Formula design and evaluation method based on AHP, SOM and TOPSIS

Abstract

The present invention discloses a recipe design and evaluation method based on AHP, SOM and TOPSIS. First, a matrix is ​​established for the collected and collated data based on hierarchical evaluation indicators such as the category of raw materials and the data source, and the weights are calculated. Then, SOM cluster analysis is performed on the weighted raw material data. Finally, a recipe is formed based on the clustering results, and TOPSIS analysis is performed to obtain the optimal solution. The present invention adopts a combined use of AHP, SOM clustering and TOPSIS, which have been used separately or in pairs in the past, to establish a method for recipe design and evaluation, analyze the raw materials in the database from multiple levels and angles, and dig out raw materials that may not have been fully used in previous application research.

Description

Technical Field

[0001] The present invention relates to application scenarios of formulation design and evaluation, and in particular to a formulation design and evaluation method based on AHP, SOM and TOPSIS. Background Art

[0002] The general approach used in formula design is to analyze the frequency of existing raw materials using algorithms such as association rules and entropy clustering to determine matching combinations. These statistical methods have been maturely applied in the fundamental stages of health food formulation screening, such as identifying core ingredients, screening for related drug pairs, and even combining possible formulations. However, these methods are relatively weak in terms of the application patterns between raw materials and formula evaluation. Furthermore, they are time-consuming, inefficient, and have limited accuracy, making them increasingly difficult to meet the current demand for personalized health foods, such as precise nutrition and precise solutions.

[0003] In the past two years, machine learning algorithms such as association rules, decision analysis, and cluster analysis have been increasingly applied to areas such as assisted medical care, drug discovery, and medical image processing. The Analytic Hierarchy Process (AHP) and the Technique for Order Preference by Similarity to Solution (TOPSIS) are two commonly used decision analysis methods. AHP uses a pairwise comparison method to construct a hierarchical matrix and ultimately calculates weighted values. While AHP offers advantages such as minimal data requirements, indirect use, and strong systematicity, it also suffers from the drawbacks of simply performing a weighted average of data to obtain a comprehensive evaluation value, ignoring the characteristics of the actual indicator values. Therefore, it must be combined with other decision-making methods in practice. TOPSIS is a comprehensive evaluation algorithm that uses a normalized raw data matrix to identify the optimal and worst options among a limited number of options and then calculates the relative proximity of the evaluation object to the optimal and worst options. SOM clustering (Self-Organizing Map) is an unsupervised, self-organizing neural learning network that automatically discovers inherent patterns and essential attributes within samples, self-organizing and adaptively changing network parameters and structure. It is often used to solve problems such as the classification of fuzzy information. Previously, it was primarily used in fields such as engineering and finance, with limited application in pharmacy or food. However, its application in conjunction with AHP and TOPSIS analysis has diverse and mature research applications. The combined application of AHP-SOM clustering is often used to select from multiple options or to establish a quality evaluation system using both qualitative and quantitative indicators. In comparison, the combined use of SOM clustering and TOPSIS can significantly reduce the number of objects in the final selection and evaluation. By combining these algorithms, it can provide support for the relationships between ingredients and the compatibility of various formulations, such as prescriptions and health foods, allowing for formulation evaluation and ultimately the selection of the most suitable formulation. Summary of the Invention

[0004] The purpose of the present invention is to make up for the situation that most formulas are simply pieced together with raw materials with the same function, and the frequency of use of some raw materials with the same function is too high, resulting in insufficient utilization of other raw materials. A method for formula design and evaluation with a certain raw material as the core is provided by combining AHP, SOM clustering and TOPSIS.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] A formulation design and evaluation method based on AHP, SOM and TOPSIS includes the following steps:

[0007] Step 1: Complete the collection of information on the API being studied by collecting approval information from the State Administration for Market Regulation and various traditional Chinese medicine information databases, including but not limited to Yaozhi Data and TCM Family.

[0008] Step 2: Standardize the data collected in step 1, and count the categories and frequencies of the raw materials on the standardized data;

[0009] Step 3: Establish hierarchical evaluation indicators based on the categories of raw materials and the sources of raw material data considered when establishing the formula;

[0010] Step 4: Create a matrix for the collected data according to the evaluation indicators established in step 3, score them, and calculate the weights;

[0011] Step 5: Perform SOM cluster analysis on the data obtained in step 4;

[0012] Step 6: Based on the functional research of the preferred raw materials shown by the clustering results in step 5, literature search is carried out, and several formulas are formulated with the support of traditional Chinese medicine compatibility theory;

[0013] Step 7: Perform TOPSIS analysis on the formula in step 6 according to the order of monarch, minister, assistant and envoy to obtain the optimal solution.

[0014] Preferably, the formula is a health food formula or composition.

[0015] Furthermore, in the above step 1, the data to be collected and the collection method are as follows:

[0016] Health food approval data: product name, approval number, function and main ingredients;

[0017] Data related to Chinese patent medicines: drug name, prescription source, prescription raw material information and dosage information;

[0018] Data related to Chinese herbal prescriptions: prescription name, origin, prescription composition, classification, function, and main indications.

[0019] The formula components are indispensable, and if other items are missing, they will be recorded as blank.

[0020] Furthermore, certain sorting methods in step 2 include but are not limited to:

[0021] The formula and medicinal information of prescriptions, Chinese patent medicines and health food approval documents are standardized with reference to the Chinese Pharmacopoeia and local standards.

[0022] (1) Change the commonly used name to the standardized name recorded in the pharmacopoeia or local standards;

[0023] (2) Changing the different preparation specifications of the same medicine to the original name;

[0024] (3) Changing the name of the medicinal material extract to the original medicinal material name;

[0025] (4) Merge data with different prescription names, Chinese patent medicine names, and different approval documents, but with consistent medicinal composition and efficacy; in addition, during the sorting process, merge approval documents with the same medicinal composition but different excipients;

[0026] (5) If the name change during the standardization process significantly affects the efficacy, the original records will be retained and entered separately.

[0027] The standardized data is used to count the categories of raw materials and the frequency of occurrence of raw materials.

[0028] Furthermore, in step three, the hierarchical evaluation indicators include: 2 levels with a total of 17 indicators are designed based on the principles of health food formula composition. The levels include medicinal flavor categories and existing prescription types. The indicator items are set according to the frequency statistical results in step two.

[0029] Furthermore, the AHP evaluation matrix mentioned in step 4 is established as follows:

[0030] With a ij Represents the selected index x that affects the selection of raw materials for health food formula i and the j-th index x j The importance ratio of a ji Represents the jth index x that affects the selection of raw materials for health food formula j and the i-th indicator x i The importance ratio of ij =1 / a ji , all comparison results are expressed in matrix A = (a ij ) n×n Indicates the maximum eigenvalue λ of A max The corresponding normalized eigenvector S = [s1, s2, ..., s j ,…s n ] T As the weight distribution vector of multiple decision-making indicators that affect the selection of health food formula raw materials, s j represents the weight of the jth indicator affecting the selection of health food formula raw materials in the decision-making system of the optimal health food formula scheme, and n represents the number of decision-making indicators affecting the selection of health food formula raw materials selected;

[0031] According to the formula, the consistency index CI is calculated to measure the degree of consistency of the judgment matrix;

[0032] Calculate the consistency ratio CR = CI / RI, where the value range of RI is between 0 and 2. Within this range, manually select the RI value corresponding to n to calculate the CR;

[0033] When CR<0.10, the consistency of the judgment matrix A is considered acceptable. Otherwise, the judgment matrix is ​​appropriately modified and adjusted, and the weight distribution vector of the decision indicator is recalculated, that is, step 4 is repeated.

[0034] As a preferred embodiment, each element in the matrix is ​​evaluated using the following method:

[0035] The relative importance of two adjacent elements can be assigned a value from 1 to 9, with 1 being equally important, 3 being slightly important, 5 being significantly important, 7 being strongly important, and 9 being extremely important. 2, 4, 6, and 8 are the median values ​​for these adjacent values.

[0036] The weight ranking of the health food raw materials is evaluated using a linear weighted method:

[0037]

[0038] And there is

[0039] Where, P is the weight value of health food raw materials obtained through AHP analysis, A i is the quantitative value of the i-th indicator, W i is the weight coefficient of the i-th indicator, and n is the number of indicators.

[0040] Furthermore, in step five, SOM cluster analysis was performed based on the weighted data of each level obtained in step four, using the SOM clustering algorithm provided by MATLAB software, and the number of neurons and the number of tests were adjusted accordingly.

[0041] Furthermore, the supporting literature in step 6 is research literature with clear functional roles. The search is conducted on CNKI and Web of Science using "ingredient name" and "function name".

[0042] Furthermore, the TOPSIS analysis data in step seven are arranged according to the five major items of the formula's monarch, minister, adjuvant and guiding ingredients and the sum of the formula's raw materials, and the numerical values ​​are obtained by linear weighting after AHP analysis.

[0043] Normalize the obtained data: Among them, Z represents the matrix to be evaluated, i and j represent the number of rows and columns of the matrix respectively, and x ij It is represented as the data in row i and column j in the matrix, and n represents the number of evaluation objects.

[0044] The normalized matrix statistics are used to obtain the maximum and minimum values, and the positive and negative ideal solutions are calculated to complete the scoring and evaluation of each formulation scheme.

[0045] Definition of maximum value:

[0046] Z + =(max{z 11 , z 21 ,...,z i1},max{z 12 , z 22 ,...,z i2}, ..., max{z 1j , z 2j ,...,z ij}).

[0047] Definition of minimum value:

[0048] Z - =(min{z 11 , z 21 ,...,z i1},min{z 12 , z 22 ,...,z i2}, ..., min{z 1j , z 2j ,...,z ij}).

[0049] Define the distance between the i-th evaluation object and the maximum value:

[0050]

[0051] Define the distance between the i-th evaluation object and the minimum value:

[0052]

[0053] The score is:

[0054]

[0055] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0056] 1. The present invention optimizes and improves the general method of formula composition, uses the analytic hierarchy process (AHP) to determine the weights of factors affecting the selection of raw materials for health food formulas, performs weighted analysis on the raw materials collected from the database, and obtains a raw material selection ranking. Then, based on the weighted data, a SOM cluster analysis is performed with a specific raw material as the center to obtain a set of raw materials that are more likely to be used in combination with this raw material. Then, based on the functional literature support of the raw material and the theory of compatibility of traditional Chinese medicine, several formulas with this raw material as the core are composed, and TOPSIS is used to evaluate and score, thereby finally obtaining a possible health food formula.

[0057] 2. The present invention adopts a combined approach to AHP, SOM clustering and TOPSIS, which were previously used separately or in pairs, to establish a method for formula design and evaluation, analyze the raw materials in the database from multiple levels of perspective, and discover raw materials that may not have been fully used in previous application research. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 Analysis Process

[0059] Figure 2 Schematic diagram of the direct distance results of the weights between the SOM clustering neighboring high-frequency raw materials

[0060] Figure 3 Schematic diagram of the distance results of the weights between the SOM clustering neighboring high-frequency raw materials DETAILED DESCRIPTION

[0061] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0062] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0063] Example 1 Health Food Formula Screening and Evaluation Results

[0064] See the analysis flow chart for Figure 1 A health food formula design and evaluation method based on AHP, SOM and TOPSIS comprises the following steps:

[0065] Step 1: Complete the collection of information on the APIs being studied by collecting data from health food approval documents compiled by the State Administration for Market Regulation and various traditional Chinese medicine information databases, including but not limited to Yaozhi Data and TCM Family.

[0066] When collecting and collating prescriptions and Chinese patent medicines, we excluded prescriptions with some identical Chinese herbal ingredients, efficacy, or indications, as well as those containing only Bletilla striata. Information on health foods was limited to those containing two or more Chinese herbal ingredients, and we excluded approval documents without formula information.

[0067] Step 2: Perform frequency statistics and other data sorting on the data collected in step 1 according to a certain sorting method;

[0068] The collected data were imported into Microsoft Excel 2019 to establish four databases. During the data processing, the formula information of prescriptions, Chinese patent medicines and health food approval documents was standardized with reference to the Chinese Pharmacopoeia and local standards: (1) the commonly used names were changed to the standardized names recorded in the pharmacopoeia or local standards, such as "Baiji" was unified as "Baiji" and "Huangqi" was unified as "Huangqi"; (2) different preparation specifications of the same medicine were changed to the original name, such as "Stir-fried Baizhu" was unified as "Baizhu" and "Propolis powder" was unified as "Propolis"; (3) the medicinal material extracts were changed to the original name, such as "Kuang root extract" was unified as "Kuang root" and "Patchouli oil" was unified as "Patchouli"; (4) data with different prescription names, Chinese patent medicine names and different approval documents but consistent medicinal composition and efficacy were merged. In addition, in the process of sorting out Chinese herbal health foods, the approval documents for the same medicinal flavors but different excipients will be merged; (5) In the process of standardization, if the efficacy of a product is significantly affected after the name is changed, the original record will be retained and entered separately, such as "Gardenia charcoal" and "Gardenia", "Shengdihuang" and "Shudihuang", etc.

[0069] Step 3: Establish hierarchical evaluation indicators based on factors such as the category of raw materials and the source of raw material data considered when establishing the formula of traditional Chinese medicine health foods;

[0070] The levels and indicators used for AHP analysis are entered according to the content and statistical items in the raw material database. The raw material medicinal flavors involved in the statistics include all high-frequency medicinal flavors in health foods, Chinese patent medicines and prescriptions that have the function of assisting in protecting the gastric mucosa and assisting in protecting chemical liver damage. The first-level evaluation indicators are the category of medicinal flavors and existing prescriptions, and the second-level evaluation indicators are the types of existing prescriptions, including health foods with the function of assisting in protecting the gastric mucosa, health foods with the function of assisting in protecting chemical liver damage, Chinese patent medicines containing Bletilla striata with the function of tonifying the spleen and replenishing qi, and prescriptions containing Bletilla striata with the function of tonifying the spleen and replenishing qi, a total of 4. By counting the frequency of high-frequency medicinal flavors in each library and the number of libraries appearing in the four libraries, the statistical table is formed in descending order and assigned a score of 1-9.

[0071] Step 4: Score the collected data according to the evaluation indicators established in step 3 and calculate the weight;

[0072] Based on the overall goal of developing a dual-function health food with stomach and liver protection, centered around Bletilla striata, and the core theory of Traditional Chinese Medicine (TCM) that treats stomach and liver disorders by tonifying the spleen and replenishing Qi, we determined the relative importance of elements within each hierarchy and constructed a pairwise comparison matrix. The importance of two adjacent elements was assigned a value from 1 to 9, with 1 indicating equal importance, 3 slightly important, 5 significantly important, 7 strongly important, and 9 extremely important. 2, 4, 6, and 8 were the median values ​​for these adjacent values.

[0073] The high-frequency medicinal flavors were further analyzed based on the weights obtained after AHP analysis. Each medicine was weighted and scored according to the linear weighted formula. Missing indicators were recorded as -1 points, and the medicines were ranked according to their final scores. The results are shown in Table 1 below.

[0074] Table 1 Total weight results and ranking of AHP analysis of high-frequency APIs

[0075]

[0076]

[0077] Step 5: Perform SOM cluster analysis on the data obtained in step 4. The clustering results are shown in Table 2 below.

[0078] Table 2 SOM cluster analysis results of high-frequency raw material medicinal flavors

[0079]

[0080] Step 6: Based on the research supporting literature on the preferred functions indicated by the clustering results in step 5 and supported by traditional Chinese medicine compatibility theory, formulate several Chinese medicine health food formulas that can be used under the functional items;

[0081] Step 7: Perform TOPSIS analysis on the formula in step 6 according to the order of monarch, minister, assistant and envoy to obtain the optimal solution. The formula and the calculated scoring results are shown in Table 3 below.

[0082] Table 3 SOM cluster analysis results of high-frequency raw material medicinal flavors

[0083]

[0084] Example 2 Pharmacodynamic Verification of Each Scheme

[0085] Alcohol has become one of the greatest threat factors to human health. People who drink alcohol are constantly facing the threats of related diseases caused by drinking. The intake of alcohol may cause various adverse health outcomes to occur simultaneously, such as cardiovascular and cerebrovascular diseases, disorders of the central nervous system, and digestive system diseases such as gastric mucosal injury and liver injury. More than 80% of these diseases are chronic diseases, which are difficult to cure. The consumption of alcohol is huge globally, and the number of consumers is large. Therefore, it is extremely important to control the occurrence and development of related diseases through external auxiliary measures such as taking health foods and enhancing exercise to improve one's own immunity from the initial stage of illness or from the sub-healthy and healthy states without illness. Based on this, liver injury and gastric mucosal injury with a high prevalence rate and difficult to cure are used as the pharmacodynamic verification indicators for the formulation of the AHP-SOM clustering-TOPSIS algorithm.

[0086] In the plan, each medicine is weighed according to an equal ratio dose. Take 15 g of each raw material and place it in a round-bottom flask, add 600 mL of water, and reflux and extract it in a water bath for 2 hours. Extract it 2 times in total. After combining the filtrates, concentrate them to obtain an extract for pharmacodynamic experiments.

[0087] 96 male C57BL / 6J mice with a body weight of 22 ± 3 g were used in the experiment. They were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. (License No.: SYXK (Beijing) 2016-0006), and were raised under the conditions of 25 ± 2 °C, humidity of 45 ± 5%, and 12 / 12 h light-dark alternating lighting, with free diet and water. After 1 week of adaptive feeding, the mice were randomly divided into 12 groups, with 8 mice in each group. The drugs were administered according to the following grouping: Normal group, Model group: gavaged with 0.2 mL of distilled water once a day for 7 days. Each drug administration group of the plan: gavaged with 0.2 mL of the extract of each plan (1 g / kg) once a day for 7 days. 1 hour after the drug administration ended on the 7th day, the Model group and the drug administration groups were given 10 mL / kg of 50% ethanol by gavage, while the Normal group was given an equal amount of normal saline treatment. After the treatment, all groups were fasted but allowed to drink water for 12 h. Then, blood was collected by eye socket puncture, centrifuged at a speed of 3000 r / min at 4 °C for 10 min, and the serum was separated for subsequent index detection. After decapitation, the stomach was taken, cut open from the greater curvature of the stomach, the length and width of the ulcer were measured, and the ulcer area was calculated. The liver was taken out, weighed, and the liver weight index was calculated. The experimental results are shown in Table 4.

[0088] Table 4 Results of the stomach-protecting and liver-protecting pharmacodynamic experiments of each plan

[0089]

[0090]

[0091] Note: Compared with the blank group, * P < 0.05, ** P < 0.01

[0092] Compared with the blank control group, the model group, the positive drug group, and each regimen group showed an increase in the gastric ulcer index and liver weight index, as well as elevated levels of SOD, MDA, ALT, and AST. With the exception of the ulcer index in regimens 8-10, the differences in all other parameters were statistically significant (P < 0.05). Silymarin and omeprazole, as positive drugs for liver injury and gastric mucosal injury, respectively, demonstrated some therapeutic efficacy in both. Furthermore, as the regimen score decreased in the TOPSIS analysis, its gastroprotective and hepatoprotective efficacy also decreased, demonstrating a positive correlation with changes in indicators of gastric mucosal and chemical liver injury.

[0093] Formula compatibility is a key principle of traditional Chinese medicine. Formulas are formulated according to the principle of "monarch, minister, adjuvant, and envoy" to enhance efficacy. The raw material ingredients for each recipe are calculated using the algorithm described above, and the respective roles of these ingredients in the recipe are determined based on their contribution to the recipe. The monarch ingredient plays a decisive role in both the calculation and validation results, playing a leading role in the efficacy of the recipe. For example, in recipes 1 and 2, the monarch ingredient is Astragalus membranaceus. Although its weight ranking is far lower than that of Bletilla striata, it scores significantly higher than the other recipes in the final TOPSIS analysis. In efficacy validation trials, its effectiveness in protecting the gastric mucosa and preventing chemical liver damage is also more pronounced than in recipes 3-10, which use Bletilla striata as the monarch ingredient. This suggests that Astragalus membranaceus, as the monarch ingredient, can significantly enhance the formula's gastric and liver protection. The minister ingredient's contribution to the recipe's efficacy is second only to that of the monarch ingredient, and it can be used in combination with the monarch ingredient to enhance its efficacy, helping it to achieve greater results. However, this role is not currently reflected in the recipes. In a compound formula, adjuvants and guiding herbs act as "messengers" that assist the monarch and minister herbs, guiding the compound's efficacy. Their individual contributions to the overall efficacy of the formula may not be significant, but when used in combination with the monarch and minister herbs, they can enhance the overall efficacy of the formula. Due to the differences in the use of astragalus and poria as adjuvants in Schemes 3 and 4, there are some differences in TOPSIS analysis scores and pharmacodynamic experiments. This also suggests that the selection of adjuvants and guiding herbs in a combination can, to a certain extent, influence the final efficacy of the formula.

[0094] In summary, the monarch drug is significantly different from other APIs in terms of dosage and efficacy, primarily due to its prominent efficacy, and generally contributes the most to the formulation. Assistant drugs, acting as auxiliary drugs to the monarch drug, generally contribute the most to the formulation's efficacy. Assistant drugs, acting as adjuvants, assist the monarch and assistant drugs in achieving therapeutic results and generally contribute the least to the formulation. The combined use of AHP, SOM clustering, and TOPSIS can achieve the goal of designing and evaluating health food formulas for any function or centered around a particular ingredient.

[0095] With the above-described preferred embodiments of the present invention as a guide, and with reference to the above description, relevant personnel are fully capable of making various changes and modifications without departing from the technical scope of this invention. The technical scope of this invention is not limited to the contents of the specification and must be determined according to the scope of the claims.

Claims

1. A formulation design method based on AHP, SOM and TOPSIS, characterized in that: The specific steps include: Step 1: Complete the collection of information on the APIs being studied by collecting approval information collected by the State Administration for Market Regulation and data from various information databases including but not limited to Yaozhi Data and TCM Family. Specifically, the APIs include: Bletilla striata, licorice, dried tangerine peel, astragalus, poria, schisandra chinensis, salvia miltiorrhiza, white peony root, fritillaria, atractylodes macrocephala, codonopsis pilosula, ginseng, costus root, yam, angelica root, coix seed, jujube, lily, oriental water plantain, clove, amomum villosum, hawthorn, evodia rutaecarpa; Step 2: Standardize the data collected in step 1, and count the categories and frequencies of the raw materials on the standardized data; Step 3: Establish hierarchical evaluation indicators based on the categories of raw materials and the sources of raw material data considered when establishing the formula; Step 4: Create a matrix for the collected data according to the evaluation indicators established in step 3, score them, and calculate the weights; Step 5: Perform SOM cluster analysis on the data obtained in step 4; Step 6: Based on the functional research of the preferred raw materials shown by the clustering results in step 5, literature search is carried out, and several formulas are formulated with the support of traditional Chinese medicine compatibility theory; Step 7: Perform TOPSIS analysis on the formula in step 6 according to the order of monarch, minister, assistant and envoy to obtain the optimal solution.

2. The formulation design method according to claim 1, wherein A hierarchical evaluation index is established based on the category of raw materials and the source of raw material data. The evaluation index includes: Evaluation index 1: the category of Chinese medicine to which the raw materials belong, including secondary indicators: tonic, dampness-removing, qi-regulating, and heat-clearing drugs, derived from the frequency statistics data in step 2; Evaluation indicator 2: Raw material source prescription type, including secondary indicators: formula with corresponding function, Chinese patent medicine and prescription containing the raw material.

3. The formulation design method according to claim 1, wherein: According to the evaluation indicators established in step 3, a matrix is ​​created for the collected data, and scores are given, and weights are calculated, including: With a ij Indicates the selected index x that affects the selection of formula raw materials i and the j-th index x j The importance ratio of a ji Indicates the jth index x that affects the selection of formula raw materials j and the i-th indicator x i The importance ratio of ij =1 / a ji , all comparison results are expressed in matrix A = (a ij ) n×n Indicates the maximum eigenvalue λ of A max The corresponding normalized eigenvector S = [s1, s2, ..., s j ,…s n ] T As the weight distribution vector of multiple decision-making indicators that affect the selection of formula raw materials, s j represents the weight of the jth indicator affecting the selection of formula raw materials in the optimal formula solution decision system, and n represents the number of decision indicators affecting the selection of formula raw materials selected; According to the formula, the consistency index CI is calculated to measure the degree of consistency of the judgment matrix; Calculate the consistency ratio CR=CI / RI, Among them, the value range of RI is between 0 and 2, and the RI value corresponding to n is manually selected within this range to calculate CR; When CR<0.10, the consistency of the judgment matrix A is considered acceptable. Otherwise, the judgment matrix is ​​appropriately modified and adjusted, and the weight distribution vector of the decision indicator is recalculated, that is, step 4 is repeated; Evaluate each element in the matrix using the following method: The importance of two adjacent elements can be assigned a score of 1-9; 1 means equally important, 3 means slightly important, 5 means obviously important, 7 means strongly important, and 9 means extremely important. 2, 4, 6, and 8 are the medians of the above adjacent judgments; The raw materials are ranked by weight using a linear weighted approach: And there is Where, P is the weight value obtained by AHP analysis of raw materials, A i is the quantitative value of the i-th indicator, W i is the weight coefficient of the i-th indicator, and n is the number of indicators.

4. The formulation design method according to claim 1, wherein: Based on the functional research of the preferred raw materials shown in the clustering results in step 5, literature search was carried out, and with the support of traditional Chinese medicine compatibility theory, several formulas were formed, including: Supporting documents are research documents and have clear functional roles; The search was conducted on China National Knowledge Infrastructure (CNKI) and Web of Science using "raw material name" and "functional name".

Images