP>1) are screened. Compared with the non-Qi stagnation group, a total of 17 first differential metabolite groups with larger differences are obtained (see Table 3 below). The OPLS-DA analysis shows that the difference between the Qi stagnation group and the non-Qi stagnation group is large, the repeatability within the group is good, and the discrimination degree is good, and the model has certain prediction ability, as shown in​​Figure 10

[0161] Table 3 Statistical table of the first difference metabolite group of the qi stagnation group compared with the non qi stagnation group

[0162]

[0163]

[0164] The correlation between the condensate difference metabolites of the qi stagnation group and the non qi stagnation group and the slope and amplitude of the 9 response channels of the odor spectrum of the qi stagnation group was evaluated by Spearman rank correlation coefficient, and P<0.05 was statistically significant, and the condensate difference metabolite group 2 with significant correlation with the slope and amplitude of the odor spectrum was obtained, and this part of the condensate difference metabolites was included in the model construction and verification, as shown in Figures 11-16

[0165] In the present application, the confusion matrix (CM) is used for model verification. The confusion matrix of a binary classification problem is defined by the true class (row name) and the predicted class (column name), which includes: true positive (TP), that is, correctly identifying samples that are actually positive; and false positive (FP), that is, mistakenly judging negative samples as positive samples; true negative (TN), that is, correctly identifying samples that are actually negative; and false negative (FN), that is, mistakenly judging positive samples as negative samples. Five evaluation indexes of the model are calculated by the CM model. The specific calculation formula is as follows:

[0166] (1) Accuracy: is a core index for measuring the effect of the model, which refers to the proportion of the number of samples correctly predicted by the model to the total number of samples, and the calculation formula of the accuracy is:

[0167] Accuracy=(TP+TN) / (TP+TN+FP+FN);

[0168] (2) Precision: the proportion of samples actually positive in the samples predicted by the model as positive, and the calculation formula of the precision is:

[0169] Precision=TP / (TP+FP);

[0170] (3) Sensitivity: true positive rate, the proportion of samples actually positive correctly identified by the model, and the calculation formula of the sensitivity is:

[0171] Sensitivity=TP / (TP+FN); ​​

[0172] (4) Specificity: True negative rate, which refers to the proportion of samples that are actually negative but are judged as negative. The specificity value is calculated using the following formula:

[0173] Specificity = TN / (TN+FP);

[0174] like Figure 17 As shown, the area under the ROC curve (AUC) is one of the most commonly used model evaluation metrics in data analysis. The ROC curve is plotted with the false positive rate ("1-specificity") on the x-axis and the true positive rate ("sensitivity") on the y-axis, at different classification thresholds, forming a curve. The model's performance is evaluated using the area under this curve, AUC, where an AUC value between 0.5 and 1 indicates that the classifier outperforms random guessing.

[0175] Figure 17 The results showed that the diagnostic performance of the combined identification model based on odor profiles and differential condensate metabolites was better than that of the identification model based on a single odor profile or condensate metabolite. The model also showed significant improvement in all evaluation indicators for the identification model of the gas stagnation group, with some improvement in other indicators (see Table 4 below).

[0176] Table 4 Evaluation Indicators for the Identification Model of Qi Stagnation Syndrome

[0177]

[0178] In a second aspect, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the TCM Qi stagnation syndrome auxiliary diagnostic method based on multimodal parameter analysis as described in the first aspect of the present invention.

[0179] The computer-readable storage medium may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory.

[0180] The non-volatile memory can be a Read Only Memory (ROM), a Programmable Read Only Memory (PROM), an Erasable Programmable Read Only Memory (EPROM), an Electrically Erasable Programmable Read Only Memory (EEPROM), a ferromagnetic random access memory (FRAM), a Flash Memory, a magnetic surface memory, an optical disk, or a Compact Disc Read Only Memory (CD ROM).

[0181] The volatile memory can be a Random Access Memory (RAM) used as an external cache. By way of example, and not limitation, many forms of RAM can be used, such as a Static Random Access Memory (SRAM), a Synchronous Static Random Access Memory (SSRAM), a Dynamic Random Access Memory (DRAM), a Synchronous Dynamic Random Access Memory (SDRAM), a Double Data Rate Synchronous Dynamic Random Access Memory (DDR SDRAM), an Enhanced Synchronous Dynamic Random Access Memory (ESDRAM), a Sync Link Dynamic Random Access Memory (SLDRAM), a Direct Rambus Random Access Memory (DRRAM). The computer readable storage medium described in the embodiments of the present application is intended to include these and any other suitable types of memory.

[0182] As Figure 18 shown in the third aspect, the present application provides an electronic device 10, comprising a processor 101 and a storage medium 102, wherein the storage medium 102 stores a computer program, and the computer program is executed by the processor 101 to realize the method for auxiliary diagnosis of TCM qi stagnation syndrome based on multi-modal parameter analysis according to the first aspect of the present application.

[0183] In some embodiments, the processor can be implemented by software, hardware, firmware or a combination thereof, and can use at least one of circuit, single or multiple application specific integrated circuits (ASIC), digital signal processors (DSP), digital signal processing devices (DSPD), programmable logic devices (PLD), field programmable gate arrays (FPGA), central processing units (CPU), controllers, microcontrollers or microprocessors, so that the processor can execute part or all steps or any combination of steps of the method for auxiliary diagnosis of TCM qi stagnation syndrome based on multi-modal parameter analysis according to various embodiments of the present application.

[0184] Finally, it should be noted that although the above embodiments have been described in the specification and drawings of the present application, the patent protection scope of the present application should not be limited. Any technical solutions obtained by replacing or modifying the equivalent structure or equivalent flow based on the essential concept of the present application, using the content described in the specification and drawings of the present application, and directly or indirectly implementing the technical solutions of the above embodiments in other related technical fields, are all included in the patent protection scope of the present application.

Claims

1. A multimodal parameter analysis-based auxiliary diagnostic method for Qi stagnation syndrome in Traditional Chinese Medicine, characterized in that, The method includes: S1: Collect the four diagnostic methods information of multiple users, perform weighted calculation on the four diagnostic methods information to obtain the qi stagnation syndrome element score, and divide users into qi stagnation syndrome group and non-qi stagnation syndrome group according to the qi stagnation syndrome element score. S2: Use a single-valve odorless aluminum foil composite membrane gas sampling bag to collect the user's breath, and continuously collect the user's breath condensate sample for several minutes using an exhaled condensate collector. S3: The breath is analyzed by a gas molecule recognition device containing a metal oxide semiconductor sensor array to obtain odor spectrum data. Odor spectrum features of each sensor channel are extracted from the odor spectrum data. The odor spectrum features include amplitude feature T and slope feature K. S4: Gas chromatography-time-of-flight mass spectrometry was performed on the condensate sample. After peak alignment and normalization of the raw data, standardized metabolite characteristic peak intensity data were obtained. S5: Orthogonal partial least squares discriminant analysis was used to screen the first differential metabolite group related to Qi stagnation syndrome from the standardized metabolite characteristic peak intensity data; S6: Identify the target metabolite group that is significantly correlated with the odor spectrum characteristics of the first differential metabolite group through correlation analysis; S7: The odor spectrum features and the characteristic peak intensities of the target metabolite group are Z-score normalized, and the two features are concatenated into a joint feature matrix; S8: Based on the joint feature matrix, divide the training set and the test set by a predetermined ratio, train the model based on the training set and the test set, and obtain the trained model. S9: Obtain the exhalation sample data of the patient to be tested, and input the exhalation sample data of the patient to be tested into the trained model; S10: Output a diagnostic report containing the probability and confidence level of the patient having the TCM Qi stagnation syndrome, wherein the confidence level is calculated based on the classification probability output by the model and the distance to the decision boundary.

2. The method for auxiliary diagnosis of Qi stagnation syndrome in Traditional Chinese Medicine using multimodal parameter analysis as described in claim 1, characterized in that, Step S1 includes: Transform the information from the four diagnostic methods into structured data and remove outliers from the structured data; Obtain an initial list of symptom weights, dynamically adjust the initial weight values ​​of each symptom based on the combination of symptom types in the current structured data, and obtain the weight correction values; The structured data is weighted according to the weight correction value to obtain the integral of the qi stagnation syndrome element; Those whose Qi stagnation syndrome element scores are greater than the first threshold are divided into the Qi stagnation syndrome group, and those whose Qi stagnation syndrome element scores are less than the first threshold are divided into the non-Qi stagnation syndrome group.

3. The method for auxiliary diagnosis of Qi stagnation syndrome in Traditional Chinese Medicine using multimodal parameter analysis as described in claim 1, characterized in that, A breath condensate sample is continuously collected from the user over several minutes using an exhaled breath condensate collector, including: The temperature of the exhaled condensate collector was maintained at [-5, -2] degrees Celsius, and condensate samples of the user's exhaled breath were collected continuously for [3-8] minutes. The collected condensate samples were then stored in a frozen environment.

4. The method for auxiliary diagnosis of Qi stagnation syndrome in Traditional Chinese Medicine using multimodal parameter analysis as described in claim 1, characterized in that, Step S5 includes: Input standardized metabolite characteristic peak intensity data, using Qi stagnation syndrome grouping as a classification variable; Calculate the variable importance projection value (VIP) for each metabolite, and screen out metabolites with a VIP value greater than 1 as potential biomarkers; By combining VIP values ​​and statistical tests, metabolites that were significantly upregulated or downregulated in the Qi stagnation syndrome group compared to the non-Qi stagnation syndrome group were screened to obtain the first differential metabolite group.

5. The method for auxiliary diagnosis of Qi stagnation syndrome in Traditional Chinese Medicine using multimodal parameter analysis as described in claim 1, characterized in that, Step S6 includes: Spearman correlation coefficients were calculated for each differential metabolite in the first differential metabolite group of the Qi stagnation syndrome group and the odor spectrum characteristics, respectively. Based on the Spearman correlation coefficient, at least one metabolite that is significantly correlated with the amplitude feature T or the slope feature K is included in the target metabolite group; The target metabolite group includes at least eight of the following: pyruvate, isobutylamine, maleimide, glycine, putrescine, β-methylaspartic acid, carvone, behenic acid, and N-acetylneuraminic acid.

6. The method for auxiliary diagnosis of Qi stagnation syndrome in Traditional Chinese Medicine using multimodal parameter analysis as described in claim 1, characterized in that, Step S7 includes: Z-score normalization was performed on the odor spectrum features and the characteristic peak intensities of the target metabolite group respectively to obtain standard odor spectrum features and standard characteristic peak intensity features of the target metabolite group; The standard odor spectrum features and the characteristic peak intensity features of the standard target metabolite group are aligned and horizontally stitched together according to the sample ID to obtain the joint feature matrix.

7. The method for auxiliary diagnosis of Qi stagnation syndrome in Traditional Chinese Medicine using multimodal parameter analysis as described in claim 1, characterized in that, Step S8 includes: The joint feature matrix is ​​divided into a training set and a test set in a 7:3 ratio; A classification model is constructed using the support vector machine algorithm, and a linear kernel or radial basis function kernel is selected for training based on the data characteristics; The model parameters are optimized by maximizing the classification boundary distance, and high-dimensional features are reduced using correlation analysis and principal component analysis to remove low-correlation features, thereby improving model performance and obtaining a well-trained model.

8. The method for auxiliary diagnosis of Qi stagnation syndrome in Traditional Chinese Medicine using multimodal parameter analysis as described in claim 1, characterized in that, The target metabolite group was determined using pathway enrichment analysis, specifically including: Pathway enrichment analysis was performed on the first differential metabolite group to screen for metabolites that were significantly enriched in energy metabolism, amino acid metabolism, or fatty acid metabolism pathways. Metabolites whose expression levels differed by a greater than a predetermined fold between the Qi stagnation syndrome group and the non-Qi stagnation syndrome group were retained; Based on KEGG pathway topology analysis, metabolites that participate in more than two related pathways or are located at key nodes in the pathway were selected. Based on the pathogenesis characteristics of Qi stagnation syndrome in Traditional Chinese Medicine, the final target metabolite group was determined.

9. A computer-readable storage medium storing computer program instructions thereon, characterized in that, The computer program instructions, when executed by a processor, implement the method as described in any one of claims 1 to 8.

10. An electronic device comprising a memory and a processor, characterized in that, The memory is used to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the processor to implement the method as described in any one of claims 1 to 8.

Images