CN106055631A - Acupuncture and moxibustion prescription main acupuncture point mining method based on fuzzy combined clustering method - Google Patents

Abstract

The invention discloses an acupuncture and moxibustion prescription main acupuncture point mining method based on a fuzzy combined clustering method, and the method comprises the steps: dividing an acupuncture point field for curing one disease into standard acupuncture points, and converting the standard acupuncture points into a co-occurrence relation of a prescription and the acupuncture points; carrying out the simultaneous clustering of the prescription and the acupuncture points through employing the fuzzy combined clustering method, and respectively finding out the prescriptions with the similar acupuncture points and the acupuncture points with the same prescriptions in a cluster; carrying out the simultaneous clustering of the prescription and the acupuncture points through employing an improved fuzzy combined clustering method when a prescription et in an acupuncture and moxibustion database is greater than a first threshold value and the number of classes is greater than a second threshold value, and respectively finding out the prescriptions with the similar acupuncture points and the acupuncture points with the same prescriptions in the cluster.

Description

Acupuncture and moxibustion prescription, acumoxibustion prescription master point method for digging based on Fuzzy Combined clustering method

Technical field

The present invention relates to technical field of data processing, be specifically related to a kind of acupuncture and moxibustion prescription, acumoxibustion prescription based on Fuzzy Combined clustering method Master point method for digging.

Background technology

Traditional Chinese medical science acupuncture points on the human body be acupuncture treatment of disease execute art position, during treatment disease, acupuncture is produced stimulation point and reaction, also It it is the traditional Chinese medical science foundation of opening acupuncture and moxibustion prescription, acumoxibustion prescription.In Chinese medicine acupuncture therapeutic process, rule of thumb with the feature of patient, different doctors Different acupuncture points on the human body treatment formula can be used for same class disease.Therefore, some support in the data base of system of acupuncture decision-making, The Different Acupoints combination of the employing of all medical treatment same diseases will form set.Research finds, can control from these Treat in the acupuncture points on the human body set of disease and find that certain general character, i.e. some acupuncture points on the human body can be commonly used by doctor when certain disease of medical needle. Thus consider to excavate relatively reliable main acupuncture points on the human body Set problem, disclose acupuncture points on the human body and use feature and Compatibility Law, how to obtain conjunction Manage effective applied research evidence to have very important significance to improving Acupuncture Clinical curative effect.

The retrieval of network multimedia and filter the most frequently used way and cluster exactly.The method is gathered according to Properties of Objects Integrated class, thus the similarity of similar is the least, between inhomogeneity, similarity is big as far as possible, and this is unsupervised mould in fact Formula recognition methods.Such as K-means clustering procedure, can be divided into several by certain object (typically with vector representation) set and gather Class or data subset, during in the most same cluster, similarity between each object clusters than other between each object Similarity is higher.

In order to promote the search efficiency of high dimensional data, data point high for similarity is gathered in one by logical General Clustering Algorithm Rise.Therefore clustering algorithm is directly or indirectly applied to a large amount of high dimensional data mechanism.But traditional clustering algorithm also has deficiency Place, because only clustering according to the similarity between object and object, does not but account for the similarity between object properties.

Summary of the invention

For the problems referred to above, the present invention provides a kind of and uses the card decision-making of Fuzzy Combined focusing solutions analysis Acupuncture Clinical to prop up Hold the prescription database in system, use associating clustering method that prescription and acupuncture points on the human body are clustered simultaneously, thus draw for certain The acupuncture and moxibustion prescription, acumoxibustion prescription master based on Fuzzy Combined clustering method of the several main acupuncture points on the human body set that the acupuncture and moxibustion prescription, acumoxibustion prescription of a kind of disease is corresponding Cave method for digging.

For reaching above-mentioned purpose, present invention acupuncture and moxibustion prescription, acumoxibustion prescription based on Fuzzy Combined clustering method master point method for digging, including: The acupuncture points on the human body field treating certain disease is resolved into standard acupuncture points on the human body and it is converted into the cooccurrence relation of prescription and acupuncture points on the human body；

Use Fuzzy Combined clustering method that prescription and acupuncture points on the human body are clustered simultaneously, find out respectively and cluster has similar acupoint The prescription in cave and the acupuncture points on the human body with similar prescription；

Further, the cluster standard of fuzzy cluster is as follows:

$\underset{i=1}{\overset{M}{\Σ}}\underset{j=1}{\overset{N}{\Σ}}{u}_{ci}{w}_{cj}{d}_{ij},c=1,\mathrm{...},C.---\left(1\right)$

In formula (1), if jth acupuncture points on the human body exists in i-th prescription, then d_ij=1；Otherwise d_ij=0；

The degree of polymerization of each bunch is by qualitative variable d_ij, prescription and the membership function u of acupuncture points on the human body_ciAnd w_cjProduct summation； By making concentration class maximize to the continuous assignment of the membership function of prescription and acupuncture points on the human body, thus form Fuzzy Cluster.Prescription and acupoint The membership function in cave is as follows:

$\underset{c=1}{\overset{C}{\Σ}}{u}_{ci}=1,u_{ci}\∈\[0,1\],i=1,\mathrm{...},M,---\left(2\right)$

$\underset{j=1}{\overset{N}{\Σ}}{w}_{cj}=1,w_{cj}\∈\[0,1\],c=1,\mathrm{...},C.---\left(3\right)$

Wherein, i and j refers to row i.e. prescription and the row i.e. acupuncture points on the human body of matrix respectively, and c indicates number of clusters, u_ciIt is that prescription is poly- Membership function in class c, w_cjIt it is acupuncture points on the human body membership function in cluster c.

When prescription set is more than Second Threshold more than first threshold and classification in acupuncture data base, uses and improve fuzzy connection Close clustering method prescription and acupuncture points on the human body are clustered simultaneously, find out the prescription in cluster with similar acupuncture points on the human body respectively with have similar The acupuncture points on the human body of prescription.

Further, in the described clustering method improving fuzzy cluster, prescription set { P regarded as by prescription bunch_ij| I=1 ... M, j=1 ... N}, the most each prescription weight sum in all of bunch is 1, and wherein M is prescription Number, N is cluster numbers, and defining the class belonging to the i-th row is C_(i)Thus propose improve model:

The most first obtain the ratio of each element in prescription set and every apoplexy due to endogenous wind all prescriptions element summation, if worked as When i-th row obtains the maximum of element, then return its affiliated class C_(i)=j, i.e. can get the concrete classification results of prescription；

Acupuncture points on the human body cluster can regard acupuncture points on the human body set { A as_ij| i=1 ... M, j=1 ... N}, the most all of acupuncture points on the human body exists The weight sum of each apoplexy due to endogenous wind is 1, and wherein M is acupuncture points on the human body number, and N is cluster numbers, every column element in definition acupuncture points on the human body bunch The minimum of value is made asIfThen take its value, thus draw and belong to the acupuncture points on the human body set that every class is new；I.e. can get the concrete classification results of acupuncture points on the human body.

Beneficial effect

Present invention acupuncture and moxibustion prescription, acumoxibustion prescription based on Fuzzy Combined clustering method master point method for digging possesses with prior art and has as follows Benefit effect:

Present invention acupuncture and moxibustion prescription, acumoxibustion prescription based on Fuzzy Combined clustering method master point method for digging, utilizes Fuzzy Combined Clustering And improve excavate the main acupuncture point of acupoint used by certain disease, test result indicate that, the present invention propose algorithm Can be good at excavating master point information used and being applied to DSS, through the mirror of Chengdu University of Traditional Chinese Medicine teacher Fixed, the method is very useful.It follows that for the user interface providing the user with close friend, the master point letter that can will excavate Breath shows in visual mode in decision system, thus improves the supporting capacity of DSS.

Detailed description of the invention

The present invention will be further described below.

Embodiment

The present embodiment acupuncture and moxibustion prescription, acumoxibustion prescription based on Fuzzy Combined clustering method master point method for digging, including: certain disease will be treated Sick acupuncture points on the human body field resolves into standard acupuncture points on the human body and it is converted into the cooccurrence relation of prescription and acupuncture points on the human body；

Use Fuzzy Combined clustering method that prescription and acupuncture points on the human body are clustered simultaneously, find out respectively and cluster has similar acupoint The prescription in cave and the acupuncture points on the human body with similar prescription；

When prescription set is more than Second Threshold more than first threshold and classification in acupuncture data base, uses and improve fuzzy connection Close clustering method prescription and acupuncture points on the human body are clustered simultaneously, find out the prescription in cluster with similar acupuncture points on the human body respectively with have similar The acupuncture points on the human body of prescription.

Fuzzy Combined cluster FCR (Fuzzy Co-clustering with Ruspini ' s condition) is the most right The row and column of matrix clusters, and advantage highlights, and is widely used in the field such as data analysis and collaborative filtering.Fuzzy in order to obtain Cluster, provides the definition of associating clustering algorithm clustering criteria, sees formula (1).Under this definition, those have phase the most each other Individuality and the attribute of closing property will gather together.To the extent that, we using the degree of polymerization as clustering criteria.Fuzzy cluster Cluster standard as follows:

$\underset{i=1}{\overset{M}{\Σ}}\underset{j=1}{\overset{N}{\Σ}}{u}_{ci}{w}_{cj}{d}_{ij},c=1,\mathrm{...},C.---\left(1\right)$

In formula (1), if jth acupuncture points on the human body exists in i-th prescription, then d_ij=1；Otherwise d_ij=0.The polymerization of each bunch Degree is by qualitative variable d_ij, prescription and the membership function u of acupuncture points on the human body_ciAnd w_cjProduct summation.By the person in servitude to prescription and acupuncture points on the human body The continuous assignment of genus degree function makes concentration class maximize, thus forms Fuzzy Cluster.The membership function of prescription and acupuncture points on the human body is as follows:

$\underset{c=1}{\overset{C}{\Σ}}{u}_{ci}=1,u_{ci}\∈\[0,1\],i=1,\mathrm{...},M,---\left(2\right)$

$\underset{j=1}{\overset{N}{\Σ}}{w}_{cj}=1,w_{cj}\∈\[0,1\],c=1,\mathrm{...},C.---\left(3\right)$

Wherein, i and j refers to row (prescription) and the row (acupuncture points on the human body) of matrix respectively, and c indicates number of clusters, u_ciIt is that prescription is poly- Membership function in class c, w_cjIt it is acupuncture points on the human body membership function in cluster c.

The associating clustering algorithm that this patent proposes is the maximization being realized the degree of polymerization by optimization object function, uses entropy Maximize and the bright sub-multiplication of glug constructs object function such as formula (4).The solution of object function is that the condition of known indispensability is led to Cross iterative processing, reach local minimum.By using the method, can once obtain the cluster result of whole set of data.Mould Stick with paste cluster after, can obtain belonging to whole prescription and acupuncture points on the human body bunch.

$\begin{array}{c}R=\underset{c=1}{\overset{C}{\Σ}}\underset{i=1}{\overset{M}{\Σ}}\underset{j=1}{\overset{N}{\Σ}}{u}_{ci}{w}_{cj}{d}_{ij}\\ -T_u\underset{c=1}{\overset{C}{\Σ}}\underset{i=1}{\overset{M}{\Σ}}{u}_{ci}{\mathrm{logu}}_{ci}-T_w\underset{c=1}{\overset{C}{\Σ}}\underset{j=1}{\overset{N}{\Σ}}{w}_{cj}{\mathrm{logw}}_{cj}\\ +\underset{i=1}{\overset{M}{\Σ}}{\mathrm{\λ}}_i(\underset{c=1}{\overset{C}{\Σ}}{u}_{ci}-1)\\ +\underset{c=1}{\overset{C}{\Σ}}{\mathrm{\γ}}_c(\underset{j=1}{\overset{N}{\Σ}}{w}_{cj}-1)\end{array}---\left(4\right)$

Wherein λ_i, γ_cIt it is different Lagrange's multipliers.(4) Section 2 in formula and Section 3 describe the maximization of entropy Becoming regular expression, this method is used in fuzzy c-means algorithm and is proposed [10] by Miyamoto et al. for the first time.Its energy Let us obtains fuzzy clustering.T_uAnd T_uBeing the weighting parameter indicating fuzziness, remaining description is degree of membership constraints, also It is exactly (1) formula and (2) formula..From the requirement of optimization object function R, namelyWithObtain Following equation:

$u_{ci}=\frac{\exp (\underset{j=1}{\overset{N}{\Σ}}{w}_{cj}{d}_{ij}/T_u)}{\underset{c=1}{\overset{C}{\Σ}}\exp (\underset{j=1}{\overset{N}{\Σ}}{w}_{cj}{d}_{ij}/T_u)},$

$w_{cj}=\frac{\exp (\underset{i=1}{\overset{M}{\Σ}}{u}_{ci}{d}_{ij}/T_w)}{\underset{j=1}{\overset{N}{\Σ}}\exp (\underset{i=1}{\overset{M}{\Σ}}{u}_{ci}{d}_{ij}/T_w)}$

According to discourse analysis above, main improve be on the basis of Fuzzy Combined clusters, prescription and acupuncture points on the human body are made into The fine grit classification of one step.We regard prescription bunch in table 1 as prescription set { P_ij| i=1 ... M, j=1 ... N}, the most often Individual prescription weight sum in all of bunch is 1.Wherein M is place's number formulary, and N is cluster numbers, defines the class belonging to the i-th row and is C_(i).Thus propose improve model:

The most first obtain the ratio of each element in prescription set and every apoplexy due to endogenous wind all prescriptions element summation, the most such as Fruit when the i-th row obtains the maximum of element, then returns its affiliated class C_(i)=j, i.e. can get the knot of specifically classifying of prescription Really.Shown below is a simple case that FCR-PA is explained.

Such as: the explanation to FRC-PA of the data for the treatment of apoplexy form 1 and form 2.

Table 1 apoplexy prescription bunch

Table 2 apoplexy acupuncture points on the human body bunch

{ i=1 ... M, j=1 ... N}, the most all of acupuncture points on the human body is 1 in the weight sum of each apoplexy due to endogenous wind: wherein M is acupoint Cave number, N is cluster numbers, and first in definition acupuncture points on the human body bunch, the minimum of each column element value is made asIfThen take its value, from And draw and belong to the acupuncture points on the human body set that every class is new:By further Specification, i.e. can get the concrete classification results of acupuncture points on the human body.Following example 1 is this model to be expanded on further, by the formulary of table 1 Conjunction can change into shown in matrix:

${\left[\begin{array}{cccc}\frac{p_{11}}{\underset{i=1}{\overset{105}{\mathrm{\Σ}}}{p}_{11}}& \frac{p_{12}}{\underset{i=1}{\overset{105}{\mathrm{\Σ}}}{p}_{12}}& \frac{p_{13}}{\underset{i=1}{\overset{105}{\mathrm{\Σ}}}{p}_{13}}& \frac{p_{14}}{\underset{i=1}{\overset{105}{\mathrm{\Σ}}}{p}_{14}}\\ \frac{p_{21}}{\underset{i=1}{\overset{105}{\mathrm{\Σ}}}{p}_{21}}& \frac{p_{22}}{\underset{i=1}{\overset{105}{\mathrm{\Σ}}}{p}_{22}}& \frac{p_{23}}{\underset{i=1}{\overset{105}{\mathrm{\Σ}}}{p}_{23}}& \frac{p_{24}}{\underset{i=1}{\overset{105}{\mathrm{\Σ}}}{p}_{24}}\\ .& .& .& .\\ .& .& .& .\\ .& .& .& .\\ \frac{p_{\mathrm{105,1}}}{\underset{i=1}{\overset{105}{\mathrm{\Σ}}}{p}_{\mathrm{105,1}}}& \frac{p_{\mathrm{105,2}}}{\underset{i=1}{\overset{105}{\mathrm{\Σ}}}{p}_{\mathrm{105,2}}}& \frac{p_{\mathrm{105,3}}}{\underset{i=1}{\overset{105}{\mathrm{\Σ}}}{p}_{\mathrm{105,3}}}& \frac{p_{\mathrm{105,4}}}{\underset{i=1}{\overset{105}{\mathrm{\Σ}}}{p}_{\mathrm{105,4}}}\end{array}\right]}_{105\×4}$

Then obtain the maximum often gone, model understand when the change corresponding to the maximum of row element every in acquirement matrix Amount j, is class C belonging to the i-th row_(i)。

The present invention uses Fuzzy Combined clustering method to cluster prescription and acupuncture points on the human body simultaneously, finds out in cluster respectively and has The prescription of similar acupuncture points on the human body and the acupuncture points on the human body with similar prescription.But when in acupuncture data base, formulary composition and division in a proportion is relatively big and classifies more Time, this algorithm is not enough to prescription and acupuncture points on the human body more preferably to cluster.The present invention proposes to use the FCR-PA algorithm improved further Fine grit classification, can clearly pick out the affiliated class of prescription and acupuncture points on the human body.It is a certain that the method can excavate treatment effectively Plant the main acupuncture points on the human body of disease, provide valuable decision support for different doctor's inquirings.

To the present invention it should be appreciated that embodiment described above, to the purpose of the present invention, technical scheme and useful effect Fruit carried out further details of explanation, these are only embodiments of the invention, be not intended to limit the present invention, every Within the spiritual principles of the present invention, done any modification, equivalent substitution and improvement etc., should be included in the protection of the present invention Within the scope of, protection scope of the present invention should be as the criterion with the protection domain that claim is defined.

Claims (3)

1. an acupuncture and moxibustion prescription, acumoxibustion prescription master point method for digging based on Fuzzy Combined clustering method, it is characterised in that including: will be treated certain The acupuncture points on the human body field planting disease resolves into standard acupuncture points on the human body and it is converted into the cooccurrence relation of prescription and acupuncture points on the human body；

Use Fuzzy Combined clustering method that prescription and acupuncture points on the human body are clustered simultaneously, find out respectively and cluster has similar acupuncture points on the human body Prescription and the acupuncture points on the human body with similar prescription；

When prescription set is more than Second Threshold more than first threshold and classification in acupuncture data base, use the Fuzzy Combined improved The further fine grit classification of clustering algorithm, the affiliated class picking out prescription and acupuncture points on the human body uses improvement Fuzzy Combined clustering algorithm pair Prescription and acupuncture points on the human body cluster simultaneously, find out the prescription in cluster with similar acupuncture points on the human body and the acupuncture points on the human body with similar prescription respectively.

Acupuncture and moxibustion prescription, acumoxibustion prescription master point method for digging based on Fuzzy Combined clustering method the most according to claim 1, its feature exists In, the cluster standard of fuzzy cluster is as follows:

$\underset{i=1}{\overset{M}{\Σ}}\underset{j=1}{\overset{N}{\Σ}}{u}_{ci}{w}_{cj}{d}_{ij},c=1,...,C.---\left(1\right)$

In formula (1), if jth acupuncture points on the human body exists in i-th prescription, then d_ij=1；Otherwise d_ij=0；

The degree of polymerization of each bunch is by qualitative variable d_ij, prescription and the membership function u of acupuncture points on the human body_ciAnd w_cjProduct summation；Pass through Make concentration class maximize to the continuous assignment of membership function of prescription and acupuncture points on the human body, thus form Fuzzy Cluster, prescription and acupuncture points on the human body Membership function is as follows:

$\underset{c=1}{\overset{C}{\Σ}}{u}_{ci}=1,u_{ci}\∈\[0,1\],i=1,...,M,---\left(2\right)$

$\underset{j=1}{\overset{N}{\Σ}}{w}_{cj}=1,w_{cj}\∈\[0,1\],c=1,...,C.---\left(3\right)$

Wherein, i and j refers to row i.e. prescription and the row i.e. acupuncture points on the human body of matrix respectively, and c indicates number of clusters, u_ciIt is that prescription is in cluster c Membership function, w_cjIt it is acupuncture points on the human body membership function in cluster c.

Acupuncture and moxibustion prescription, acumoxibustion prescription master point method for digging based on Fuzzy Combined clustering method the most according to claim 1, its feature exists In, in the described clustering algorithm improving fuzzy cluster, prescription set { P regarded as by prescription bunch_ij| i=1 ... M, j=1 ... N}, the most each prescription weight sum in all of bunch is 1, and wherein M is place's number formulary, and N is cluster numbers, defines belonging to the i-th row Class be C_(i)Thus propose improve model:

The most first obtain the ratio of each element in prescription set and every apoplexy due to endogenous wind all prescriptions element summation, if when i-th When row obtains the maximum of element, then return its affiliated class C_(i)=j, i.e. can get the concrete classification results of prescription；

Acupuncture points on the human body cluster can regard acupuncture points on the human body set { A as_ij| i=1 ... M, j=1 ... N}, the most all of acupuncture points on the human body is at each The weight sum of apoplexy due to endogenous wind is 1, and wherein M is acupuncture points on the human body number, and N is cluster numbers, and in definition acupuncture points on the human body bunch, each column element value is Lower bound is made asIfThen take its value, thus draw and belong to the acupuncture points on the human body set that every class is new；I.e. obtain the concrete classification results of acupuncture points on the human body.