Method for screening and identifying traditional Chinese medicine effective components based on data mining and AI

Abstract

The invention discloses a data mining and AI-based traditional Chinese medicine effective prescription screening and identifying method. The method comprises the following steps: S1, screening related literatures of traditional Chinese medicine prescriptions for stroke treatment in an academic journal database; s2, performing name standardization treatment on the traditional Chinese medicinal materials; s3, performing clustering analysis and correlation analysis on the traditional Chinese medicinal materials in the screened traditional Chinese medicinal prescription according to the monarch, minister, assistant and guide law in prescription compatibility; s4, screening core traditional Chinese medicinal materials, determining the correlation among the traditional Chinese medicinal materials in the law of monarch, minister, assistant and guide, and optimizing the compatibility rule to obtain a traditional Chinese medicine prescription for the optimal compatibility relationship of stroke treatment; the invention discovers the rule of treating stroke by traditional Chinese medicine, forms a paradigm method, optimizes the algorithm based on the theory of traditional Chinese medicine, reveals the compatibility rule of traditional Chinese medicinal materials in the traditional Chinese medicine prescription, screens core medicinal materials according to the monarch, minister, assistant and guide method, and obtains a specific paradigm of compatibility of seven traditional Chinese medicinal materials of angelica sinensis, lumbricus, ligusticum wallichii, peach kernel, astragalus membranaceus, safflower and red paeony root for stroke treatment.

Description

Technical Field

[0001] The present invention relates to a screening and identification method for effective Chinese medicine prescriptions, and in particular to a screening and identification method for effective Chinese medicine prescriptions based on data mining and AI. Background Art

[0002] Clinical treatment of stroke requires more alternatives and new natural products (NPs) to overcome the limitations of current stroke drug treatments. This is where NPs can help. Since prehistoric times, humans have used natural extracts from various resources such as plants and animals to alleviate or treat diseases. With the development of biomedicine, NPs have received increasing attention. Most NPs have unique structures and biological effects, which can be used to develop new / alternative drugs.

[0003] Especially in China, TCM is a treasure trove of NPs and has made great contributions in medical theories and ways of thinking. As TCM has been widely used in China throughout history, it has the unparalleled advantage of rich clinical experience. TCM has significant advantages in the comprehensive treatment and overall regulation of multi-site and multi-target diseases such as stroke. Recent studies have shown that TCM has a significant effect on neuroprotection and neurorepair, and active compounds such as resveratrol have been shown to be effective in the treatment of stroke. If more active NPs for stroke can be identified from TCM, it will make stroke drug development more promising.

[0004] Although there are many successful records of using traditional Chinese medicine to treat stroke, there are still some challenges that hinder these traditional Chinese medicines from developing into mature pharmaceutical approaches. First, the use of traditional Chinese medicine to treat diseases has an interesting philosophical background and a long history, but it has been increasingly questioned due to the lack of evidence-based efficacy proven by high-quality trials. Second, traditional Chinese medicine views the human body as a complex dynamic system and uses a dialectical treatment strategy to treat diseases. The variety of methods for treating the same disease has led to inconsistent medical practices, and coupled with the different processing and naming methods of drugs, it is easy to cause confusion. Summary of the invention

[0005] In order to solve the shortcomings of the above-mentioned technologies, the present invention provides a screening and identification method for effective Chinese medicine prescriptions based on data mining and AI.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is: a screening and identification method for effective Chinese medicine prescriptions based on data mining and AI, comprising the following steps:

[0007] S1. Screen relevant literature on traditional Chinese medicine prescriptions for stroke treatment in academic journal databases;

[0008] S2. Standardize the names of Chinese medicinal materials in Chinese herbal prescriptions with oral herbal usage in relevant literature according to the 2020 edition of the Pharmacopoeia of the People's Republic of China;

[0009] S3. Conduct cluster analysis and association analysis on the selected Chinese medicinal materials in the Chinese medicine prescriptions according to the principle of monarch, minister, adjuvant and envoy in the prescription compatibility;

[0010] S4. Screen core Chinese medicinal materials, clarify the correlation between each Chinese medicinal material in the principle of monarch, minister, adjuvant and guiding, optimize the compatibility rules of Chinese medicine prescriptions, and obtain the Chinese medicine prescription with the most optimized compatibility relationship for stroke treatment.

[0011] Furthermore, the academic journal database of S1 was searched separately with the keywords stroke treatment, TCM, and Chinese herbal medicine, and the retrieved data were aggregated and added.

[0012] Further, the following steps are performed on the relevant literature:

[0013] A1. Screen the relevant literature of clinical trials, clinical observations and human controlled experiments, and filter the relevant literature of reviews, viewpoints, systematic reviews, animal experiments, cell experiments, basic research and data research;

[0014] A2. Screen the literature on oral herbal medicine use, and filter the literature on diagnosis, acupuncture, fumigation, and massage;

[0015] A3. Screen relevant literature with detailed descriptions of single prescriptions, and filter relevant literature without detailed ingredients, formulas or with variations;

[0016] A4. Screen relevant literature with significant and / or positive results, and exclude relevant literature with insignificant effects and / or negative results.

[0017] Furthermore, the top twenty Chinese medicinal materials with the highest frequency of occurrence in the Chinese medicinal prescriptions of S3 were selected for cluster analysis, using ML-based clustering algorithms, including the hierarchical clustering (HC) method and the K-modes clustering method.

[0018] Furthermore, the hierarchical clustering (HC) method converts the presence of each Chinese herbal medicine in different Chinese herbal medicine prescriptions into a sparse matrix, and uses the average linkage method to measure the differences between Chinese herbal medicines. The average linkage distance between two clusters is measured as follows:

[0019]

[0020] In the formula, A represents the Chinese medicinal materials in cluster A; B represents the Chinese medicinal materials in cluster B; ∑ i∈A ∑ j∈Bd(i, j) represents the sum of all pairwise distances between points in cluster A and points in cluster B; |A| and |B| represent the number of data points in cluster A and B respectively;

[0021] then,

[0022]

[0023] In the formula, x ik 、x jk They represent the coordinates of data points i and j in the kth dimension respectively; n is the dimension, representing the total number of Chinese medicinal materials; d(i, j) represents the distance between the i-th Chinese medicinal material and the j-th Chinese medicinal material.

[0024] Furthermore, the K-modes clustering method, i.e., a method that initially specifies the number of cluster centers and approximates the optimal clustering result through interactive calculation, identifies combinations of Chinese medicinal materials with similar characteristics by calculating the distance between them and classifies them into the same category, where the distance is the Hamming distance.

[0025] Furthermore, the Hamming distance is defined as the number of mismatches in all feature items for two data points. The Hamming distance calculation formula is used to measure the difference between two discrete data, namely, the medicinal material attribute vectors. The calculation formula is as follows:

[0026]

[0027] This formula indicates whether the i-th feature item is matched, where the feature item refers to the presence of each Chinese medicinal material in different Chinese medicine prescriptions.

[0028] Wherein, x represents the attribute vector of the first Chinese medicinal material; y represents the attribute vector of the second Chinese medicinal material; and n represents the total number of attributes.

[0029] Furthermore, the association analysis in S3 adopts the ML-based Apriori algorithm. In the Apriori algorithm, three standard indicators, support, confidence and lift, are used to quantitatively evaluate the association relationship between Chinese medicinal materials. Support represents the frequency of simultaneous occurrence of certain combinations of Chinese medicinal materials, confidence represents the probability of other medicinal materials appearing when certain medicinal materials are known, and lift represents the degree of influence of the appearance of certain medicinal materials on the appearance of other medicinal materials.

[0030] Furthermore, the calculation formula of the Apriori algorithm is as follows:

[0031]

[0032] In the formula, X represents the antecedent set; Y represents the consequent set; ALL represents all events in the data set; Support(X) represents the support of the medicinal material combination X; Confidence(X→Y) represents the confidence from the medicinal material combination X to Y; represents the improvement of the medicinal material combination X on Y.

[0033] Furthermore, the most optimized traditional Chinese medicine prescription for stroke treatment is composed of seven Chinese medicinal materials: angelica, earthworm, chuanxiong, peach kernel, astragalus, safflower, and red peony root.

[0034] A screening and identification method for effective Chinese medicine formulas based on data mining and AI uses a variety of statistical methods and several machine learning algorithms for overall analysis to discover the rules of Chinese medicine treatment for stroke and form a paradigm method. The algorithm is optimized based on the theory of traditional Chinese medicine. Cluster analysis and association analysis are used to study the combination rules between medicinal materials, revealing the compatibility rules of Chinese medicine formulas. Core medicinal materials are screened according to the principle of monarch, minister, adjuvant and envoy. The optimized Chinese medicine formula for stroke treatment is a specific example of the compatibility of seven Chinese medicines: angelica, earthworm, chuanxiong, peach kernel, astragalus, safflower and red peony root. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 This is a schematic diagram of document collection for the present invention.

[0036] Figure 2 It is a schematic diagram of the usage frequency of the Chinese medicinal materials of the present invention.

[0037] Figure 3 It is the hierarchical clustering branch diagram of the present invention.

[0038] Figure 4 This is a graph showing the relationship between the number of clusters and the distance threshold of the present invention.

[0039] Figure 5 This is a graph showing the clustering results after PCA of the present invention.

[0040] Figure 6 This is a graph showing the results of the association analysis of the present invention. DETAILED DESCRIPTION

[0041] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0042] This embodiment is about the screening and identification method of effective Chinese medicine prescriptions based on data mining and AI. By forming a paradigm approach and combining the theory and clinical practice of traditional Chinese medicine, the algorithm is optimized to better analyze the composition, efficacy and compatibility of Chinese medicine prescriptions, thereby improving the accuracy and reliability of the research.

[0043] The paradigm method of this embodiment includes the following steps:

[0044] S1. Screen relevant literature on traditional Chinese medicine prescriptions for stroke treatment in academic journal databases;

[0045] S2. Standardize the names of Chinese medicinal materials in Chinese herbal prescriptions with oral herbal usage in relevant literature according to the 2020 edition of the Pharmacopoeia of the People's Republic of China;

[0046] S3. Conduct cluster analysis and association analysis on the selected Chinese medicinal materials in the Chinese medicine prescriptions according to the principle of monarch, minister, adjuvant and envoy in the prescription compatibility;

[0047] S4. Screen core Chinese medicinal materials, clarify the correlation between each Chinese medicinal material in the principle of monarch, minister, adjuvant and guiding, optimize the compatibility rules of Chinese medicine prescriptions, and obtain the Chinese medicine prescription with the most optimized compatibility relationship for stroke treatment.

[0048] Example 1

[0049] S1. Screen relevant literature on traditional Chinese medicine prescriptions for stroke treatment in academic journal databases;

[0050] In this embodiment, in order to collect medical records of successful treatment of stroke with Chinese herbal medicine, we prefer Figure 1 The three major academic journal databases shown, namely, China National Knowledge Infrastructure, VIP Information and Wanfang Data, are included in the search as authoritative knowledge bases. It should be noted that this embodiment does not limit the specific selection of academic journal databases; the academic journal databases are searched separately with stroke treatment, TCM and Chinese herbal medicine as keywords, among which TCM refers to Traditional Chinese Medicine, and then the retrieved data are collected and added to obtain a total of 1,160 relevant documents. It should be noted that the data disclosed in this embodiment are relevant documents in the past 10 years (2014-2024).

[0051] On this basis, if Figure 1 As shown, the following steps are performed on the relevant literature:

[0052] A1. Screen relevant literature on clinical trials, clinical observations and human controlled experiments, filter out relevant literature on reviews, viewpoints, systematic reviews, animal experiments, cell experiments, basic research and data research, and after this step, screen the remaining 390 relevant literatures;

[0053] A2. Screening the literature on oral herbal medicine usage, filtering the literature on diagnosis, acupuncture, fumigation, and massage, and after this step, 101 relevant literatures were screened;

[0054] A3. Screen relevant literature with detailed description of single prescriptions, filter relevant literature without detailed ingredients, formula or with variants, and screen the remaining 57 relevant literature after this step;

[0055] A4. Screen relevant literature with significant and / or positive results, and exclude relevant literature with insignificant effects and / or negative results. After this step, 54 remaining relevant literature were screened, including 56 effective Chinese medicine prescriptions.

[0056] According to Figure 2 The frequency of use of the herbs (Chinese medicinal materials) shown shows that there are seven Chinese medicinal materials ranked at the top in frequency. These seven Chinese medicinal materials are Chinese angelica, earthworm, chuanxiong, peach kernel, astragalus, safflower, and red peony root.

[0057] S2. Standardize the names of Chinese medicinal materials in Chinese herbal prescriptions with oral herbal usage in relevant literature according to the 2020 edition of the Pharmacopoeia of the People's Republic of China;

[0058] It should be noted that Chinese medicine prescriptions are usually composed of a variety of herbs / animals, and the dosage of Chinese medicines is often adjusted according to the properties of the herbs or the characteristics of the patient. Therefore, this embodiment only considers whether the materials exist, and does not perform quantitative division;

[0059] The purpose of name standardization in this embodiment is to prevent confusion caused by the processing methods, translation modes, and image evolution of Chinese medicinal materials. The specific standardization relies on external name mapping to ensure that the Chinese medicinal materials used in this embodiment comply with the standards of the 2020 edition of the "Pharmacopoeia of the People's Republic of China" and improve reliability.

[0060] S3. Conduct cluster analysis and association analysis on the selected Chinese medicinal materials in the Chinese medicine prescriptions according to the principle of monarch, minister, adjuvant and envoy in the prescription compatibility;

[0061] In order to reveal the usage patterns of Chinese medicinal materials in these different Chinese medicine prescriptions and the interactions between different Chinese medicinal materials, this embodiment selects the top 20 Chinese medicinal materials with the highest frequency of appearance in Chinese medicine prescriptions for cluster analysis, and adopts ML-based clustering algorithms, including hierarchical clustering (HC) method and K-modes clustering method.

[0062] The hierarchical clustering (HC) method converts the presence of each Chinese herbal medicine in different Chinese herbal medicine prescriptions into a sparse matrix, and uses the average linkage method to measure the differences between Chinese herbal medicines. The average linkage distance measurement formula between two clusters is as follows:

[0063]

[0064] In the formula, A represents the Chinese medicinal materials in cluster A; B represents the Chinese medicinal materials in cluster B; ∑ i∈A ∑ j∈B d(i, j) represents the sum of all pairwise distances between points in cluster A and points in cluster B; |A| and |B| represent the number of data points in cluster A and B respectively;

[0065] Here, the definition of distance follows the classic Euclidean distance:

[0066]

[0067] In the formula, x ik 、x jk They represent the coordinates of data points i and j in the kth dimension respectively; n is the dimension, representing the total number of Chinese medicinal materials; d(i, j) represents the distance between the i-th Chinese medicinal material and the j-th Chinese medicinal material.

[0068] In the average link method disclosed in this embodiment, the distance d(i, j) between different Chinese medicinal materials is calculated, and then the distances between all Chinese medicinal materials are averaged to determine the average link distance between cluster A and cluster B (cluster of medicinal materials, i.e., a combination of multiple medicinal materials). This is helpful in finding similar combinations of medicinal materials and clustering them together.

[0069] Based on the above content, in the hierarchical clustering (HC) of this embodiment, the distance threshold is taken as 4, and then there are 5 clusters. At this time, the 20 Chinese medicinal materials selected in the preamble are arranged in 5 clusters. The 5 clusters are specifically:

[0070] Cluster 1. Angelica and Astragalus;

[0071] Cluster 2. Chuanxiong, peach kernel, red peony root and safflower;

[0072] Cluster 3. Poria cocos;

[0073] Cluster 4. Earth dragon;

[0074] Cluster 5. In addition to the twelve herbs included in clusters 1-4, Figure 3 shown.

[0075] In order to verify the ideality of the distance threshold, such as Figure 4 As shown in the figure, when the distance threshold is 4, the number of clusters is significantly reduced, which indicates that the distance threshold point is ideal.

[0076] The K-modes clustering method, i.e., a method that initially specifies the number of cluster centers and approximates the optimal clustering result through interactive calculation, identifies combinations of Chinese medicinal materials with similar characteristics by calculating the distance between them and classifies them into the same category, where the distance is the Hamming distance.

[0077] The Hamming distance calculation formula is used to measure the difference between two discrete data, namely, the medicinal material attribute vectors. The calculation formula is as follows:

[0078]

[0079] This formula indicates whether the i-th feature item is matched, where the feature item refers to the presence of each Chinese medicinal material in different Chinese medicine prescriptions.

[0080] Wherein, x represents the attribute vector of the first Chinese medicinal material; y represents the attribute vector of the second Chinese medicinal material; and n represents the total number of attributes.

[0081] It should be noted that the Hamming distance calculation is used to measure the difference between two discrete data (attribute vectors of Chinese medicinal materials). By calculating the number of different attributes in the two medicinal material attribute vectors, the similarity or difference between them can be determined, which can be used for clustering of Chinese medicinal materials in the K-modes algorithm, which is conducive to discovering common features between Chinese medicinal materials and providing guidance for the combination of Chinese medicinal materials in Chinese medicine prescriptions.

[0082] In this embodiment, for the K-modes clustering method, when the number of clusters is 2, the clustering performance is optimal, and the distance between the two clusters is wide and obvious; after clustering, PCA, i.e., principal component analysis, is performed to reduce the dimension of the clustering results for proper visualization;

[0083] like Figure 5 The K-modes clustering result after PCA is shown. The boundaries of the two clusters are very far, which shows that the classification is clear. It is worth noting that Figure 5 The cluster on the right, i.e. the cluster inside the oval circle, contains 7 different Chinese medicinal materials. The above Chinese medicinal materials are Figure 2 The Chinese medicinal materials with the highest frequency of use were selected through the above operations to reveal the regularity of Chinese herbal prescriptions for treating stroke.

[0084] Association analysis is used to reveal the co-occurrence rules of Chinese medicinal materials and screen core Chinese medicinal materials. The association analysis adopts the ML-based Apriori algorithm. In this embodiment, the Apriori algorithm uses three standard indicators, namely support, confidence and lift, to quantitatively evaluate the association relationship between Chinese medicinal materials. Support represents the universality of the antecedent set, and in this embodiment, it represents the frequency of simultaneous occurrence of certain Chinese medicinal material combinations; confidence represents the possibility of co-occurrence of the latter set and the antecedent set in the data set, and in this embodiment, it represents the probability that other medicinal materials will also appear when certain medicinal materials are known; lift represents the probability that the latter item increases when a specific antecedent item exists, and in this embodiment, it represents the degree of influence of the appearance of certain medicinal materials on the appearance of other medicinal materials.

[0085] The calculation formula of the Apriori algorithm is as follows:

[0086]

[0087] In the formula, X represents the antecedent set; Y represents the consequent set; ALL represents all events in the data set; Support(X) represents the support of the medicinal material combination X; Confidence(X→Y) represents the confidence from the medicinal material combination X to Y; represents the improvement of the medicinal material combination X on Y.

[0088] In this embodiment, the thresholds of support ≥ 15%, confidence ≥ 80%, and lift ≥ 1.2 are specifically set, and 52 association rules remain, including 2 two-item set association rules and 50 three-item set association rules. The detailed rules are shown in the following table:

[0089]

[0090]

[0091]

[0092]

[0093] After calculating the correlation of common multi-flavor Chinese medicinal materials, we can get a visualization like Figure 6 The results of the association analysis shown in the figure show that the higher the frequency of occurrence, the larger the display size of the herb, e.g. Figure 6 There are 7 core nodes shown, namely Angelica sinensis, earthworm, Chuanxiong, peach kernel, astragalus, safflower, and red peony root. This shows that these seven Chinese medicinal materials are not only used most frequently, but also interact closely with other medicinal materials. Not only that, the interaction between these seven Chinese medicinal materials is also very close, forming an absolute core.

[0094] S4. Screen core Chinese medicinal materials, clarify the correlation between each Chinese medicinal material in the principle of monarch, minister, adjuvant and guiding, optimize the compatibility rules of Chinese medicine prescriptions, and obtain the Chinese medicine prescription with the most optimized compatibility relationship for stroke treatment; in this embodiment, the most optimized Chinese medicine prescription for stroke treatment is composed of seven Chinese medicinal materials: angelica, earthworm, chuanxiong, peach kernel, astragalus, safflower and red peony root.

[0095] The present application discloses a screening and identification method for effective Chinese medicine formulas based on data mining and AI, which uses a variety of statistical methods and several machine learning algorithms for overall analysis to discover the rules of Chinese medicine treatment of stroke, form a paradigm method, optimize the algorithm based on the theory of traditional Chinese medicine, and specifically use cluster analysis and association analysis to study the combination rules between medicinal materials, revealing the compatibility rules of Chinese medicine formulas, and screening core medicinal materials according to the principle of monarch, minister, adjuvant and envoy. A specific example of the optimal Chinese medicine formula for stroke treatment is obtained, which is composed of seven Chinese medicines: angelica, earthworm, chuanxiong, peach kernel, astragalus, safflower and red peony root.

[0096] The above implementation modes are not limitations of the present invention, and the present invention is not limited to the above examples. Any changes, modifications, additions or substitutions made by technicians in this technical field within the scope of the technical solution of the present invention also belong to the protection scope of the present invention.

Claims

1. A screening and identification method for effective Chinese medicine prescriptions based on data mining and AI, characterized in that: The following steps are involved: S1. Screen relevant literature on traditional Chinese medicine prescriptions for stroke treatment in academic journal databases; S2. Standardize the names of Chinese medicinal materials in Chinese herbal prescriptions for oral herbal use in the relevant literature according to the 2020 edition of the Pharmacopoeia of the People's Republic of China; S3. Conduct cluster analysis and association analysis on the selected Chinese medicinal materials in the Chinese medicine prescriptions according to the principle of monarch, minister, adjuvant and envoy in the prescription compatibility; S4. Screen core Chinese medicinal materials, clarify the correlation between each Chinese medicinal material in the principle of monarch, minister, adjuvant and guiding, optimize the compatibility rules of Chinese medicine prescriptions, and obtain the Chinese medicine prescription with the most optimized compatibility relationship for stroke treatment.

2. The method for screening and identifying effective Chinese medicine prescriptions based on data mining and AI according to claim 1, characterized in that: The academic journal database of S1 was searched separately with the keywords "stroke treatment", "TCM" and "Chinese herbal medicine", and the retrieved data were aggregated and added.

3. The method for screening and identifying effective Chinese medicine prescriptions based on data mining and AI according to claim 2, characterized in that: The following steps are performed on the relevant documents: A1. Screen relevant literature on clinical trials, clinical observations and human controlled experiments, and filter relevant literature on reviews, viewpoints, systematic reviews, animal experiments, cell experiments, basic research and data research; A2. Screen the literature on oral herbal medicine use, and filter the literature on diagnosis, acupuncture, fumigation, and massage; A3. Screen relevant literature with detailed descriptions of single prescriptions, and filter relevant literature without detailed ingredients, formulas or with variations; A4. Screen relevant literature with significant and / or positive results, and exclude relevant literature with insignificant effects and / or negative results.

4. The method for screening and identifying effective Chinese medicine prescriptions based on data mining and AI according to claim 1, characterized in that: The top 20 Chinese medicinal materials with the highest frequency of occurrence in the Chinese medicinal prescriptions of S3 were selected for cluster analysis, and an ML-based clustering algorithm was used, including a hierarchical clustering (HC) method and a K-modes clustering method.

5. The method for screening and identifying effective Chinese medicine prescriptions based on data mining and AI according to claim 4, characterized in that: The hierarchical clustering (HC) method converts the presence of each Chinese medicinal material in different Chinese medicine prescriptions into a sparse matrix, and uses the average linkage method to measure the differences between Chinese medicinal materials, wherein the average linkage distance measurement formula between two clusters is as follows: In the formula, A represents the Chinese medicinal materials in cluster A; B represents the Chinese medicinal materials in cluster B; ∑ i∈A ∑ j∈B d(i, j) represents the sum of all pairwise distances between points in cluster A and points in cluster B; |A| and |B| represent the number of data points in cluster A and B respectively; then, In the formula, x ik 、x jk They represent the coordinates of data points i and j in the kth dimension respectively; n is the dimension, representing the total number of Chinese medicinal materials; d(i, j) represents the distance between the i-th Chinese medicinal material and the j-th Chinese medicinal material.

6. The method for screening and identifying effective Chinese medicine prescriptions based on data mining and AI according to claim 4, characterized in that: The K-modes clustering method, i.e., a method of initially specifying the number of cluster centers and approximating the optimal clustering result through interactive calculation, identifies combinations of medicinal materials with similar characteristics by calculating the distance between Chinese medicinal materials and classifies them into the same category, wherein the distance is the Hamming distance.

7. The method for screening and identifying effective Chinese medicine prescriptions based on data mining and AI according to claim 6, characterized in that: The Hamming distance is defined as the number of mismatches in all feature items for two data points. The Hamming distance calculation formula is used to measure the difference between two discrete data, namely, the medicinal material attribute vectors. The calculation formula is as follows: This formula indicates whether the i-th feature item is matched, where the feature item refers to the presence of each Chinese medicinal material in different Chinese medicine prescriptions. Wherein, x represents the attribute vector of the first Chinese medicinal material; y represents the attribute vector of the second Chinese medicinal material; and n represents the total number of attributes.

8. The method for screening and identifying effective Chinese medicine prescriptions based on data mining and AI according to claim 1, characterized in that: The association analysis in S3 adopts the ML-based Apriori algorithm, in which three standard indicators, support, confidence and lift, are used to quantitatively evaluate the association relationship between Chinese medicinal materials. The support represents the frequency of simultaneous occurrence of certain combinations of Chinese medicinal materials, the confidence represents the probability that other medicinal materials will also appear when certain medicinal materials are known, and the lift represents the degree of influence of the appearance of certain medicinal materials on the appearance of other medicinal materials.

9. The method for screening and identifying effective Chinese medicine prescriptions based on data mining and AI according to claim 8, characterized in that: The calculation formula of the Apriori algorithm is as follows: In the formula, X represents the antecedent set; Y represents the consequent set; ALL represents all events in the data set; Support(X) represents the support of the medicinal material combination X; Confidence(X→Y) represents the confidence from the medicinal material combination X to Y; represents the improvement of the medicinal material combination X on Y.

10. The method for screening and identifying effective Chinese medicine prescriptions based on data mining and AI according to claim 1, characterized in that: The optimal Chinese medicine prescription for treating stroke is prepared by combining seven Chinese medicinal materials: angelica sinensis, earthworm, chuanxiong, peach kernel, astragalus, safflower and red peony root.

Images