A novel integrated analytical platform for the quality assessment of complex natural products.

Abstract

The present invention provides a novel integrated platform that enables the verification of manufacturing processes for complex natural products, as well as the standardization and evaluation of complex natural products. This platform allows for the "system-wide" characterization of complex mixtures contained in therapeutic and / or health-promoting products based on or composed of natural substances, without limiting the approach to reductionist chemical approaches targeting single markers. This is achieved through the integration of qualitative and quantitative fingerprint characterization data metabolomics and physical and biophysical activity data for the mixtures, thereby providing a product characterization that forms the basis for verifying manufacturing processes that can ensure the reproducibility of efficacy and safety across different batches of the product. The analytical platform described herein enables the assurance of a complex natural product as a whole, by considering all the characteristics of its quality and efficacy, through a multidimensional analytical approach, in order to ensure reproducibility regarding efficacy and safety.

Description

[Technical Field]

[0001] Technical field of inventions This invention relates to the verification, standardization, and evaluation of manufacturing processes for complex natural products. In particular, the invention provides a novel integrated platform that enables the verification, standardization, and evaluation of complex natural products, which allows for the characterization of complex mixtures based on or composed of natural substances, used in the manufacture of therapeutic products (e.g., medical devices) or products that support the physiological functions of organisms (e.g., dietary supplements), as a “whole system,” and this characterization is not limited to a chemical approach targeting a single marker. This platform enables the integration of qualitative and quantitative fingerprint characterization data metabolomics of the mixture with physical and biophysical activity data, thereby providing product characterization that forms the basis for a process of verifying manufacturing processes that can ensure reproducibility of efficacy and safety between different batches of the product.

[0002] The analytical platform described herein enables the holistic assurance of the quality of complex natural products by considering their entirety in order to guarantee reproducibility regarding efficacy and safety through a multidimensional analytical approach. [Background technology]

[0003] Conventional technology In recent years, complex natural product families have become increasingly important for treating so-called "self-limiting (spontaneously resolving) diseases" (e.g., cough, sore throat, constipation) in specific patient groups (such as children and the elderly), and also for addressing important "still unmet medical needs," particularly syndromes and functional disorders where classical pharmacological approaches fail to provide satisfactory answers in terms of risk-benefit ratios. This last parameter, which plays a crucial role in characterizing treatments, has gradually evolved in recent years, moving from evaluations focused solely on patients to also assessing the environmental impact of treatments. This is because the introduction of non-biodegradable substances into the environment can cause direct and undeniable adverse effects on human health, creating an unsustainable vicious cycle.

[0004] In this new relationship, which can be defined as a "risk-benefit relationship extended to the relationship between humans and the environment," it is clear that using natural products that are biodegradable by definition presents an opportunity to consider all of their potential benefits.

[0005] Regulation 2017 / 745 concerning medical devices sets forth a clear regulatory framework that adds to the existing regulatory framework for pharmaceuticals, particularly by specifying the possibility of using complex natural products for the treatment and prevention of medical conditions. The fundamental difference between these two regulatory situations relates to the mechanism of action that achieves therapeutic activity: in the case of pharmaceuticals, this therapeutic activity is obtained through pharmacological (or immunological, or metabolic) actions leading to the interaction of a specific substance with a cellular target (usually a receptor), whereas in the case of medical devices, the therapeutic purpose must be achieved by a mechanism of action other than those mentioned above. Therefore, in accordance with this regulation, therapeutic compositions / formulations are defined as either pharmaceuticals or medical devices based on the mechanism of action by which they exert their therapeutic effect. In the case of natural substances that have therapeutic effects, these substances can be classified as either pharmaceuticals or medical devices. However, in the case of pharmaceuticals, it is necessary to demonstrate that their therapeutic activity is achieved through specific substances contained in natural materials (generally defined as markers or active ingredients) that act on specific cellular targets, whereas in the case of medical devices, therapeutic activity is related to the entire complex matrix and cannot be traced back to a specific mechanism of action.

[0006] From a technical and regulatory standpoint, the above means that pharmaceuticals must have the ability to restore, modify, or alter physiological functions, considering that they achieve therapeutic effects through pharmacological, immunological, or metabolic mechanisms of action, either by a single molecule or a combination of molecules (Directive 2001 / 83, Article 2), whereas medical devices must function through non-pharmacological, immunological, or metabolic actions, thereby altering physiological or pathological processes or conditions (Regulation 2017 / 745, Article 2) [Capone, L., Geraci, A., Giovagnoni, E., Marcoaldi, R., and Palazzino, G. (2012) “Elementi di valutazione e di discernimento tra dispositivo medico e medicinale. Roma: Istituto Superiore di Sanita. Rapporti ISTISAN 12 / 30; Racchi, M., Govoni, S., Lucchelli, A., Capone, L., and Giovagnoni, E. (2016). Insights into the definition of terms in European medical device regulation. Expert Rev. Med. Devices 13 (10), 907-917. doi:10.1080 / 17434440.2016. 1224644].

[0007] In medical devices composed of natural substances, where the therapeutic effect of the natural substance is attributable to a set of defining components, establishing criteria for characterizing and verifying these innovative products is necessary to ensure the reproducibility of the risk-benefit ratio that justifies their therapeutic use.

[0008] Considering the high variability of natural substances, it is necessary to provide manufacturing processes and quality control approaches that can ensure that a single product batch remains within a certain range of variability, which is determined experimentally and has been confirmed to ensure the maintenance of a demonstrated risk-benefit ratio.

[0009] This characterization needs to be positioned as the basis for demonstrating the effectiveness and safety of medical devices in accordance with the provisions of Regulation 2017 / 745, Article 13.3, Annex I, and needs to be stated on the product label as a guarantee of such characterization.

[0010] This result can be achieved by an approach that is not limited to determining the composition of the product from a quantitative chemical perspective. In fact, this approach is sufficient to characterize products composed of purified substances, but it cannot reveal the reproducibility of complex substances, and for that, it is necessary to be able to describe the product from physicochemical and biological perspectives.

[0011] As described above, complex substances obtained from natural biological matrices (UVCB substances of unknown or variable composition, biologically derived substances) cannot be defined like pure chemical substances, but through analytical techniques such as fingerprinting and physicochemical and biological property analysis, the variability of complex substances can be revealed and verified within a specific range.

[0012] Therefore, a comprehensive and sophisticated analytical approach is required to characterize its structural and performance effectiveness and to reveal the correlations between various parameters.

[0013] Correlation studies between fingerprint data and activity of complex natural products are beginning to appear in the literature. For example, the paper "A novel strategy to evaluate the quality of traditional Chinese medicine based on the correlation analysis of chemical fingerprint and biological effect" by Jing Wang, Hongwei Kong, Zimin Yuan, Peng Gao, Weidong Dai, Chunxiu Hu, Xin Lu, and Guowang Xu (Journal of Pharmaceutical and Biomedical Analysis 83 (2013) 57-64) describes an analysis of the correlation between the fingerprints of a certain number of compounds contained in complex natural products and their biological effects, in order to identify a certain number of marker components related to the quality of such natural products in relation to their biological effects. However, even in these studies, standardization is based on a single marker.

[0014] Currently, there are no available procedures that allow for the characterization of complex mixtures contained in natural products as a whole "system," ensuring the validation of manufacturing processes and / or the confirmation of reproducibility between different batches of complex natural products without limiting characterization to a select few markers. Furthermore, the lack of such procedures makes it impossible to ensure the reproducibility of risk-benefit ratios in clinical trials. [Overview of the project]

[0015] This invention provides an innovative analytical platform that enables the characterization of complex mixtures contained in natural products "as a whole system" by integrating advanced analytical techniques, rather than limiting characterization to reductionist chemical approaches targeting a single marker. This is achieved by integrating qualitative and quantitative fingerprint characterization data, metabolomics, and physical and biophysical activity data for the mixture. The inventors were able to provide product characterization, such as products that form the basis of verification procedures for manufacturing processes, which also enables the assurance of reproducibility between different batches of this product.

[0016] The analytical platform described herein enables the holistic assurance of complex natural products by considering all characteristics of product quality and efficacy, through a multidimensional analytical approach, in order to ensure reproducibility regarding efficacy and safety.

[0017] In fact, complex natural systems exhibit properties that cannot be explained by a single technique. This invention provides a method that enables correlation between qualitative and quantitative parameters and efficacy parameters, thereby providing an evaluation of complex products.

[0018] The present invention provides a process for revealing the properties of a product having therapeutic and / or health-promoting properties, wherein the product comprises a complex natural system (for example, a composition obtained by mixing natural substances, their extracts and fractions, preferably obtained through a physical manufacturing process with minimal environmental impact; the composition may exist in various pharmaceutical forms such as syrup, tablets, granule packets, capsules, sprays, suspensions, aerosols, aerosol powders, etc.), and the process comprises the following steps:

[0019] a. A step of performing qualitative and quantitative fingerprint analysis of a standard formulation of the said product and a formulation different from the said standard formulation, wherein the difference from the standard formulation is due to the presence of the same component of the same quality in different proportions, or the presence of the same component of different quality in the same proportions, thereby obtaining X blocks of fingerprint data.

[0020] b. A step of analyzing one or more biophysical properties of a standard formulation of the product and a formulation different from the standard formulation, wherein the difference from the standard formulation is due to the presence of the same component of the same quality in different proportions, or the presence of the same component of different quality in the same proportions, thereby obtaining a data block Y of biophysical properties.

[0021] c. A step of integrating data based on correlation using multivariate analysis, which involves (requires) the correlation between data block X and data block Y, which is integrated based on the correlation between each data obtained in a. and each data obtained in b., thereby obtaining a regression line for each pair of values ​​and its regression coefficient R 2 And calculate the mean squared error RMSEE, and R 2 The step provides a model that enables the prediction of the respective biophysical properties of standard formulations from the fingerprints used by selecting data pairs such that the ratio is ≥ 0.8.

[0022] The present invention also provides a process for verifying the manufacture of one or more batches of products having therapeutic and / or health-promoting properties, which include complex natural systems, the process comprising the following steps:

[0023] a. A step of performing qualitative and quantitative fingerprint analysis of a standard formulation of the said product and a formulation different from the said standard formulation, wherein the difference from the standard formulation is due to the presence of the same component of the same quality in different proportions, or the presence of the same component of different quality in the same proportions, thereby obtaining X blocks of fingerprint data.

[0024] b. A step of analyzing one or more biophysical properties of a standard formulation of the product and a formulation different from the standard formulation, wherein the difference from the standard formulation is due to the presence of the same component of the same quality in different proportions, or the presence of the same component of different quality in the same proportions, thereby obtaining a Y block of biophysical property data.

[0025] c. A step of integrating data based on correlation using multivariate analysis, which involves (requires) the correlation between data block X and data block Y, which is integrated based on the correlation between each data obtained in a. and each data obtained in b., thereby obtaining a regression line for each pair of values ​​and its regression coefficient R 2 And calculate the mean squared error RMSEE, and R 2 The step provides a model that allows for the prediction of the respective biophysical properties of standard formulations from the fingerprints used by selecting data pairs such that the result is ≥0.8.

[0026] A step in which a qualitative and quantitative fingerprint analysis is performed on one or more batches of the product to be validated using the same method as used in a, using the regression model developed in c. to predict the biophysical properties described in db.

[0027] Glossary Products having health and / or therapeutic properties, including the complex natural systems of the present invention, refer to products composed of natural substances, such as medical devices that fall under requirement 13.3 of EC Regulation 745 / 2017, Annex I (products having therapeutic properties), or nutritional supplements made of complex natural substances that fall under Article 2 of Directive 2002 / 46 / EC and Article 2 of DLgs 169 / 2004 (products having health-promoting effects).

[0028] In all parts of the Description and Claims, "product" means a product that is not naturally occurring but is produced by a complex combination of natural substances. Accordingly, the term "product having health and / or therapeutic properties, comprising a complex natural system" may be replaced in all parts of the Text and Claims with "a formulation having health and / or therapeutic properties, comprising (or consisting of) a complex natural system," or "a composition having health and / or therapeutic properties, comprising (or consisting of) a complex natural system," or "a mixture having health and / or therapeutic properties, comprising (or consisting of) a complex natural system." Furthermore, in any part of the description and claims, the phrase "contains a complex natural system" may be replaced with "consisting of a complex natural system."

[0029] "Products with healthy activity" refers to products that, when taken through the appropriate route, provide health benefits or are beneficial to health, as described above, but it particularly refers to products that benefit the health of healthy individuals. Products with healthy activity include supplements, special medical foods, and mixtures of nutrients (proteins, vitamins, enzymes, oils, etc.).

[0030] The terms “formulation” or “quality sample,” as also defined herein as the “standard formulation” or “target product” of product x, refer to a formulation of a product having therapeutic and / or health-promoting properties, including the complex natural systems described above, which correspond to the properties required to be defined as a medical device under EC Regulation 745 / 2017, Requirement 13.3, Annex I, or as a dietary supplement composed of complex natural substances in accordance with Directive 2002 / 46 / EC Article 2 and DLgs 169 / 2004 Article 2. The definition of a medical device, in accordance with the aforementioned regulations, includes therapeutic compositions comprising substances or groups of substances, such as biological materials, which can take various pharmaceutical forms, such as syrups, tablets, granule packets, capsules, sprays, suspensions, aerosols, and aerosol powders. Such formulations may include health-promoting compositions, such as the dietary supplements described herein.

[0031] The term "low-quality formulation (or sample or component) (of the product of the present invention)" refers to a formulation that differs from the standard formulation (or sample) due to the presence of the same component in a different proportion than the sample, or due to the use of the same component as present in the standard formulation but having different quality characteristics than those used in the definition of the product, and is defined as a low-quality formulation.

[0032] Data integration refers to building correlation models between qualitative and quantitative fingerprint data and biophysical activity, which makes it possible to predict corresponding data (e.g., biophysical activity) by measuring one data point (e.g., NIR fingerprint) in a new batch.

[0033] Obtaining this information across multiple batches allows for the definition of very strict quality ranges by considering it from multiple perspectives according to qualitative and quantitative fingerprint data and correlation models of biophysical activity. In this way, quality ranges can be defined individually depending on the product in question. In this explanation, the term "characterization" refers to a qualitative and quantitative description (definition) of the complex natural product and its biophysical properties.

[0034] Qualitative and quantitative descriptions of complex natural products: This refers to the results of product analysis performed using fingerprinting techniques such as NIR spectroscopy and / or mass spectrometry, or metabolomics. NIR spectroscopy: This method investigates the absorption and emission of light in the infrared region (4000-12500 cm-1) by materials. Mass spectrometry: This method involves examining and measuring the unique spectrum of a sample. In mass spectrometry, the sample is passed through an electromagnetic field, and then its mass is measured by the mass-to-charge ratio.

[0035] Description of the biophysical properties of complex systems: This means conducting tests to reveal the biophysical properties of a product that have been found to correlate with its effectiveness, such as tests measuring barrier effect, adhesion activity and antioxidant activity, neutralization activity, moisturizing effect, and lubricating effect.

[0036] In this invention, the term "Poliresin" is used to refer to a plant complex based on polysaccharides, resins, and flavonoids obtained by subjecting three starting plant materials (plantain (a herbaceous plant of the genus Musa), Grindelia (a plant of the genus Grindelia in the Asteraceae family), and Helichrysum (a plant of the genus Helichrysum in the Asteraceae family)) to a physical extraction process with minimal environmental impact (LPME process, liquid-phase microextraction method, a process characterized by crushing plant parts into extremely fine particles and performing liquid-phase extraction), the said process generates extracted fractions and makes it possible to standardize their content in the final product.

[0037] In this specification, OPLS (Orthogonal Partial Least Squares / Orthogonal Latent Structure Projection) refers to a well-known multivariate data analysis method that efficiently utilizes all relevant data with minimal loss of information. This is a recent expansion of the PLS known in the literature. OPLS separates systematic or parallel changes from unpredictable or orthogonal changes.

[0038] PLS (Latent Structure Projection using Partial Least Squares) is a well-known statistical method for relating two matrices X and Y using a multivariate linear model. PLS determines whether the relationship between the data block Y (dependent variable) and the variable block X (predictor variable) is linear or polynomial. The objective function of PLS ​​is to maximize the covariance between X and Y.

[0039] PLS2 (Partial Least Squares 2) is the first algorithm proposed for performing PLS regression, now known in the literature, and is currently one of the most widely used implementations of PLS. PLS2 refers to a model with multiple dependent variables. Given a Y matrix of response variables and an X matrix of predictor variables, the objective of PLS2 is to compute a regression coefficient matrix B that constructs a linear regression model.

[0040] The complex natural ingredients described herein are components found in products based on natural substances, and are therefore not synthetic, but comprise multiple components, such as cut or crushed parts of plants, plant extracts or plant parts obtained by aqueous alcohol extraction, fractions of said extracts (e.g., fractions obtained by semipermeable membrane filtration (microfiltration, ultrafiltration, nanofiltration)), or those obtained by selective precipitation by controlled evaporation of aqueous alcohol solutions, or those obtained by treatment with adsorbent resins. [Brief explanation of the drawing]

[0041] Detailed description of the drawing [Figure 1]The OPLS algorithm (orthogonal partial least squares / orthogonal latent structure projection) uses NIR profiles (near-infrared fingerprints, fingerprints or profiles obtained by near-infrared spectroscopy) and adhesion effects (PINCL) to more effectively separate systematic changes in information by separating orthogonal information from parallel information and considering only predictive or parallel information. Information in block X is given by NIR fingerprints, and information in block Y is given by adhesion effects (PINCL). This figure shows bar graphs (summary of goodness of fit) of the cumulative diagnostic parameters R2 and Q2 (in light gray and dark gray, respectively) representing the performance of the OPLS model regarding the correlation between NIR fingerprint and adhesion effect data. R2 is an index representing the goodness of the model (fit), and Q2 is an index representing the goodness of the prediction (predictive excellence). The validation of the diagnostic parameters R2 and Q2 is considered to be satisfied when the two parameters are close to each other and approach 1 (y axis). The number of components used is indicated on the x-axis and enclosed in square brackets. The two bars on the left represent the R2 and Q2 values ​​for parallel components and zero orthogonal components (comp[1+0]). The two bars in the middle represent the R2 and Q2 values ​​for parallel components and the first orthogonal component (comp[1+1]). The two bars on the right represent the R2 and Q2 values ​​for parallel components and the second orthogonal component (comp[1+2]). Therefore, in this case, the optimal OPLS model is the one using one parallel component and two orthogonal components (comp[1+2]) because the two values ​​of R2 and Q2 tend to be close to each other and approach 1. The graph in the lower right shows the software used and the date of reprocessing. [Figure 2a]A scatter plot of NIR fingerprint data (near-infrared) integrated with the adhesion effect value to the inclined surface (PINCL) was created using the OPLS algorithm (orthogonal partial least squares / orthogonal latent structure projection). The x-axis shows block information as predicted (or calculated) data. Each circle shown in the graph represents an experimental measurement of adhesion to the inclined surface for each sample analyzed (see Table 7), but these samples are characterized by different compositions as described in the text (see Table 6). The samples are identified by the acronym "DOE_EXP 1~8", corresponding to the samples referred to as "Experiment 1~8" in Table 7. The dotted line in the center of the graph represents the relationship between the data and is characterized by the equation y = 0.9966x + 0.1152, with a correlation coefficient R² of 0.993, shown in the upper left of the graph. This straight line indicates that the estimated error is low, with an RMSEE (Root Mean Square Error of Estimates) of 2.03219, and the cross-validation error is small, with an RMSE CV (Root Mean Square Error of Cross-Validation) of 2.6701, resulting in a very accurate prediction of biophysical activity values ​​(Y data block) from the NIR fingerprint (X data block). The RMSEE and RMSE CV values ​​are listed under the x-axis item YPRED[1](PINCL). The vertical bars on the right represent the experimental values ​​of the Y block used to create the model (in this case, measurements of adhesion to the inclined surface) using a color gradient. The graph in the lower right shows the software used and the date of reprocessing. [Figure 2b]This scatter plot was created using the OPLS algorithm (orthogonal partial least squares / orthogonal latent structure projection) of NIR fingerprint data (near-infrared) integrated with the value of adhesion effect to inclined surfaces (PINCL), including the prediction of the TARGET adhesion effect from NIR fingerprints. This graph is basically the same as the graph in Figure 2a, but with the addition of TARGET samples (DoE_Target) to the prediction. Therefore, all the details explained in the previous graph are also valid here. Furthermore, in this case, the NIR fingerprint of the TARGET sample is added to the information of block X, which is composed of NIR fingerprints. The sample data used to build the model is represented by light gray circles, and the TARGET sample in the prediction model is shown by dark gray circles. By using a regression model, the information of block Y is predicted using the fingerprint information of the TARGET sample, in this case predicting the adhesion information to inclined surfaces. The dotted line in the center of the graph represents the interpolation between the data, including the TARGET sample, and is characterized by the equation y = 0.9969x - 0.2445, with a correlation coefficient R² of 0.9893, which is shown in the upper left of the graph. This straight line indicates a low value of prediction error and a RMSEP (root mean square error of prediction) of 3.13957, which indicates that the prediction of biophysical activity values ​​is very accurate. The RMSEP value is listed below YPredPS [1](PINCL) on the x axis. [Figure 3]The R2 and Q2 of the OPLS model using NIR profiles (near-infrared fingerprints, fingerprints or profiles obtained by near-infrared spectroscopy) and barrier effects. The OPLS algorithm (orthogonal partial least squares / orthogonal latent structure projection) separates orthogonal information from parallel information, considering only predictive or parallel information, thus more effectively separating systematic changes in information. Information in block X is given by the NIR fingerprint, and information in block Y is given by the barrier effect (BARR). This figure shows bar graphs (summary of goodness of fit) of the cumulative diagnostic parameters R2 and Q2 (in light gray and dark gray, respectively), representing the model's performance (model performance) regarding the correlation between NIR fingerprint and barrier effect data. R2 is an index representing the goodness of fit, and Q2 is an index representing the excellence of prediction (predictive excellence). The validation of the diagnostic parameters R2 and Q2 is considered to be satisfied when the two parameters are close to each other and approach 1 (y-axis). The number of components used is indicated on the x-axis and enclosed in square brackets. The two bars on the left represent the R2 and Q2 values ​​for the parallel component and zero orthogonal components (comp[1+0]). The middle bar represents the R2 and Q2 values ​​for the parallel component and the first, second, and third orthogonal components (comp[1+1], comp[1+2], comp[1+3]), respectively. The last two bars on the right represent the R2 and Q2 values ​​for the parallel component and the fourth orthogonal component (comp[1+4]). Therefore, in this case, the optimal OPLS model is the one using one parallel component and four orthogonal components (comp[1+4]) because the two values ​​of R2 and Q2 tend to be close together and approach 1. The graph in the lower right shows the software used and the date of reprocessing. [Figure 4a]Scatter plots of NIR fingerprint data (near-infrared) integrated with barrier effect (BARR) measurements, constructed using the OPLS algorithm (orthogonal partial least squares / orthogonal latent structure projection). The x-axis shows block information as predicted (or calculated) data. Each circle shown in the graph represents an experimental measurement of the barrier effect for each sample analyzed (see Table 7), but these samples feature different compositions as described in the text (see Table 6). The samples are identified by the acronym "DOE_EXP 1~8", corresponding to the samples referred to as "Experiment 1~8" in Table 7. The dotted line in the center of the graph represents the relationship between the data, characterized by the equation y = 1x + 9.003e10 - 5, with a correlation coefficient R² of 0.9991, shown in the upper left of the graph. This straight line indicates that the estimated error is low, with an RMSEE (Root Mean Square Error of Estimates) of 1.07321, and the cross-validation error is small, with an RMSECV (Root Mean Square Error of Cross-Validation) of 2.14445, resulting in a very accurate prediction of biophysical activity values ​​(Y data block) from the NIR fingerprint (X data block). The RMSEE and RMSE CV values ​​are listed below the x-axis label YPRED[1](BARR EFFECT). The vertical bars on the right represent the experimental values ​​of the Y block used to create the model using a color gradient. The graph in the lower right shows the software used and the date of reprocessing. [Figure 4b]This scatter plot was created using the OPLS algorithm (orthogonal partial least squares / orthogonal latent structure projection) for NIR fingerprint data (near-infrared) integrated with barrier effect values ​​(BARR), including predictions of the TARGET barrier effect from NIR fingerprints. This graph is basically the same as the graph in Figure 4a, but with the addition of TARGET samples (DoE_Target) to the prediction. Therefore, all the details explained in the previous graph are also valid here. Furthermore, in this case, the NIR fingerprints of the TARGET samples are added to the information of block X, which is composed of NIR fingerprints. The sample data used to build the model is represented by light gray circles, and the TARGET samples in the prediction model are shown by dark gray circles. By using a regression model, the information of block Y is predicted using the fingerprint information of the TARGET samples, in this case predicting the barrier effect information. The dotted line in the center of the graph represents the interpolation between data including the TARGET sample and is characterized by the equation y = 1.003x + 0.1146, with a correlation coefficient R² of 0.9892, displayed in the upper left of the graph. This straight line indicates a low prediction error value and a RMSEP (root mean square error of prediction) of 3.76766, which indicates that the prediction of biophysical activity values ​​is very accurate. The RMSEP value is listed under the x-axis notation YPredPS [1] (BARR EFFECT). [Figure 5]The R2 and Q2 values ​​of the OPLS model using UHPLC-qToF profiles (ultrahigh-performance liquid chromatography / quadrupole time-of-flight mass spectrometry fingerprints, fingerprints or profiles obtained by ultrahigh-performance liquid chromatography / quadrupole time-of-flight mass spectrometry) and antioxidant activity. The OPLS algorithm (orthogonal partial least squares / orthogonal latent structure projection) separates orthogonal information from parallel information, considering only predictive or parallel information, thus more effectively separating systematic changes in information. Information in block X is given by the UHPLC-qToF fingerprint (UNT), and information in block Y is given by antioxidant activity (ORAC). This figure shows bar graphs (summary of goodness of fit) of the cumulative diagnostic parameters R2 and Q2, which represent the model performance (model performance), regarding the correlation between UHPLC-qToF fingerprints and antioxidant activity (shown in light gray and dark gray, respectively). R2 is an index representing the goodness of fit of the model, and Q2 is an index representing the goodness of prediction (predictive performance). The validation of diagnostic parameters R2 and Q2 is considered satisfied when the two parameters are close to each other and approach 1 (y-axis). The number of components used is indicated on the x-axis and enclosed in square brackets. The two bars on the left represent the values ​​of R2 and Q2 for parallel components and zero orthogonal components (comp[1+0]). In this case, the orthogonal components were not used because the values ​​of R2 and Q2 were very high and only close to 1 for the parallel components. Therefore, the optimal OPLS model is the model using one parallel component and zero orthogonal components (comp[1+0]). The graph in the lower right shows the software used and the date of reprocessing. [Figure 6a]Scatter plots were created using the OPLS algorithm (Orthogonal Least Squares / Orthogonal Latent Projection) of UHPLC-qToF (Ultrahigh Performance Liquid Chromatography / Quadrupole Time-of-Flight Mass Spectrometry) fingerprint data integrated with antioxidant activity (ORAC) measurements. The x-axis shows block information as predicted (or calculated) data. Each circle shown in the graph represents the experimental measurement of antioxidant activity for each analyzed sample (see Table 7), but these samples feature different compositions as described in the text (see Table 6). The samples are identified by the acronym "DOE_EXP 1~8", corresponding to the samples referred to as "Experiments 1~8" in Table 7. The dotted line in the center of the graph represents the relationship between the data and is characterized by the equation y = 1x + 8.809e10 - 5, with a correlation coefficient R² of 0.9591, shown in the upper left of the graph. This straight line indicates that the estimated error is low, with an RMSEE (Root Mean Square Error of Estimates) of 296.847, and that the cross-validation error is small, with an RMSE CV (Root Mean Square Error of Cross-Validation) of 339.527, resulting in a very accurate prediction of biophysical activity values ​​(Y data block) from the UHPLC-qToF fingerprint (X data block). The RMSEE and RMSE CV values ​​are listed under the x-axis item name YPRED[1](ORAC). The vertical bars on the right represent the experimental values ​​of the Y block used to create the model, expressed in a color gradient. The graph in the lower right shows the software used and the reprocessing date. [Figure 6b]This scatter plot was constructed using the OPLS algorithm (orthogonal partial least squares / orthogonal latent structure projection) of UHPLC-qToF (ultrahigh-performance liquid chromatography / quadrupole time-of-flight mass spectrometry) fingerprint data integrated with antioxidant activity (ORAC) measurements, including predictions of the antioxidant activity of TARGET from UHPLC-qToF fingerprints. This graph is essentially the same as Figure 6a, but with the addition of the TARGET sample (DoE_Target) to the prediction. Therefore, all the details described in the previous graph are also valid here. Furthermore, in this case, the UHPLC-qToF fingerprint of the TARGET sample is added to the information of block X, which consists of UHPLC-qToF fingerprints. The sample data used to build the model is represented by light gray circles, and the TARGET sample in the prediction model is shown by dark gray circles. By using a regression model, the information of block Y is predicted using the fingerprint information of the TARGET sample, in this case predicting the antioxidant activity information. The dotted line in the center of the graph represents the interpolation between the data, including the TARGET sample, and is characterized by the equation y=1x-38.7, with a correlation coefficient R² of 0.9513, displayed in the upper left of the graph. This straight line indicates a low value of prediction error, with an RMSEP (root mean square error of prediction) of 346.658, which indicates that the prediction of biophysical activity values ​​is very accurate. The RMSEP value is listed below YPredPS [1](ORAC) on the x axis. [Modes for carrying out the invention]

[0042] Detailed explanation The present invention provides a method for characterizing a product having therapeutic and / or health-promoting properties, wherein the product includes (or is composed of) complex natural systems, and the method includes the following steps:

[0043] a. A step of performing qualitative and quantitative fingerprint analysis of a standard formulation of the said product and a formulation different from the said standard formulation, wherein the difference from the standard formulation is due to the presence of the same component of the same quality in different proportions, or the presence of the same component of different quality in the same proportions, thereby obtaining X blocks of fingerprint data.

[0044] b. A step of analyzing one or more biophysical properties of a standard formulation of the product and a formulation different from the standard formulation, wherein the difference from the standard formulation is due to the presence of the same component of the same quality in different proportions, or the presence of the same component of different quality in the same proportions, thereby obtaining a data block of biophysical properties.

[0045] c. A step of integrating data based on correlation using multivariate analysis, which involves (requires) the correlation between data block X and data block Y, which is integrated based on the correlation between each data obtained in a. and each data obtained in b., thereby obtaining a regression line for each pair of values ​​and its regression coefficient R 2 And calculate the mean squared error RMSEE, and R 2 A step is provided in which a model is provided that allows for the prediction of the respective biophysical properties of standard formulations from the fingerprints used, by selecting data pairs such that the ratio is ≥0.8.

[0046] Thus, regression models make it possible to monitor product quality not only from the perspective of qualitative and quantitative fingerprints, but also from the perspective of required biophysical activity, representing an unprecedented advance in the quality assessment of complex natural products.

[0047] The data obtained in c. can be used to classify products as either "conforming quality" or "non-conforming quality" based on specific product standards. Conformity standards are variable and depend on the individual product.

[0048] The present invention also provides a process for verifying the manufacture of one or more batches of products having therapeutic and / or health properties, including complex natural systems, the process comprising the following steps:

[0049] a. A step of performing qualitative and quantitative fingerprint analysis of a standard formulation of the product and a formulation different from the standard formulation, wherein the difference from the standard formulation is due to the presence of the same component of the same quality in different proportions, or the presence of the same component of different quality in the same proportions, thereby resulting in a block.

[0050] b. A step of analyzing one or more biophysical properties of a standard formulation of the product and a formulation different from the standard formulation, wherein the difference from the standard formulation is due to the presence of the same component of the same quality in different proportions, or the presence of the same component of different quality in the same proportions, thereby obtaining a Y block of biophysical property data;

[0051] c. A step of integrating data based on correlation using multivariate analysis, which involves (requires) the correlation between data block X and data block Y, which is integrated based on the correlation between each data obtained in a. and each data obtained in b., thereby obtaining a regression line for each pair of values ​​and its regression coefficient R 2 And calculate the mean squared error RMSEE, and R 2 By selecting data pairs such that the ratio is ≥0.8, a model is provided that enables the prediction of the respective biophysical properties of standard formulations from the fingerprints used, in the above step.

[0052] A step in which a qualitative and quantitative fingerprint analysis is performed on one or more batches of the product to be validated using the same method as used in a, using the regression model developed in c. to predict the biophysical properties described in db.

[0053] The data obtained in d. can be used to classify products as "quality that meets the standards" or "quality that does not meet the standards" based on specific product standards. The standards for conformity may vary and depend on the individual product, but the aim is to maintain qualitative and quantitative parameters and biophysical activity parameters.

[0054] For each specific product, the acceptance criteria are established through fingerprinting and biophysical measurements performed on products that differ from the standard formulation. These values ​​represent the limits of conformity for each characteristic, and batches exceeding these limits must be investigated.

[0055] The effectiveness of this analytical platform makes it possible to monitor qualitative and quantitative parameters and biophysical activity parameters of various production batches, and furthermore, to discard deviant batches that cannot be guaranteed to maintain biophysical activity within a predetermined range based on individual control charts.

[0056] By obtaining this information across numerous batches, it becomes possible to define extremely precise quality ranges by considering various aspects based on qualitative and quantitative fingerprints and correlation models of biophysical activity. This allows the quality range to be specifically defined according to the product in question.

[0057] Thus, regression models enable monitoring of product quality not only from a qualitative and quantitative fingerprinting perspective but also from a biophysical activity perspective, representing an unprecedented advance in the quality assessment of complex natural products.

[0058] According to the present invention, the therapeutic product comprising a complex natural system is preferably a product made of natural substances, for example, a medical device that falls under EC Regulation 745 / 2017, Requirement 13.3, Annex I, and / or a nutritional supplement having health benefits made of complex natural substances as defined in Directive 2002 / 46 / EC Article 2 and Decree 169 / 2004 Article 2.

[0059] Therefore, examples of such products include the medical devices or nutritional supplements described above, which are obtained by mixing natural substances and / or extracts thereof and / or fractions of said extracts. The ingredients used in the preparation of the medical devices or nutritional supplements are preferably obtained by physical manufacturing processes that have minimal environmental impact.

[0060] Non-limiting examples of natural substances in the present invention include parts of plants, powders of those parts, extracts of plants or parts thereof (leaves, flowers, roots, bark, stems, seeds, fruits, petioles, parts of flowers, etc.), fractions of said extracts (hydrated alcohol fractions or other fractions, etc.), essential oils, natural gums, natural resins, minerals, honey, propolis, animal substances, mineral substances, and the like.

[0061] This product can take various forms, such as syrup, tablets, capsules, granules, powder, solution, suspension, aerosol, aerosol powder, powder sachets, granule sachets, operculum, spray, cream, emulsion, soft gelatin capsule or hard gelatin capsule, or other suitable formulations commonly used in the industry.

[0062] According to the present invention, a formulation that differs from a standard product in that it contains the same components of the same quality in different proportions, or contains the same components of different quality in the same proportions, is a formulation of inferior quality. Such formulations can be produced by changing the components that are essential for the therapeutic or health-promoting activity of the product.

[0063] To rationally examine the contribution of each component to the properties of complex formulations, it is necessary to create a suitable number of formulations that allow for the evaluation of their effects on the product's properties. Therefore, an appropriate design of experiment (DOE) is required to discover the conditions under which components affect the response.

[0064] For classic pharmaceuticals containing one or two specific active ingredients (compounds), a "full factorial design" (DOE) is used, resulting in the creation of 2,000 formulations (K = number of active ingredients). In the case of complex natural products with multiple components contributing to the therapeutic effect, the number of low-quality formulations in item b. is immeasurable.

[0065] For example, in the case of a product consisting of 11 complex natural ingredients, it would be necessary to create over 1,000 formulations to describe all the low-quality variations, centered around the standard formulation.

[0066] Therefore, preliminary studies can identify the component requirements necessary to maintain the product's characteristics. This can be achieved by using quality (or standard) samples (which serve as the target or central point of the experimental design) and low-quality samples (which differ from the quality sample in that they contain the same component of the same quality in different proportions, or the same component of different quality in the same proportions). To select lower-quality formulations, the ingredients of the target products can be grouped together.

[0067] In embodiments of the present invention, lower-quality formulations can be selected by grouping components. These groups consist of three basic groups that, when modified, alter the properties of the formulation (in Table 4 of the Examples section, these groups are defined as “factors” and are A, B, and C). Further groups can be identified using all other components within the product, which can be defined as independent factor groups (in Table 4, these components belong to the group defined as factor D). A “computer-aided” experimental design using a non-orthogonal factorial design allows for random variations in the proportions of the grouped components (factors A, B, and C), with factor D being combined to complement each other to 100%. Through this computational process, it is possible to select a “low-quality” formulation “suitable for generating a regression line,” i.e., the formulation with the highest probability of constructing a line to be used as a model to predict new data. For the products analyzed in the Examples section, eight formulations were selected, as shown in Table 6. Therefore, the number of low-quality formulations to be considered in "full factorial planning" type DOE studies is significantly reduced.

[0068] The qualitative and quantitative fingerprint analysis in item a. can be performed using known techniques that can characterize the complexity of the product in question: for example, infrared (IR) and near-infrared (NIR) spectroscopy, mass spectrometry (MS, and various coupling possibilities), nuclear magnetic resonance (NMR), ultraviolet-visible spectroscopy (UV-Vis), etc. According to one embodiment, the fingerprint analysis is performed by UHPLC-qToF, recording of NIR spectra, recording of IR spectra, or by acquiring chromatograms using other mass spectrometers (ion trap, triple quadrupole, single quadrupole) coupled to various chromatography systems (e.g., gas chromatography, liquid chromatography), or by direct injection into a mass spectrometer, as well as NMR and UV-Vis spectroscopy coupled to chromatography systems (e.g., liquid chromatography).

[0069] In accordance with the present invention, item b. analyzes one or more biophysical properties that correlate with the therapeutic effect of the product in question. Non-limiting examples of such properties include mucosal adhesion properties, barrier effect, antioxidant effect, neutralizing properties, moisturizing properties, and lubricity.

[0070] Therefore, in one embodiment of the present invention, the analysis of the biophysical properties in item b. may be carried out by measuring one or more of the following: mucosal adhesion ability, barrier effect, antioxidant activity, neutralization activity, moisturizing activity, and lubrication activity.

[0071] The analysis of these properties can be carried out using any method known to those skilled in the art, but non-limiting examples of such techniques include, for example, measuring the (mucosal) adhesion effect by testing on an inclined surface, in vitro testing using human fibroblasts, measuring the barrier effect by testing the reduction of dextran-fluorescein molecules (molecular weight 70,000) permeability in a semipermeable membrane (transwell), and measuring the antioxidant effect by ORAC testing. In a preferred embodiment, the integration of data in item c. is performed using the OPLS algorithm.

[0072] OPLS separates data into predictive (or parallel) information and uncorrelated (or orthogonal) information, removing variability unrelated to Y (characteristic variable block) from X (descriptive variable block) and generating a sophisticated predictive correlation model. Therefore, OPLS models are superior because they separate orthogonal information from parallel information and consider only predictive or parallel information.

[0073] Suitable algorithms or functions for processing NIR fingerprint data by scaling include, for example, the SNV (Standard Normal Variance) algorithm, first derivative, and second derivative; for fingerprints obtained by mass spectrometry, the PARETO and MEAN-CENTERING algorithms; and for biophysical data, the UV (Unit Variance) or CENTERING algorithm, all of which are well known to those skilled in the art.

[0074] Algorithms or functions suitable for processing by conversion of data related to the analysis of each biophysical property include, for example, exponential functions (e.g., square exponent, 1 / 4 exponent) or logarithmic functions.

[0075] The processed variable is R, a diagnostic parameter of the model 2 and Q 2 which can be evaluated by calculation. Appropriate R 2 value data pairs are selected, for example, R 2 ≧0.8, preferably R 2 ≧0.9, preferably ≧0.95, more preferably 0.99 or more.

[0076] In any part of this specification, the term "including" can be replaced by the term "consisting of".

[0077] The following examples are based on the following standard formulation of Syrup A (Grintuss by Aboca) and formulations different from the standard formulation, the difference being due to the same components of the same quality being present in different proportions or the same components of different qualities being present in the same proportion. The provided examples are intended to illustrate an executable method of implementing the present invention or an executable method of performing a part of the steps of the claimed process and should not be construed as limiting the present invention in any sense. (Example)

[0078] Method used 1. NIR fingerprint measurement method - reflection mode 1.1. Measuring means Bruker NIR spectrometer, model MPA (multi-purpose analyzer)

[0079] 1.2. Parameters of the reflection method Resolution: 16 cm-1 ; Wave number reproducibility: 0.04 cm -1 Less than; wavenumber precision: 0.1 cm -1 Less than; Photometric accuracy: 0.1% T; Wavenumber range: 4000~12000 cm -1 Background scan: 64; Sample data acquisition scan: 64

[0080] 1.3. Sample Preparation For NIR analysis, 3 ml of each syrup sample was dropped into the center of a Petri dish using a disposable plastic pipette. A metal spacer was placed on the dropper to eliminate any air bubbles that might form between the spacer and the plate.

[0081] 1.4. Data acquisition from samples For each sample, three technical iterative experiments were performed, and data acquisition was repeated three times, allowing for monitoring of data reproducibility. Background correction was performed before each data acquisition.

[0082] 1.5. Data Preprocessing To develop a reprocessing method, it was also necessary to evaluate data preprocessing, and we used SIMCA® software, which made it possible to select a preprocessing method that included the use of SNV normalization.

[0083] 1.6. Data evaluation using SIMCA® New samples introduced into the NIR model are evaluated by statistical tests to show that they fall within the range of variances validated by the model. The aforementioned diagnostic tests are Hotellings T 2 And DModX.

[0084] 2. NIR Fingerprint Measurement Method - Transmission Mode This method differs from conventional methods in that it requires the use of a test tube with an optical path length of 2 mm to provide the sample for reading using the instrument.

[0085] 3. UHPLC-qToF Fingerprint Measurement Method 3.1. Reagents Methanol / ultrapure water 50:50 (v / v), HPLC grade methanol (MeOH), ultrapure water

[0086] 3.2. Materials Column: Waters Cortecs® C18 1.6μm 2.1×100 mm; Pre-column: Waters Cortecs C18 1.6μm 2.1×5mm VanGuard® pre-column 3 / Pk; Syringe filter, 0.2μm RC membrane, Phenomenex

[0087] 3.3. Measuring Instruments UHPLC-qToF system coupled to a 6545 quadrupole time-of-flight mass spectrometer (qToF) with a resolution of 2 GHz and an Electrospray Dual AJS ESI ion source: Agilent 1290 Infinity (Agilent Technologies). Ultrasonic processor. Chemical balance (resolution 0.00001 g).

[0088] 3.4. Sample Preparation Using a glass Pasteur pipette, weigh 0.500 g ± 0.005 g of the sample into a 50 mL volumetric flask. Soluble the sample in approximately 45 mL of 50% methanol for LC-MS by sonication at 35 °C ± 3 °C for 15 minutes. Dilute to 50 mL at room temperature using the same extraction solvent. Filter 3 mL of the extract through a 0.2 μm cellulose acetate filter and transfer to a 4 mL brown vial, discarding the first 0.5 mL of filtrate. Using a graduated 200 μL micropipette, take 200 μL of the filtered sample and transfer it to a brown vial with a glass insert and screw cap. Inject into the UHPLC-qToF system with negative polarity [ESI(-)] for metabolome profile measurement.

[0089] 3.5. UHPLC chromatography conditions Injection volume: 3 μl; Column temperature: 40°C. Eluent: A. Water, 0.1% acetic acid; B. Methanol, 0.1% acetic acid Time interval: 0 Waste; 2 min MS; Flow rate: 0.3 mL / min. Elution method:

[0090] [Table 1] Post-infusion time: 3 minutes; Iso pump: 0.05 mL / min

[0091] 3.6. Reprocessing of fingerprint data obtained by UHPLC-qToF As the first step, we will perform feature extraction (data extraction) using the XCMS package (version 3.2.0), which uses the following algorithm: • centWave This algorithm performs operations such as peak picking and alignment. The parameters shown in Table 2 were used to run the algorithm.

[0092] [Table 2] The second step involves using a filtering function on the obtained variables, which is called preprocessing. This function was executed using the parameters shown in Table 3.

[0093] [Table 3] The final step in variable extraction is subtracting the analysis blank. This is done using the blanksubtract.centWave function (Table 4).

[0094] [Table 4]

[0095] The resulting final data matrix is ​​converted into an a.csv file, which is included in the SIMCA(registered trademark) software, in order to proceed with statistical reprocessing using OPLS.

[0096] 4. Method for measuring adhesion activity on inclined surfaces A reference standard solution (RSS) is prepared by dissolving 0.9% w / w sodium alginate in physiological saline. The test sample (SAMPLE) (syrup, SYRUP A) is analyzed without any modifications.

[0097] Before starting the analysis, ensure that the temperature of the RSS and SAMPLE samples is 20°C.

[0098] Using a micropipette, dispense the same number of drops of SAMPLE and RSS onto the tip of a plate with a predetermined incline. The droplets are placed at equal intervals along the same line. Due to gravity, the sample and RSS move along the plate. If the adhesion capacity of SAMPLE is higher than that of RSS, the SAPMLE test meets the criteria.

[0099] 5. Method for measuring barrier effectiveness This study is based on the principle that the reduction in the permeation of dextran-fluorescein molecules (molecular weight 70,000) after application of the product is considered an indicator of the protective efficiency of the formulation, compared to an untreated control (no formulation applied, which serves as a positive control).

[0100] The test requires the installation of two chambers physically separated by a semipermeable membrane (Transwell). The product is applied to the surface of the Transwell, and dextran-fluorescein is applied 10 minutes later. One hour after dextran application, the solution on the receiving side is collected, and considering that the barrier effect can be determined by fluorescence measurement of the well solution in the receiving plate, the readings are taken with a plate reader set to excitation 494 nm and emission 518 nm. A blank (product application without dextran) is also considered. This condition is used to evaluate the contribution of the portion of the product that can diffuse inside the receiving chamber.

[0101] After calculating the amount of fluorescein dextran in the sample and positive control, calculate the percentage of barrier effect using the following formula: Barrier effectiveness rate (%) = 100 - ((Sample dextran concentration / Positive control mean dextran concentration)) × 100

[0102] Reagents and equipment Dextran fluorescein D1823 70,000 MW (Invitrogen) • PBS (Gibco) • HTS 96 3391 - Corning® HTS Transwell® • Corning 3605-96 Round-bottomed white plate + lid • Fluorescent plate reader (Varioskan - Thermo Fisher) Eppendorf 1.5ml Thermo Fisher p10, p20, p100, p1000 pipettes + tips • Multipet R Xstream Eppendorf + 5 ml tip • Falcon 15 ml and 50 ml Vortex · timer

[0103] Assay execution The reagents required to perform the test are as follows:

[0104] • Dextran fluorescein: As instructed in the datasheet, the powder needs to be resuspended in PBS at a concentration of 25 mg / ml. This will be the stock solution, which should be divided into fixed volumes and stored at -20°C (the stock solution should be vortexed before dividing). The concentration to be used in the test is 500 μg / ml, so it needs to be diluted 1:50 with PBS. • Sample: Mix the sample before starting the test and transfer approximately 5 ml to a 15 ml Falcon tube. • PBS: Pour the required amount of PBS into a 50 ml bottle. Store the PBS at room temperature.

[0105] The following is the experimental setup and steps for conducting the test. Each sample must be in a triple configuration.

[0106] Positive control (CTRL+): 15 μl of PBS on top, 90 μl of PBS on the bottom, and 2 μl of dextran fluorescein at a concentration of 500 μg / ml on top. Sample: 15 μl of sample at the top, 90 μl of PBS at the bottom, and 2 μl of dextran fluorescein at a concentration of 500 μg / ml at the top.

[0107] Blank: 15 μl of sample at the top, 90 μl of PBS at the bottom, and 2 μl of PBS at the top. This condition is used to evaluate the contribution of the portion of the product that can diffuse inside the receiving chamber. 1. Mix the test samples. 2. Prepare a 500 μg / ml dextran fluorescein solution and store it in a dark place. 3. Add 15 μl of syrup (for sample and blank) or 15 μl of PBS (for ctrl+) to the surface of the HTS 96 transwell and leave at room temperature for 10 minutes. Replace the tip for each transwell. 4. After 10 minutes, add 90 μl of PBS to the bottom. For the sample and ctrl+ samples, first vortex 500 μg / ml dextran fluorescein solution, then add 2 μl to the top. For the blank sample, add 2 μl of PBS to the top. 5. Incubate in a dark place at room temperature for 1 hour. 6. Prepare a calibration curve using Eppendorf tubes and hold it in a dark place at room temperature for 1 hour. 7. Transfer 75 μl (2 strips) from each calibration curve point and sample to a 96-well U-bottom white plate and measure fluorescence using Varioskan (excitation 494 nm, emission 521 nm) (excitation 494 nm, emission 521 nm).

[0108] Determination of dextran fluorescein After performing fluorescence measurements with the excitation wavelength set to 494 nm and the emission wavelength to 521 nm, a dextran fluorescein calibration curve is created using GRAPHPAD PRISM software. A function is set to subtract the average fluorescence value of the blank wells from the sample fluorescence value. Then, the dextran fluorescein concentrations in the sample and positive control are calculated by interpolating the fluorescence values ​​of the corresponding wells using a standard curve.

[0109] Analysis of results After calculating the amount of dextran fluorescein contained in the sample and positive control, the percentage of barrier effect is calculated using the following formula: Barrier effectiveness rate (%) = 100 - ((Dextran concentration in the sample / Mean dextran concentration of the positive control + ctrl)) × 100 The reduction in dextran fluorescein permeation after application of the product is considered an indicator of the protective efficiency of the formulation compared to the untreated control (no formulation application).

[0110] 6. Method for measuring antioxidant properties using ORAC measurement (oxygen radical absorption / antioxidant capacity) This method was developed to evaluate the protective effects of antioxidants on living organisms from reactive hydroxides and peroxides. It is designed for testing human and animal serum samples, plant products, foods, food components, pharmaceuticals, and pet food. Currently, it is considered the only method capable of measuring the inhibitory capacity that antioxidants can exert against free radicals. The unit of measurement for antioxidant power is named ORAC units.

[0111] This method is highly sensitive and utilizes 2,2'-azobis(2-aminedinopropane) dihydrochloride (AAPH), a water-soluble azo compound that generates water-soluble peroxide radicals at a constant rate by thermal decomposition, using fluorescein as a fluorescent marker. Fluorescein fluorescence is extremely sensitive to the three-dimensional structure and chemical integrity of the molecule itself. Under appropriate conditions, a decrease in fluorescence in the presence of peroxide radicals serves as an indicator of oxidative damage caused by the active nuclide. In a given sample, the inhibition of fluorescein decomposition due to the protective effect of present antioxidants, reflected in the retention of fluorescein fluorescence, serves as an indicator of the sample's antioxidant capacity against the active nuclide. Antioxidants protect the fluorescent marker from decay as long as they can capture radicals: once the antioxidant effect wears off, the radicals react with fluorescein, and fluorescence decreases. The fluorescence decay time is proportional to the amount and activity of antioxidants present in the sample.

[0112] Therefore, the ORAC assay measures the time-dependent decrease in the fluorescence of the marker molecule fluorescein, which is a result of damage caused by oxygen-centered radicals.

[0113] Assay results are quantified by showing the sum of the area under the reaction rate curve compared to a blank reaction without antioxidants after the reaction is complete. The area under the curve (AUC) is proportional to the concentration of all antioxidants present in the sample. This method allows for the simultaneous calculation of two parameters—inhibition time and inhibition rate of the active nuclide by all antioxidants—as a single quantity: therefore, this method is superior to similar methods that use inhibition rate at a fixed time and inhibition time at a fixed inhibition rate.

[0114] The final result is calculated using the difference in fluorescein decay AUC (relative fluorescence intensity vs. time) between the test sample and the blank.

[0115] Trolox® standard substance (6-hydroxy-2,5,7,8-tetramethylchroman-2-carboxylic acid) is a water-soluble vitamin E analog.

[0116] ORAC units = (AUC sample - AUC blank) / (AUC standard - AUC blank) × [standard] / [sample] = μmol / g The measurement results are reported based on the equivalence relationship that 1 ORAC unit = equivalent amount of 1 μM Trolox® per 100 g of sample. [Examples]

[0117] Characterization of complex products Grintuss (also referred to herein as "Syrup A"), a cough suppressant, is produced through a characterized and highly standardized production process and consists of the following mixture of ingredients:

[0118] Polyresin, based on polysaccharides, resins, and flavonoids (LPME extract fractions of plantain, grindelia, and helichrysum obtained by subjecting three starting plant materials (plantain (a herbaceous plant of the genus Musa), grindelia (a plant of the genus Grindelia in the Asteraceae family), and helichrysum (a plant of the genus Helichrysum in the Asteraceae family)) to a physical extraction process with minimal environmental impact (LPME process, liquid-phase microextraction method, a process characterized by crushing plant parts into extremely fine particles and extracting them in the liquid phase), honey, sugarcane, water, eucalyptus essential oil, star anise essential oil, lemon essential oil, natural lemon fragrance, gum arabic, and xanthan gum.

[0119] All of the experiments described below were conducted in sets of three.

[0120] 1. Characterization of experimental design (DOE) and experimental design samples To obtain a correlation model that integrates the qualitative and quantitative characteristics of a product with its corresponding biophysical properties, we defined different quality formulations of the original complex natural product. This model, used for characterizing numerous quality samples, enables the rigorous definition of quality levels for the same product.

[0121] -In the case of Syrup A, Table 5 shows all components deemed important for examining the influence on the properties of the complex natural product, classified into Factor A (represented by the components Polyresin, gum arabic, and xanthan gum), Factor B (represented by the honey components), and Factor C (represented by the sugar and water components). Factor D is listed as representing other components of the formulation (components: water, lemon flavoring, lemon essential oil, and eucalyptus essential oil), but was considered an independent factor in this study.

[0122] [Table 5]

[0123] Starting with the composition of the TARGET product (Syrup A), computer-aided design (CAD) was applied to increase or decrease factors A, B, and C, covering a box of compositions centered on TARGET and generating a grid of possible combinations. Only grid elements that met the formulation constraints were included in the candidate experimental set, which was sampled to select eight formulations to be considered for examining the response. Thus, Table 6 shows the recognized factor combinations with percentage increases or decreases compared to the target. In particular, factor A was found to vary within a range of ±80%, and factors B and C varied within a range of ±20%.

[0124] The combination procedure performed by the computer aimed to obtain the DOE matrix with the minimum number of conditions. Thus, this formulation experimental design procedure reduced the distance between the candidate formulation and the complete factor design experiments to the minimum, and therefore to 23 (2K), because it considered three factors (A, B, C), which corresponds to eight experiments.

[0125] Therefore, Table 6 lists the formulations to be considered. The above DOE matrix supports linear and interaction models with fewer than 3.6 conditions.

[0126] [Table 6] The formulations referred to as Experiments 1-8 were used to record fingerprints and biophysical characteristics.

[0127] 1.1 Fingerprint analysis using near-infrared (NIR) spectroscopy 4000~12500 cm -1 The near-infrared region was selected, and background noise was removed. Spectral recording can be performed in transmission or reflection mode, and in optical spectroscopy, transmittance T is the transmitted ray intensity (I t The ratio of the reflected light intensity (I0) to the incident light intensity (I0) is the reflectance. r This is the ratio of (I0) to the incident light intensity.

[0128] The NIR spectra are transformed into a data matrix with variables characterized by wavelength and corresponding transmittance or reflectance values. The matrix containing data from all the samples examined is then aligned and preprocessed, namely by transformation using the SNV (Standard Normalization) algorithm and scaling using the MEAN-CENTERING function. Additional preprocessing may include the first or second derivative, or a combination of these with the SNV algorithm. The resulting final data matrix is ​​then used in the next step of integrating it with biophysical data using OPLS.

[0129] Software packages available for performing NIR spectral analysis are specific to NIR instruments; for example, Bruker offers OPUS software.

[0130] Another software used for NIR spectral analysis is SIMCA® (Sartorius).

[0131] 1.2 Fingerprint analysis by UHPLC-qToF mass spectrometry. Feature extraction (variables) from the fingerprint was performed using peak picking and peak operation alignment. The parameters shown in Table 2 were used to run the algorithm.

[0132] In the second stage, a filter function was applied to the variables obtained using the parameters listed in Table 3. The final stage of variable extraction involves subtracting analysis blanks (Table 4). The resulting final data matrix is ​​converted to an a.csv file and used for integration with the active data in the next stage, where statistical reprocessing using OPLS correlation is performed.

[0133] Examples of software packages available for performing data extraction, data processing, and data analysis include XCMS and SIMCA® (Sartorius).

[0134] 1.3 Measurement of adhesion characteristics This biophysical property was measured by placing a complex natural substance (e.g., syrup A) on an inclined surface and allowing it to move due to gravity. The tested product had to have higher adhesion than the standard, which was verified by measuring the distance (mm) traveled by the product and the standard substance on the inclined surface.

[0135] To ensure reliable reproducibility of the measurements, they are performed under constant temperature conditions using an inclined surface with a fixed angle.

[0136] The detailed procedure used in this study is described as an example in the section on materials and methods.

[0137] 1.4 Measurement of barrier properties These biophysical properties were evaluated by placing the product on a semipermeable membrane (transwell) and measuring its ability to inhibit the permeation of a dextran fluorescein compound (molecular weight 70,000).

[0138] The detailed procedure used in this study is described as an example in the section on materials and methods.

[0139] 1.5 Measurement of antioxidant properties This property was measured using ORAC testing in accordance with standard protocols well known to those skilled in the art. The detailed procedure used in this section is described as an example in the Materials and Methods section.

[0140] 2. Calculation of multiple linear regression The experimental measurements of the biophysical properties used in the calculation of multiple linear regressions are reported in Table 7.

[0141] [Table 7]

[0142] Before proceeding with the evaluation of the relationship between fingerprints and biophysical responses (using OPLS), we analyzed the biophysical responses based on multiple linear regression (classical least squares regression) to verify the existence of linearity in the interaction model between the three factors A, B, and C in eight different formulations (Experiments 1-8). Based on equation (1), various different correlation coefficients were evaluated, and the results are shown in Table 8.

[0143] Equation (1) f(Y) = aA + bB + cC + dAB + eAC + fBC + g In the formula: f(Y) is the transformation function of the response to be evaluated. a,b,c = Regression coefficients representing the effects of variations in parameters A, B, and C related to antioxidant biophysical activity (a), adhesion effect (b), and barrier effect (c). d,e,f = Regression coefficients representing the effects of variations in binding parameters AB, AC, and BC related to antioxidant biophysical activity (a), adhesion effect (b), and barrier effect (c). A, B, C = Variable component mixture in Experiments 1-8 g= intercept

[0144] The data on biophysical parameters were transformed using base-10 logarithmic, exponential 1 / 4, and exponential square functions for antioxidant activity, adhesion activity, and barrier effect, respectively (also known as the Box-Cox transformation; Box, GEP and Cox, DR (1964). An analysis of transformations, Journal of the Royal Statistical Society, Series B, 26, 211-252).

[0145] Table 8 shows the results of the multiple linear regression evaluation. In the three activity models evaluated, Q 2 Value and R 2 The values ​​are good.

[0146] [Table 8]

[0147] Since the coefficient 'a' indicates the correlation of factor A in mixtures 1-8, a good correlation (positive proportionality) with antioxidant activity and a fairly good correlation (inverse proportionality) with adhesion effect are observed. Factor A is associated with the presence of polyresin and gums (xanthan gum and gum arabic).

[0148] The coefficient b shows the correlation of factor B for mixtures 1-8, and there is a fairly good correlation (inverse proportion) with the adhesion effect and a fairly good correlation (inverse proportion) with the barrier effect. Factor B is associated with the presence of honey.

[0149] The coefficient c shows the correlation of the C factor for mixtures 1-8, showing a good correlation (positive proportionality) with the adhesion effect and a fairly good correlation (inverse proportionality) with the barrier effect. The C factor is related to the presence of sugars and water. The coefficient f represents the correlation between the composite BC factors of mixtures 1-8 and shows a good correlation (direct proportionality) with the adhesion effect. The composite BC factors are related to the presence of honey, sugars, and water.

[0150] By substituting the data from Table 8 into Equation 1, the corresponding regression equation is obtained, which allows us to examine the relationships between the responses of factors A, B, and C in experimental formulations 1-8 (Table 7) that consider the interaction model. Box-Cox transformations were applied to these responses to obtain normally distributed residuals. The ORAC response values ​​were transformed using a base-10 logarithmic transformation, giving a linear model represented by the following equation: log10(ORAC) = 3.259 (p<0.001) + 0.317 (p<0.001) A - 0.085 (p=0.014) C

[0151] The number of conditions was 1.81, and the p-value of the analysis of variance (ANOVA) was less than 0.001. The goodness of fit of the multiple linear regression model was R 2 = 0.97. Q was obtained by leave-one-out(loo) cross-validation. 2 = 0.93 was obtained.

[0152] By converting the adhesion response to the inclined surface to the power of 0.25, the following interaction model is obtained:

[0153] (Adhesion to inclined surfaces) 0.25 = 2.296 (p<0.001) -0.288 (p<0.001) A-0.262 (p<0.001) B-0.555 (p<0.001) C+0.068 (p=0.022) BC The "number of conditions" was 2.96, and the ANOVA result p-value was less than 0.001. This model is R 2 =0.99, and Q 2 It showed =0.98.

[0154] The "barrier effect" response was transformed by squaring, yielding the following linear model:

[0155] (Barrier effect) 2 = 8820 (p<0.001) - 1130 (p=0.047) A - 1300 (p=0.129) B - 3300 (p=0.011) C

[0156] The values ​​from Experiment 5, where the response was found to be saturated, were excluded from the calculation. The "number of conditions" was 3.46, and the p-value by ANOVA was 0.022. The model was R 2 =0.78, and Q 2 The result was 0.21. Therefore, the response in Experiment 5 was estimated to be 110.

[0157] 3. Integration of qualitative and quantitative data and biophysical properties through OPLS correlation. Based on favorable results obtained from multiple linear regression studies of biophysical properties measured for mixtures 1-8, we proceeded to construct a correlation model between qualitative and quantitative fingerprint data and biophysical data using the Orthogonal Latent Structure Projection (OPLS) algorithm. In OPLS, experimental data is separated into predictive (or parallel) information and uncorrelated (orthogonal) information, thereby removing variability unrelated to Y (characteristic variable) from X (descriptive variable). Therefore, the OPLS model is superior because it separates predictive or parallel information from unpredictable or orthogonal information and considers only that predictive or parallel information.

[0158] Similar correlations can also be obtained using the PLS and PLS2 algorithms (see pages 14-15-16 for definitions; see Mattoli L, Gianni M, Burico M. Mass spectrometry-based metabolomics analysis as a tool for quality control of natural complex products. Mass Spectrom Rev. 2022 Mar 3:e 21773. doi: 10.1002 / mas.21773. Epub ahead of print. PMID: 35238411; see pages 430-431; see User Guide to SIMCA 15 By Sartorius (https: / / www.sartorius.com / download / 544940 / simca-15-user-guide-en-b-00076-sartorius-data.pdf; see below: Stocchero M., Riccadonna S., Franceschi P., Projection to latent structures with orthogonal constraints for metabolomics data, Journal of Chemometrics. 2018;e 2987. https: / / doi.org / 10.1002 / cem.2987).

[0159] 3.1 Integration of NIR fingerprint data and adhesion data using OPLS correlation. Fingerprint data was preprocessed using the SNV (Standard Normal Variable) algorithm and scaled using the MEAN-CENTERING function. Data related to adhesion activity measurements were transformed using the exponential function 1 / 4 (or 0.25) and scaled using the UV (Unit Variance) function. Generally, the transformation algorithm can be an exponential or logarithmic function such as Log, NegLog, Logit, or power. Other transformations can be evaluated with the aim of obtaining normally distributed variables.

[0160] The variables processed in this way become the model's diagnostic parameter R 2 and Q 2 It was evaluated by calculating this R 2 and Q 2 The results were very good, with values ​​of 0.996 and 0.984, respectively. Therefore, this model consists of one parallel component and two orthogonal components, as seen in Figure 1, and is also written as [1+2] (the diagnostic parameter R of the model for one parallel component and two orthogonal components). 2 and Q 2 (OPLS_NIR & Effetto sticker_Graph).

[0161] The graph in Figure 2a shows the experimental values ​​of adhesion to an inclined surface on the vertical axis and their predicted values ​​on the horizontal axis. This graph shows the regression line representing the experimental data. In an ideal line, if the regression coefficient is 1, the predicted values ​​will be equal to the experimental values. Theoretical data always have a deviation compared to experimental values, so the regression coefficient R 2 By calculating the root mean square error of the estimated values, and the RMSEE (root mean square error of the estimated values), it is possible to predict the fit of the model and predict the deviation of the predicted values ​​compared to the experimental values.

[0162] The model shown in the graph of Figure 2a results in high correlation between samples, and therefore high prediction, and R 2 The R is very high, >0.99, and the RMSEE (Root Mean Square Error of Estimate) is 2.0 on a scale up to 70. In Figure 2b, the model is used to predict the adhesion effect of the TARGET sample to the inclined surface from the NIR fingerprint. As shown in the figure, R 2 The accuracy is very high, >0.99, and the mean squared error of the prediction, RMSEP (root mean squared error of the prediction), is 3.1 on a scale up to 70, indicating that the prediction is excellent.

[0163] In summary, Figure 2 shows an OPLS scatter plot of the NIR fingerprint data integrated with the adhesion values ​​to the plane, where a correlation is observed between the experimental value of the adhesion effect (vertical coordinate) and the predicted value of the adhesion effect (horizontal coordinate) (Figure 2a), clarifying the prediction of the target adhesion effect from the NIR fingerprint (Figure 2b).

[0164] The graph in Figure 2 shows that when predicting samples obtained from new samples, such as manufacturing batches or samples obtained from considering changes in product composition, the error is very small, indicating that the effect estimates are extremely accurate.

[0165] 3.2 Integration of NIR fingerprint data and barrier effect data using OPLS correlation Fingerprint data was preprocessed using the SNV (Standard Normal Value) algorithm and scaled using the MEAN-CENTERING function. Data related to barrier effect measurements were transformed using the squared exponential function and scaled using the UV (Unit Variance) function. Generally, the transformation algorithm can be an exponential or logarithmic function such as Log, NegLog, Logit, or power. Other transformations can be evaluated with the aim of obtaining normally distributed variables.

[0166] The variables processed in this way become the model's diagnostic parameter R 2 and Q 2 It was evaluated by calculating this R 2 and Q 2 The results were very good, with values ​​of 0.994 and 0.954, respectively. Therefore, this model consists of one parallel component and four orthogonal components, as seen in Figure 3, and is also written as [1+4] (the diagnostic parameter R of the model for one parallel component and four orthogonal components). 2 and Q 2 (OPLS_NIR & Effet barrier_Graph).

[0167] In the graph in Figure 4a, the vertical axis represents the experimental values ​​of the barrier effect, and the horizontal axis represents their predicted values. This graph shows the regression line representing the experimental data. In an ideal line, if the regression coefficient is 1, the predicted values ​​will be equal to the experimental values. Theoretical data always have a deviation compared to experimental values, so the regression coefficient R 2 By calculating the root mean square error of the estimated values, and the RMSEE (root mean square error of the estimated values), it is possible to predict the fit of the model and predict the deviation of the predicted values ​​compared to the experimental values.

[0168] The model shown in the graph in Figure 4a results in high correlation between samples, and therefore high prediction, and R 2 The R is very high, >0.99, and the RMSEE (Root Mean Square Error of Estimate) is 1.0 on a scale of 100. In Figure 4b, the model is used to predict the barrier effect of the TARGET sample from the NIR fingerprint. As shown in the figure, R 2 The accuracy is very high, >0.99, and the mean squared error of the prediction, RMSEP (root mean squared error of the prediction), is 3.7 on a scale of 100, indicating that the prediction is excellent.

[0169] In summary, Figure 4 shows an OPLS scatter plot of NIR fingerprint data integrated with barrier effect values, where experimental values ​​of the barrier effect (vertical coordinate) are shown in comparison with predicted values ​​of the barrier effect (horizontal coordinate) (Figure 4a), clarifying the prediction of the target's barrier effect from the NIR fingerprint (Figure 4b).

[0170] The graph in Figure 4 shows that when predicting samples obtained from new samples, such as manufacturing batches or samples obtained from considering changes in product composition, the error is very small, indicating that the effect estimate is extremely accurate.

[0171] 3.3. Integration of UHPLC-qToF fingerprint data and antioxidant activity data using OPLS correlation. UHPLC-qToF fingerprint data were scaled using the PARETO and MEAN-CENTERING functions to obtain a normal distribution of the variables. Data related to antioxidant activity measurements were not transformed, only scaled using the UV (unit variance) function. Generally, transformation algorithms can be exponential or logarithmic functions such as Log, NegLog, Logit, or power. Other transformations can be evaluated with the aim of obtaining normally distributed variables.

[0172] The variables processed in this way become the model's diagnostic parameter R 2 and Q 2 It was evaluated by calculating this R 2 and Q 2 The results were very good, with values ​​of 0.959 and 0.929, respectively. Therefore, this model consists of one parallel component and zero orthogonal components, as seen in Figure 5, and is also written as [1+0] (the diagnostic parameter R of the model for one parallel component and zero orthogonal components). 2 and Q 2 (OPLS UHPLC-qToF & Antioxidant effect graph).

[0173] The graph in Figure 6a shows the experimental values ​​of antioxidant activity on the vertical axis and their predicted values ​​on the horizontal axis. This graph shows the regression line representing the experimental data. In an ideal line, if the regression coefficient is 1, the predicted values ​​will be equal to the experimental values. Theoretical data always have a deviation compared to experimental values, so the regression coefficient R 2 By calculating the mean squared error and RMSEE (root mean squared error of the estimate), it is possible to predict the fit of the model and predict the deviation of the predicted values ​​compared to the experimental values.

[0174] The model shown in the graph in Figure 6a results in high correlation between samples, and therefore high prediction, and R 2The value is very high, >0.95, and the RMSEE (root mean square error of the estimate) is 297 on a scale up to 4000. In Figure 6b, the model is used to predict the antioxidant activity of the TARGET sample from the UHPLC-qToF fingerprint. As shown in the figure, R 2 The value is very high, >0.95, and the mean squared error of the prediction, RMSEP (root mean squared error of the prediction), is 347 on a scale up to 4000, so the prediction is very good.

[0175] In summary, Figure 6 shows an OPLS scatter plot of surface adhesion values ​​and integrated UHPLC-qToF fingerprint data, where a correlation is observed between experimental values ​​of antioxidant activity (vertical coordinate) and predicted values ​​of antioxidant activity (horizontal coordinate) (Figure 6a), clarifying the prediction of Target's antioxidant activity from UHPLC-qToF fingerprints (Figure 6b).

[0176] As shown in the graph in Figure 6, when predicting the effects of new samples, such as those obtained from manufacturing batches or product composition changes, the estimated values ​​are extremely accurate because the error is very small.

Claims

1. A method for evaluating the properties of a product having therapeutic and / or health-promoting properties, wherein the product comprises a complex natural system, and the method is: a. A step of performing qualitative and quantitative fingerprint analysis of a standard formulation of the said product and a formulation different from the said standard formulation, wherein the difference from the standard formulation is due to the presence of the same component of the same quality in different proportions, or the presence of the same component of different quality in the same proportions, thereby obtaining a block, b. A step of analyzing one or more biophysical properties of a standard formulation of the product and a formulation different from the standard formulation, wherein the difference from the standard formulation is due to the presence of the same component of the same quality in different proportions, or the presence of the same component of different quality in the same proportions, thereby obtaining block Y of biophysical property data. c. A step of integrating data based on correlation using multivariate analysis, which involves the correlation between data block X and data block Y, integrated based on the correlation between each data obtained in a. and each data obtained in b., thereby obtaining a regression line for each pair of values ​​and its regression coefficient R 2 And calculate the mean squared error RMSEE, and R 2 By selecting data pairs such that the ratio is ≥0.8, a model is provided that enables the prediction of the respective biophysical properties of standard formulations from the fingerprints used, in the above step. The method, including the method described above.

2. A method for verifying the manufacture of one or more batches of products having therapeutic and / or health properties, including complex natural systems, the method being: a. A step of performing qualitative and quantitative fingerprint analysis of a standard formulation of the said product and a formulation different from the said standard formulation, wherein the difference from the standard formulation is due to the presence of the same component of the same quality in different proportions, or the presence of the same component of different quality in the same proportions, thereby obtaining X blocks of fingerprint data, b. A step of analyzing one or more biophysical properties of a standard formulation of the product and a formulation different from the standard formulation, wherein the difference from the standard formulation is due to the presence of the same component of the same quality in different proportions, or the presence of the same component of different quality in the same proportions, thereby obtaining a Y block of biophysical property data. c. A step of integrating data based on correlation using multivariate analysis, which involves the correlation between data block X and data block Y, integrated based on the correlation between each data obtained in a. and each data obtained in b., thereby obtaining a regression line for each pair of values ​​and its regression coefficient R 2 And calculate the mean squared error RMSEE, and R 2 By selecting data pairs such that the ratio is ≥0.8, a model is provided that enables the prediction of the respective biophysical properties of standard formulations from the fingerprints used, in the above step. To predict the biophysical properties described in db, the regression model developed in c. is used, and qualitative and quantitative fingerprint analysis is performed on one or more batches of the product to be validated using the same method as used in a. The method, including the method described above.

3. The method according to claim 1 or 2, wherein the qualitative and quantitative fingerprint analysis is performed by recording NIR spectra, recording IR spectra, or by acquiring chromatograms from a chromatography system coupled to a mass spectrometer (e.g., gas chromatography, liquid chromatography, and qToF, single quadrupole, triple quadrupole, ion trap), or by mass spectrometry by direct injection into the mass spectrometer, similarly by chromatography systems, NMR coupled to liquid chromatography, UV-Vis spectroscopy.

4. The method according to any one of claims 1 to 3, wherein one or more of the aforementioned biophysical properties are selected from mucosal adhesion properties, barrier effect, antioxidant effect, neutralizing activity, moisturizing properties, and lubricity.

5. The method according to any one of claims 1 to 4, wherein the integration and correlation in item c are carried out using PLS, OPLS, or PLS2.

6. The method according to any one of claims 1 to 5, wherein the product having therapeutic properties, including a complex natural system, takes the form of a medical device composed of natural substances, and the product having health properties takes the form of a nutritional supplement composed of natural substances.

7. The method according to any one of claims 1 to 6, wherein the natural substance is selected from parts of plants, powders of such parts, extracts of plants or parts thereof (such as leaves, flowers, roots, bark, stems, seeds, fruits, petioles, parts of flowers, etc.), fractions of such extracts, essential oils, natural gums, natural resins, minerals, honey, propolis, animal substances, and mineral substances.

8. The method according to any one of claims 1 to 7, wherein the product takes the form of a syrup, tablet, capsule, lozenge, granule, powder, solution, suspension, aerosol, aerosol powder, powder sachet, granule sachet, oper column, spray, cream, emulsion, soft gelatin capsule, or hard gelatin capsule.

9. The aforementioned R 2 The method according to any one of claims 1 to 8, wherein is ≥ 0.

9.

10. The method according to any one of claims 1 to 9, wherein the integration by correlation of the data in item c is performed using OPLS.

Images