CN114049930A - Traditional Chinese medicine prescription relocation method based on heterogeneous network representation learning - Google Patents
Traditional Chinese medicine prescription relocation method based on heterogeneous network representation learning Download PDFInfo
- Publication number
- CN114049930A CN114049930A CN202111342174.2A CN202111342174A CN114049930A CN 114049930 A CN114049930 A CN 114049930A CN 202111342174 A CN202111342174 A CN 202111342174A CN 114049930 A CN114049930 A CN 114049930A
- Authority
- CN
- China
- Prior art keywords
- chinese medicine
- prescription
- medicine prescription
- traditional chinese
- vector
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 239000003814 drug Substances 0.000 title claims abstract description 57
- 238000000034 method Methods 0.000 title claims abstract description 30
- 201000010099 disease Diseases 0.000 claims abstract description 34
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 claims abstract description 34
- 238000013135 deep learning Methods 0.000 claims abstract description 9
- 239000011159 matrix material Substances 0.000 claims abstract description 8
- 238000004364 calculation method Methods 0.000 claims abstract description 7
- 230000014509 gene expression Effects 0.000 claims abstract description 7
- 230000003993 interaction Effects 0.000 claims abstract description 6
- 238000007781 pre-processing Methods 0.000 claims abstract description 6
- 239000013598 vector Substances 0.000 claims description 39
- 238000009826 distribution Methods 0.000 claims description 8
- 238000013507 mapping Methods 0.000 claims description 6
- 230000000694 effects Effects 0.000 claims description 5
- 241000287196 Asthenes Species 0.000 claims description 3
- 230000002776 aggregation Effects 0.000 claims description 3
- 238000004220 aggregation Methods 0.000 claims description 3
- 230000006870 function Effects 0.000 claims description 3
- 230000008447 perception Effects 0.000 claims description 3
- 238000012546 transfer Methods 0.000 claims description 3
- 238000012935 Averaging Methods 0.000 claims description 2
- 230000005540 biological transmission Effects 0.000 claims description 2
- 239000013604 expression vector Substances 0.000 claims 1
- 239000000463 material Substances 0.000 claims 1
- 238000012163 sequencing technique Methods 0.000 abstract description 2
- 229940079593 drug Drugs 0.000 description 5
- 238000002474 experimental method Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 3
- 241001522129 Pinellia Species 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 235000008216 herbs Nutrition 0.000 description 2
- 238000010926 purge Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 239000013589 supplement Substances 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 229940126673 western medicines Drugs 0.000 description 2
- 241000202726 Bupleurum Species 0.000 description 1
- 239000009220 Liuwei Dihuang Decoction Substances 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000009472 formulation Methods 0.000 description 1
- 241000411851 herbal medicine Species 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 230000000144 pharmacologic effect Effects 0.000 description 1
- 239000006187 pill Substances 0.000 description 1
- 239000000843 powder Substances 0.000 description 1
- 239000000047 product Substances 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
- 239000008539 xiaoyao Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H20/00—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
- G16H20/10—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance relating to drugs or medications, e.g. for ensuring correct administration to patients
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Public Health (AREA)
- Biomedical Technology (AREA)
- Theoretical Computer Science (AREA)
- Epidemiology (AREA)
- Primary Health Care (AREA)
- Physics & Mathematics (AREA)
- Biophysics (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Medicinal Chemistry (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Chemical & Material Sciences (AREA)
- Databases & Information Systems (AREA)
- Pathology (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The invention discloses a traditional Chinese medicine prescription relocation method based on heterogeneous network representation learning, which sequentially comprises the following steps: step 1, preprocessing the Chinese medicine according to input data, and constructing a Chinese medicine prescription heterogeneous network; step 2, according to the heterogeneous graph embedding method, a network pharmacology learning module is executed to learn semantic associated information in a heterogeneous network of the traditional Chinese medicine prescription and obtain embedded representation of the traditional Chinese medicine prescription and diseases; and 3, obtaining an interaction influence scoring matrix of the traditional Chinese medicine prescription and the disease by using a deep learning recommendation algorithm according to the embedded expression of the traditional Chinese medicine prescription and the disease obtained in the step 2. And finally, sequencing the scoring matrix results to obtain a disease recommendation result of Top @ K of the traditional Chinese medicine prescription. According to the invention, the disease recommendation of the traditional Chinese medicine prescription can be obtained through intelligent calculation, so that manual operation is omitted, and the efficiency of repositioning the traditional Chinese medicine prescription is effectively improved.
Description
Technical Field
The invention relates to a Chinese medicine prescription relocation method, in particular to a Chinese medicine prescription relocation method based on heterogeneous network representation learning.
Background
Drug relocation aims at finding new applications for existing drugs, has the advantages of low waste rate, reduced cost and short time, and draws attention of many researchers. Although research work on relocation of western medicines is many, the relocation method of western medicines cannot be directly applied to relocation of traditional Chinese medicine prescriptions aiming at the characteristics of multiple components, multiple target points and multipath regulation of traditional Chinese medicine prescriptions. In addition, although some researchers have studied the relocation of traditional Chinese medicine formulations using cyber pharmacology, they still use a small number of tools for manual exploration. Currently, there is no universal prescription relocation prediction method.
Disclosure of Invention
The purpose of the invention is as follows: the invention aims to provide a Chinese Medicine prescription relocation method (HNPL) based on Heterogeneous Network representation Learning and aims to solve the problem that the existing feature generation method cannot process irrelevant or redundant feature generation, and the method for representing and Learning by using the Heterogeneous Network
The technical scheme is as follows: in order to achieve the above purpose, the method for relocating the traditional Chinese medicine prescription based on heterogeneous network representation learning sequentially comprises the following steps of:
step 3, repositioning the prescription: according to the characteristic embedding representation obtained in the step 2, calculating by using a deep learning recommendation method to obtain an interaction scoring matrix Y between the traditional Chinese medicine prescription and the diseasepd(ii) a And then ranking the influence scores of the related diseases of each prescription to obtain the first K recommended results.
Further, step 1 specifically comprises: and establishing the association of corresponding nodes in the connection network to form the Chinese medicine prescription heterogeneous network. For example, if there is an association between the prescription p and the drug h, there is an edge between two nodes in the corresponding network G, otherwise there is no.
Further, step 2 specifically comprises:
step 21, defining meta relation: for an edge linked from a source node s to a target node t, its meta-relationship is represented asWherein, τ (. sup.) andrepresenting the type of node and edge, respectively. In TCMHIN, it is assumed that there may be multiple types of relationships between different types of nodes. For example, there is an inclusion relationship between prescriptions and nodes of herbs, etc.
And step 22, heterogeneous mutual attention calculation based on meta-relation perception:
given a target node prescription p (or disease d), all its neighbors h e N (p) may belong to different distributions, their mutual Attention Attention () must be computed from their meta-relationships. The attention weight e ═ h, p for k-head for each edge is calculated as follows:
in these formulas, ATT-head for the ith attention headi(h, e, p) projecting a τ (h) type source node s to the ith Key vector Ki(h) Middle, linear projectionWhere n is the number of attention heads and m is the vector dimension of each head, then linear projection is usedAnd projecting the target node p into the ith Query vector. (2) U in the (1) is a prior tensor, represents each element relation triple, and performs attention adaptive scaling operation based on the element relation.
And step 23, heterogeneous information transmission:
for edge e ═ h, p, the multi-headed Message (h, e, p) can be calculated as equation (5), (6):
to obtain the MSG-headi(h,e,p)Using linear mappingTo map a source node h of type τ (h) as the ith message vector. Matrix arrayFor integrating edge dependencies.
Step 24, heterogeneous message aggregation:
averaging corresponding information from the source node h by using the attention vector as a weight, and adopting outer product calculationObtaining an updated vector:
updating vectorsThen, to obtain the first layer E [ p ]]The final output of (2) is to map the target node p and return the distribution of its corresponding class, with τ (p) of the target node as an index. Then linearly mapping A-Linearτ(p)Applied to the update vector, and then concatenates the original vectors of the previous layer p as residuals:
to transfer information from the source node h to the destination node p, the computation of mutual attention is performed in parallel. The purpose is to combine the meta-relations of different edges into the message passing process to ease the difference of different types of nodes and edge distributions.
By this way of network learning representation, E can be obtained(l)[p]And E(l)[d]Can be used as input to the downstream prescription relocation module task.
Further, step 3 specifically comprises:
step 31, calculating the interaction influence score between the prescription and the disease:
obtaining an embedded vector of a prescription node p, namely e, through a network representation learning module*p; similarly, for disease node d, the vector is represented asThen, the representation vector is used as an input vector of a deep learning recommendation algorithm Recommend (). The calculation method of the prediction score is shown as the formula (9) according to the relationship between the traditional Chinese medicine prescription p and the disease d.
To learn the optimal parameters, the objective function is set to:
in this formula, λ is the regularization parameter, | | is the corresponding regularization term, random gradient descent is used to optimize the parameter, the model is iterated continuously until the effect is best, and finally the prediction result is evaluated.
Has the advantages that:
the invention provides a traditional Chinese medicine prescription relocation method based on heterogeneous network representation learning, which converts the traditional Chinese medicine prescription relocation problem into a recommendation problem by means of the thought of combining heterogeneous network representation learning and a deep learning recommendation algorithm. The model firstly integrates the existing Chinese medicine prescription data to construct a related Chinese medicine prescription heterogeneous network, and then learns prescriptions, diseases and topological neighborhood representation thereof from the network by using the graph neural network technology to extract network structure information and semantic relations. Finally, the model finds potential application of the traditional Chinese medicine prescription through a deep learning recommendation method on the basis of fully learning network pharmacology.
It includes the following advantages:
(1) the semantic relation is extracted from the Chinese medicine prescription heterogeneous network by using a transform-based heterogeneous network representation learning technology, so that the characteristics and the association of the prescription and the diseases can be better expressed.
(2) The model applies meta-relationships and attention mechanisms to automatically explore pharmacological information between prescriptions and diseases, thereby avoiding the required domain knowledge.
Drawings
FIG. 1 is an overall framework diagram of the HNPL model of the present invention;
FIG. 2 is a schematic diagram of a network pharmacology learning module;
FIG. 3 is a graph of TCMHIN-ETCM embedded dimension experimental results;
FIG. 4 is a graph of the experimental results of TCMHIN-TCMID embedding dimension.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described below with reference to specific embodiments and illustrative drawings, it being understood that the preferred embodiments described herein are for the purpose of illustration and explanation only and are not intended to limit the present invention.
The invention relates to a traditional Chinese medicine prescription relocation method based on heterogeneous network representation learning, which sequentially comprises the following steps of:
1. constructing a heterogeneous network of the traditional Chinese medicine prescription:
preprocessing input data, including representing various entities and relations, and obtaining a Chinese medicine prescription heterogeneous network G after representation; and establishing the association of corresponding nodes in the connection network to form the Chinese medicine prescription heterogeneous network. For example, if there is an association between the prescription p and the drug h, there is an edge between two nodes in the corresponding network G, otherwise there is no.
2. Network pharmacology learning:
obtaining Chinese medicine prescription and embedded expression of disease E [ p ] by heterogeneous network expression learning method]And E [ d ]](ii) a Step 21, first defining meta-relations. For an edge linked from a source node s to a target node t, its meta-relationship is represented asIn TCMHIN, it is assumed that there may be multiple types of relationships between different types of nodes. For example, there is an inclusion relationship between prescriptions and nodes of herbs, etc.
To compute the heterogeneous mutual attention based on meta-relationship perception first, given a target node prescription p (or disease d), all its neighbors h e N (p) may belong to different distributions, and their mutual attention must be computed according to their meta-relationship. The attention weight e ═ h, p for k-head for each edge is calculated as follows:
in these formulas, ATT-head for the ith attention headi(h, e, p) projecting a τ (h) type source node s to the ith Key vector Ki(h) Middle, linear projectionWhere n is the number of attention heads and m is the vector dimension of each head, then linear projection is usedAnd projecting the target node p into the ith Query vector.
At the heterogeneous information dissemination layer for edge e ═ h, p, a multi-headed message can be calculated as equation (5), (6):
to obtain the MSG-headi(h,e,p)Using linear mappingTo map a source node h of type τ (h) as the ith message vector. Matrix arrayFor integrating edge dependencies.
The heterogeneous message aggregation layer uses the attention vector as a weight to average corresponding information from the source node h to obtain an updated vector:
updating vectorsThen, to obtain the first layer E [ p ]]The final output of (2) is to map the target node p and return the distribution of its corresponding class, with τ (p) of the target node as an index. Then linearly mapping A-Linearτ(p)Applied to the update vector, and then concatenates the original vectors of the previous layer p as residuals:
to transfer information from the source node h to the destination node p, the computation of mutual attention is performed in parallel. The purpose is to combine the meta-relations of different edges into the message passing process to ease the difference of different types of nodes and edge distributions.
By this way of network learning representation, E can be obtained(l)[p]And E(l)[d]Can be used as input to the downstream prescription relocation module task.
3. Prescription relocation module:
calculating to obtain an interaction scoring matrix Y between the Chinese medicinal prescription and the disease by using a deep learning recommendation methodpd. Obtaining an embedded vector of a prescription node p, namely e, through a network representation learning module*p; similarly, for disease node d, the vector is represented asThen, the representation vector thereof is taken as an input vector of the deep learning recommendation algorithm recammend (). The calculation method of the prediction score is shown as the formula (9) according to the relationship between the traditional Chinese medicine prescription p and the disease d.
To learn the optimal parameters, the objective function is set to:
in this formula, λ is a regularization parameter. A random gradient descent is used to optimize the parameters. And continuously iterating the model until the effect is the best, finally evaluating the prediction result and sequencing the influence scores of the related diseases of each prescription to obtain the first K recommended results.
Experiment:
to verify the effectiveness of the model for the relocation of the traditional Chinese medicine prescription, the patent performed experiments on two data sets, the data set used being shown in table 1.
TABLE 1 Experimental data set
The TCMHIN is based on ETCM and TCMID databases and is supplemented with new supplement of YaTCM, TCMSP and TCM-ID to form two Chinese medicine prescription heterogeneous information network data sets of TCMHIN-TCMID and TCMHIN-ETCM. For example, in the ETCM database, there are cases where the herbal medicines contained in the prescription are incomplete or do not correspond to other documents or databases, and the corresponding prescription is searched in YaTCM and TCMSP to supplement and update TCMHIN-ETCM. The data sources are shown in Table 2.
TABLE 2 data sources
Tables 3 and 4 show a comparison of the performance of the different models on the two data sets. As can be observed from the table, the results of the HNPL model in the evaluation indexes of Precision @10, Recall @10 and NDCG @10 are the highest, which shows that compared with other models, the HNPL model training effect is better.
TABLE 3 Experimental results on different methods TCMHIN-ETCM data set
Model (model) | P@10 | R@10 | NDCG@10 |
FM | 0.9602 | 0.0187 | 0.9609 |
NFM | 0.9622 | 0.0188 | 0.9626 |
CKE | 0.9297 | 0.0175 | 0.9304 |
HERec | 0.9354 | 0.0177 | 0.9315 |
HAN-NFM | 0.9626 | 0.0284 | 0.9714 |
KGAT | 0.9735 | 0.0356 | 0.9794 |
HNPL | 0.9824 | 0.0415 | 0.9876 |
TABLE 4 results of experiments on TCMHIN-TCMID data sets by different methods
Model (model) | P@10 | R@10 | NDCG@10 |
FM | 0.9423 | 0.2394 | 0.9498 |
NFM | 0.9465 | 0.2456 | 0.9495 |
CKE | 0.9243 | 0.2345 | 0.9386 |
HERec | 0.9365 | 0.2473 | 0.9397 |
HAN-NFM | 0.9517 | 0.2604 | 0.9524 |
KGAT | 0.9675 | 0.2688 | 0.9682 |
HNPL | 0.9848 | 0.2746 | 0.9852 |
Fig. 3 and 4 illustrate the effect of the dimensions of the embedded vector on the results. In the experiments, the best results occur when d equals 64, indicating that increasing the dimension d appropriately can improve the performance of the model, but too large a dimension value may risk overfitting.
Table 5 shows the coverage of case document study relocation results in case validation. The disease predicted by the model is marked in italics. Therefore, the potential diseases recommended by the pinellia ternate heart-fire-purging decoction and the bupleurum-cassia twig decoction cover all the disease results in the research. Similarly, the recommendations for LIUWEIDIHUANG pill and XIAOYAO powder cover 2 of the 3 diseases in the study. The coverage rate of the big bupleurum decoction reaches 60 percent.
TABLE 5 recall results for case verification
Table 6 we used bolded font marks for predicted diseases, verified by searching for relevant literature. For the pinellia ternate decoction for purging the heart, the HNPL predicted residual disease rate can reach 75% through literature verification. As for the Liuwei Dihuang Wan, 5 related documents were also found in the remaining 8 diseases. This result confirms to some extent the effectiveness of the HNPL model.
TABLE 6 predictive disease of top 10
Claims (4)
1. A traditional Chinese medicine prescription relocation method based on heterogeneous network representation learning is characterized by sequentially comprising the following steps of:
step 1, constructing a heterogeneous network of a traditional Chinese medicine prescription: preprocessing the Chinese medicine prescription according to input data, wherein the preprocessing comprises representing various entities and relations, and a Chinese medicine prescription heterogeneous network G is obtained after representation;
step 2, network pharmacology learning: according to the Chinese medicine prescription heterogeneous network data obtained in step 1, the Chinese medicine prescription and embedded expressions of diseases E p and E d are obtained by a heterogeneous network expression learning method;
step 3, repositioning the prescription: according to the characteristic embedding representation obtained in the step 2, calculating by using a deep learning recommendation method to obtain an interaction scoring matrix Y between the traditional Chinese medicine prescription and the diseasepd(ii) a And then ranking the influence scores of the related diseases of each prescription to obtain the first K recommended results.
2. The method for relocating traditional Chinese medicine prescription based on heterogeneous network representation learning according to claim 1, wherein the step 1 specifically comprises: establishing association for connecting corresponding nodes in the network to form a Chinese medicine prescription heterogeneous network; if the prescription p and the medicinal material h are associated, an edge exists between two nodes in the corresponding network G, otherwise, no edge exists.
3. The method for relocating traditional Chinese medicine prescription based on heterogeneous network representation learning according to claim 2, wherein the step 2 specifically comprises:
step 21, defining meta relation: for an edge linked from a source node s to a target node t, its meta-relationship is represented asWherein, τ (. sup.) andrespectively representing the types of nodes and edges; in TCMHIN, it is assumed that there may be multiple types of relationships between different types of nodes; if the inclusion relationship exists between the prescription and the herbal nodes;
and step 22, heterogeneous mutual attention calculation based on meta-relation perception:
given a target node prescription p or disease d, all its neighbors h e N (p) may belong to different distributions, their mutual Attention attentions () must be computed from their meta-relations; the attention weight e ═ h, p for k-head for each edge is calculated as follows:
in these formulas, ATT-head for the ith attention headi(h, e, p) projecting a τ (h) type source node s to the ith Key vector Ki(h) Middle, linear projection Where n is the number of attention heads and m is the vector dimension of each head, then linear projection is usedProjecting a target node p into an ith Query vector; (2) u in the three-dimensional relationship is a prior tensor, represents each element relationship triple, and performs attention self-adaptive scaling operation based on the element relationship;
and step 23, heterogeneous information transmission:
for edge e ═ h, p, the multi-headed Message (h, e, p) representation can be calculated as equation (5), (6):
to obtain the MSG-headi(h,e,p)Using linear mapping Mapping a source node h with the type of tau (h) to obtain an ith message vector; matrix arrayDependencies for edges to integrate;
step 24, heterogeneous message aggregation:
averaging the corresponding information from the source node h using the attention vector as a weight, taking the outer productCalculating to obtain an updated vector:
updating vectorsThen, to obtain the first layer E [ p ]]The final output of (2) needs to map the target node p and return the distribution of its corresponding class with τ (p) of the target node as an index, and then linearly map A-Linearτ(p)Applied to the update vector, and then concatenates the original vectors of the previous layer p as residuals:
to transfer information from the source node h to the target node p, the computation of mutual attention is performed in parallel; by this way of network learning representation, E can be obtained(l)[p]And E(l)[d]Can be used as input to the downstream prescription relocation module task.
4. The method for relocating traditional Chinese medicine prescription based on heterogeneous network representation learning according to claim 3, wherein the step 3 is specifically as follows:
step 31, calculating the interaction influence score between the prescription and the disease:
obtaining an embedded vector of a prescription node p, namely e, through a network representation learning module* p(ii) a Similarly, for disease node d, the vector is represented asThen, the expression vector is used as an input vector of a deep learning recommendation algorithm Recommend (x), and a prediction score calculation method is shown as a formula (9) according to the relation between a traditional Chinese medicine prescription p and a disease d:
to learn the optimal parameters, the objective function is set to:
in this formula, λ is the regularization parameter, | × | is the corresponding regularization term, random gradient descent is used to optimize the parameter, the model is iterated continuously until the effect is best, and finally the prediction result is evaluated.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111342174.2A CN114049930A (en) | 2021-11-12 | 2021-11-12 | Traditional Chinese medicine prescription relocation method based on heterogeneous network representation learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111342174.2A CN114049930A (en) | 2021-11-12 | 2021-11-12 | Traditional Chinese medicine prescription relocation method based on heterogeneous network representation learning |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114049930A true CN114049930A (en) | 2022-02-15 |
Family
ID=80208864
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111342174.2A Pending CN114049930A (en) | 2021-11-12 | 2021-11-12 | Traditional Chinese medicine prescription relocation method based on heterogeneous network representation learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114049930A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114613452A (en) * | 2022-03-08 | 2022-06-10 | 电子科技大学 | Drug relocation method and system based on drug classification map neural network |
CN114613437A (en) * | 2022-03-08 | 2022-06-10 | 电子科技大学 | miRNA and disease associated prediction method and system based on heteromorphic image |
CN114912033A (en) * | 2022-05-16 | 2022-08-16 | 重庆大学 | Knowledge graph-based recommendation popularity deviation adaptive buffering method |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20180247224A1 (en) * | 2017-02-28 | 2018-08-30 | Nec Europe Ltd. | System and method for multi-modal graph-based personalization |
WO2019231624A2 (en) * | 2018-05-30 | 2019-12-05 | Quantum-Si Incorporated | Methods and apparatus for multi-modal prediction using a trained statistical model |
CN111916145A (en) * | 2020-07-24 | 2020-11-10 | 湖南大学 | Novel coronavirus target prediction and drug discovery method based on graph representation learning |
CN113505294A (en) * | 2021-06-15 | 2021-10-15 | 黄萌 | Heterogeneous network representation recommendation algorithm fusing meta-paths |
-
2021
- 2021-11-12 CN CN202111342174.2A patent/CN114049930A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20180247224A1 (en) * | 2017-02-28 | 2018-08-30 | Nec Europe Ltd. | System and method for multi-modal graph-based personalization |
WO2019231624A2 (en) * | 2018-05-30 | 2019-12-05 | Quantum-Si Incorporated | Methods and apparatus for multi-modal prediction using a trained statistical model |
CN111916145A (en) * | 2020-07-24 | 2020-11-10 | 湖南大学 | Novel coronavirus target prediction and drug discovery method based on graph representation learning |
CN113505294A (en) * | 2021-06-15 | 2021-10-15 | 黄萌 | Heterogeneous network representation recommendation algorithm fusing meta-paths |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114613452A (en) * | 2022-03-08 | 2022-06-10 | 电子科技大学 | Drug relocation method and system based on drug classification map neural network |
CN114613437A (en) * | 2022-03-08 | 2022-06-10 | 电子科技大学 | miRNA and disease associated prediction method and system based on heteromorphic image |
CN114613452B (en) * | 2022-03-08 | 2023-04-28 | 电子科技大学 | Drug repositioning method and system based on drug classification graph neural network |
CN114613437B (en) * | 2022-03-08 | 2023-05-26 | 电子科技大学 | Method and system for predicting association of miRNA and diseases based on different patterns |
CN114912033A (en) * | 2022-05-16 | 2022-08-16 | 重庆大学 | Knowledge graph-based recommendation popularity deviation adaptive buffering method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN114049930A (en) | Traditional Chinese medicine prescription relocation method based on heterogeneous network representation learning | |
Geng | Label distribution learning | |
WO2023000574A1 (en) | Model training method, apparatus and device, and readable storage medium | |
CN105468605B (en) | Entity information map generation method and device | |
Mansingh et al. | Using ontologies to facilitate post-processing of association rules by domain experts | |
CN105117428A (en) | Web comment sentiment analysis method based on word alignment model | |
CN107122809A (en) | Neural network characteristics learning method based on image own coding | |
CN113066526B (en) | Hypergraph-based drug-target-disease interaction prediction method | |
CN108804419A (en) | Medicine is sold accurate recommended technology under a kind of line of knowledge based collection of illustrative plates | |
Kunjir et al. | Data mining and visualization for prediction of multiple diseases in healthcare | |
CN111477344B (en) | Drug side effect identification method based on self-weighted multi-core learning | |
Hernández-Julio et al. | Framework for the development of data-driven Mamdani-type fuzzy clinical decision support systems | |
CN111191048A (en) | Emergency call question-answering system construction method based on knowledge graph | |
CN109903854A (en) | A kind of core drug recognition methods based on TCM Literature | |
CN102799671A (en) | Network individual recommendation method based on PageRank algorithm | |
CN109271486A (en) | A kind of similitude reservation cross-module state Hash search method | |
CN116364299B (en) | Disease diagnosis and treatment path clustering method and system based on heterogeneous information network | |
Skupin et al. | Spatialization | |
Pan et al. | Low-rank tensor regularized graph fuzzy learning for multi-view data processing | |
CN113764112A (en) | Online medical question and answer method | |
Sumathi et al. | Improved fuzzy weighted‐iterative association rule based ontology postprocessing in data mining for query recommendation applications | |
CN106529138A (en) | Method and device for information pushing | |
CN114969369A (en) | Knowledge graph human cancer lethal prediction method based on mixed network and knowledge graph construction method | |
CN113706251A (en) | Commodity recommendation method and device based on model, computer equipment and storage medium | |
CN116386899A (en) | Graph learning-based medicine disease association relation prediction method and related equipment |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |