CN112652373B - Prescription recommendation method based on tongue image retrieval - Google Patents
Prescription recommendation method based on tongue image retrieval Download PDFInfo
- Publication number
- CN112652373B CN112652373B CN202011433472.8A CN202011433472A CN112652373B CN 112652373 B CN112652373 B CN 112652373B CN 202011433472 A CN202011433472 A CN 202011433472A CN 112652373 B CN112652373 B CN 112652373B
- Authority
- CN
- China
- Prior art keywords
- prescription
- tongue
- feature
- similarity
- 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.)
- Active
Links
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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Library & Information Science (AREA)
- Databases & Information Systems (AREA)
- Chemical & Material Sciences (AREA)
- Medicinal Chemistry (AREA)
- Epidemiology (AREA)
- Medical Informatics (AREA)
- Primary Health Care (AREA)
- Public Health (AREA)
- Image Analysis (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The invention discloses a prescription recommendation method based on tongue image retrieval, which comprises the following steps: s1, a server side analyzes prescription information of a database and vectorizes and represents a prescription; s2, training a feature extractor for obtaining a tongue picture according to vector representation of the prescription; s3, pre-extracting a feature vector of each tongue picture in the database by using a feature extractor; s4, submitting tongue pictures to be inquired by the client; s5, the server side extracts the feature vector of the tongue picture by using a feature extractor; s6, the server side calculates the feature similarity between the tongue picture and each tongue picture in the database by using a similarity calculation module; and S7, the server uses a prescription recommendation module to sort according to the feature similarity obtained by calculation in the S6, and returns the prescription corresponding to the K tongue pictures with higher similarity to the client. The invention combines deep learning and prescription recommendation, and realizes the prescription recommendation method based on tongue pictures for the first time.
Description
Technical Field
The invention relates to image retrieval, target detection and prescription recommendation in the traditional Chinese medicine field in the computer vision field, in particular to a prescription recommendation method based on tongue image retrieval.
Background
In the age of rapid development of artificial intelligence technology, diagnosis of traditional Chinese medicine diseases mainly depends on abundant knowledge and experience of experts. As the case database grows day by day, intelligent adjuvant therapy is a necessary trend based on historical case information. However, manually searching for a referenceable case becomes time-consuming and labor-consuming due to the large amount of case data. Therefore, how to apply the artificial intelligence technology to effectively retrieve the case similar to the current patient from the database, and assist the doctor in assisting the diagnosis of the patient's condition, especially the difficult and complicated conditions, becomes a core problem, and further, directly provides the final decision reference. The artificial intelligence technology is applied to the diagnosis and treatment process of the traditional Chinese medicine, and the diagnosis efficiency can be greatly improved no matter the control of the state of an illness is carried out. It is also an important component in the modernization process of traditional Chinese medicine.
The traditional Chinese medical theory of visceral manifestation and holistic concept holds that the pathological changes of the zang-fu organs will be manifested through the external expression, and the tongue manifestation is one of the windows that can reflect the pathological changes of the zang-fu organs. The clinical tongue examination method mentioned that neither of the two types of syndromes will show their shapes, and the tongue will be stained with the color of the internal and external syndromes. In the book "discussing insomnia differentiation and treatment based on tongue diagnosis theory in traditional Chinese medicine", it is mentioned that the tongue is dark purple, or there is ecchymosis on the tongue edge, or the tongue has blood stasis in the sublingual collaterals and purple, and the tongue coating is thin and white, indicating blood stasis blockage or qi stagnation and blood stasis, and it is necessary to combine with blood-activating herbs such as Chuan Xiong, red peony root, corydalis tuber, curcuma rhizome, salvia miltiorrhiza, etc. to promote qi circulation, activate blood circulation and remove blood stasis. Therefore, the internal state of the patient can be reflected by the tongue picture, and the tongue picture has an indispensable effect on the diagnosis and treatment of the traditional Chinese medicine.
TCM synthesizes various information and combines the experience of doctors to make diagnosis and treatment. However, for inexperienced doctors, it is often necessary to use a comparative analysis to diagnose the doctor with reference to the cases of other doctors. Meanwhile, a more accurate judgment can be made on the current illness state by referring to similar historical cases.
However, no prescription recommendation method takes into account the dosage information in the prescription, and the current prescription recommendation method has the following disadvantages. A deep neural network-based traditional Chinese medicine recommendation method (publication No. 108182967 a) recommends medicinal materials using tongue pictures, but cannot guarantee that the recommended medicinal materials can form an effective prescription. A hidden semantic model-based traditional Chinese medicine formula recommendation method and system (publication number 11477295A) recommend a prescription according to symptoms, and compared with tongue images, the method and system have higher subjectivity.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention provides a prescription recommendation method for tongue image retrieval.
The invention is realized by at least one of the following technical schemes.
A prescription recommendation method based on tongue image retrieval comprises the following steps:
s1, a server side analyzes prescription information of a database and vectorizes and represents a prescription;
s2, training a feature extractor for obtaining a tongue picture according to vector representation of the prescription;
s3, pre-extracting a feature vector of each tongue picture in the database by using a feature extractor;
s4, submitting tongue pictures to be inquired by the client;
s5, the server side extracts the feature vector of the tongue picture by using a feature extractor;
s6, the server side calculates the feature similarity between the tongue picture and each tongue picture in the database by using a similarity calculation module;
and S7, the server uses a prescription recommendation module to sort according to the feature similarity obtained by calculation in the S6, and returns the prescription corresponding to the K tongue pictures with higher similarity to the client.
Preferably, the vectorization representation of the prescription in step S1 includes the medicinal materials used by the prescription and the dosage information thereof.
Preferably, the step S1 specifically includes the steps of:
s11, sorting database records, and eliminating prescriptions which do not accord with traditional Chinese medicine standards and are wrongly input;
s12, respectively standardizing the dose of each medicinal material, compressing the dose of each medicinal material to the same range, and eliminating dimensional interference;
s13, representing each prescription in the database into a prescription vector T with dimension H, wherein the ith element T of the prescription vector T i ＝M+C i ，C i The standardized dosage used for the ith medicine in the prescription is shown, M is a hyperparameter, and H is the total medicine number in the database.
Preferably, the feature extractor takes a tongue picture as input, and extracts a D-dimensional feature vector through the feature extractor.
Preferably, the feature extractor includes an image classification network BONE and a fully-connected neural network MLP, which replace the classifier with a fully-connected network with an output dimension D.
Preferably, the fully-connected neural network MLP is composed of a plurality of fully-connected layers, and the output dimension of the last fully-connected layer FCn is H, which is the same as the dimension H of the prescription vector.
Preferably, the training of step S2 is a process of repeated iteration, and each iteration specifically includes the following steps:
s21, randomly extracting two tongue pictures as a pair, and standardizing the two tongue pictures according to the standardized mode of an image classification network BONE; respectively extracting the two tongue picture feature vectors by using an image classification network BONE to obtain two feature vectors V 1 ,V 2 (ii) a Will V 1 ,V 2 Inputting into a fully-connected neural network MLP to obtain a predicted prescription vector O 1 ,O 2 ；
S22, calculating the feature similarity S between the two tongue pictures and the prescription similarity L corresponding to the two tongue pictures, calculating the similarity S between feature vectors through a similarity calculation module, calculating by using the inner product of the vectors, calculating the similarity L between the real prescription T by using the cosine similarity of the prescription vectors, and adjusting the parameters of BONE and MLP by using an error back propagation algorithm through two optimization losses.
Preferably, in step S2, when training the feature extractor, iteratively optimizing two optimization losses repeatedly until convergence to adjust parameters of the deep convolutional neural network, and performing target optimization according to the predicted D-dimensional feature vector.
Preferably, the specific optimization comprises the following steps:
1) For each tongue picture which is randomly extracted, minimizing the difference between the predicted prescription vector O of the tongue picture and the corresponding real prescription vector T;
2) For a pair of tongue pictures which are randomly extracted, the difference between the predicted feature similarity S and the true prescription similarity L of the two is minimized.
Preferably, in step S7, the prescription recommendation module sorts the tongue pictures according to feature similarity between feature vectors of the tongue pictures provided by the client and feature similarities of all tongue pictures in the database, recommends the prescription corresponding to the K tongue pictures with the highest similarity, and returns the prescription to the client.
Compared with the prior art, the invention has the beneficial effects that:
1. the prescription is recommended directly according to the tongue picture, so that the method is more objective.
2. The tongue picture information can be directly obtained without the judgment of a doctor, and is convenient and quick.
3. The recommended prescription is more practical in consideration of dosage information.
Drawings
FIG. 1 is a flow chart of a prescription recommendation method based on tongue image retrieval;
fig. 2 is a block diagram of a tongue picture feature extraction network.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiment is one embodiment of the present invention, and not all embodiments.
As shown in fig. 1 and fig. 2, a prescription retrieval method based on tongue image of the present embodiment includes the following steps:
in the preparation phase of the system, namely the offline training phase, the method comprises the following steps:
s1, a server side analyzes prescription information of a database and vectorizes and represents a prescription, and the method specifically comprises the following steps:
s11, establishing a tongue prescription database according to clinical data, wherein each record of the database comprises a tongue picture and a corresponding treatment prescription, and each treatment prescription comprises a plurality of medicinal materials and the dosage of each medicinal material. Sorting out records, and eliminating prescriptions which do not accord with the traditional Chinese medicine standard and are wrongly input;
s12, respectively standardizing the dose of each medicinal material, compressing the dose of each medicinal material to the same range, and eliminating dimensional interference;
s13, representing each prescription in the database into a prescription vector T with dimension H, wherein the ith element T of the prescription vector T i ＝M+C i ，C i The standardized dose used for the ith drug in the prescription is expressed, M is the hyperparameter, and H is the total number of drugs in the database.
S2, training a feature extractor for obtaining a tongue picture by the server according to vector representation of the prescription, and specifically comprising the following steps:
s21, designing a tongue picture feature extractor, which specifically comprises the following processes: selecting any publicly available image classification network pre-trained in ImageNet, and replacing a classifier at the last layer of the image classification network with a full connection layer FC with an output dimension D; this part is the feature extractor BONE. And adding a fully-connected network MLP with an input dimension D and an output dimension H after the feature extractor BONE to predict a prescription vector O of the tongue picture. In this embodiment, the image classification network is selected as ResNet18, the feature dimension D output by the feature extractor BONE is 512, and the fully-connected network MLP for predicting the prescription vector is set to include two fully-connected layers, where the first fully-connected layer has an input dimension D and an output dimension D, and the second fully-connected layer has an input dimension D and an output dimension H, where the dimension H of the predicted prescription vector is 634, the total number of the medicinal materials in the database.
S22, repeating and iterating the step until the optimization loss is not reduced any more, thereby completing the training process of the feature extractor, which specifically comprises the following steps: randomly extracting two tongue pictures as a pair, and standardizing the two pictures according to the standardization mode of an image classification network BONE; respectively extracting the two tongue picture feature vectors by using BONE to obtain two feature vectors V 1 ,V 2 (ii) a Will V 1 ,V 2 Input into MLP to obtain predicted prescription vector O 1 ,O 2 . And calculating the feature similarity S between the two tongue pictures and the corresponding prescription similarity L of the two tongue pictures. Similarity S-pass between feature vectorsAnd the over-similarity calculation module calculates by using the inner product of the vectors, calculates the similarity L between the real prescriptions T by using the cosine similarity of the prescription vectors, and adjusts parameters of BONE and MLP by using an error back propagation algorithm through two optimization losses. The calculation formula of the inner product of any two vectors x and y is as follows:
two eigenvectors V i 、V j The feature similarity S of (A) is the inner product of the two<V i ,V j >Two prescription vectors T i 、T j The similarity calculation formula is as follows:
optimizing two optimization penalties L using Adam optimizer p ,L r The learning rate thereof is set to 0.003. Loss function L p For minimizing the error between the predicted prescription vector O and the true prescription vector T. Loss function L r For minimizing the difference of the ranking results sorted using feature similarity and using prescription vector similarity, L p The calculation formula of (2) is as follows:
loss function L r The calculation formula of (2) is as follows:
L r ＝(1-L)log(1+exp(α(S+β)))+Llog(1+exp(-α(S+β)))
the hyper-parameters alpha, beta are set to 10 and 0.5 respectively, the overall optimization objective is L total ＝L p +L r 。
S3, for the tongue picture I of any database, extracting the feature vector V of the tongue picture I by using the trained feature extractor BONE in the step S22 I ；
During the actual operation of the system, namely the on-line operation stage, the steps are
S4, the client submits a tongue picture to the server side as a query;
s5, extracting a tongue picture Q submitted by a client as a query by using a feature extractor BONE to obtain a feature representation V Q ；
S6, calculating a feature vector V of the query tongue picture Q by using a similarity calculation module Q And the feature vector V of any database picture I I Feature similarity of<V Q ,V I >；
And S7, reordering all tongue-prescription pairs (I, T) in the database by using the prescription recommending module and taking the feature similarity obtained by calculation in the step S5 as a standard, recommending a plurality of prescriptions with high similarity for the tongue picture Q query, and returning the prescriptions to the client.
The above embodiments are preferred embodiments of the present invention, but the present invention is not limited to the above embodiments, and any other changes, modifications, substitutions, combinations, and simplifications which do not depart from the spirit and principle of the present invention should be construed as equivalents thereof, and all such changes, modifications, substitutions, combinations, and simplifications are intended to be included in the scope of the present invention.
Claims (6)
1. A prescription recommendation method based on tongue image retrieval is characterized by comprising the following steps:
s1, a server side analyzes prescription information of a database and vectorizes and represents a prescription, and the method specifically comprises the following steps:
s11, sorting database records, and eliminating prescriptions which do not accord with traditional Chinese medicine standards and are wrongly input;
s12, respectively standardizing the dose of each medicinal material, compressing the dose of each medicinal material to the same range, and eliminating dimensional interference;
s13, representing each prescription in the database into a prescription vector T with dimension H, wherein the ith element T of the prescription vector T i ＝M+C i ，C i The normalized dosage of the ith medicinal material in the prescription is shown, M is a hyperparameter, and H is the total dosage in the databaseThe number of medicinal materials;
s2, training a feature extractor for obtaining a tongue picture according to vector representation of the prescription;
the feature extractor of the tongue picture is as follows: selecting an image classification network pre-trained in ImageNet, and replacing a classifier at the last layer of the image classification network with a full-connection layer FC with an output dimension of D; this part is the feature extractor BONE; after the feature extractor BONE, adding a fully-connected neural network MLP with an input dimension D and an output dimension H for predicting a prescription vector O of a tongue picture, wherein the feature dimension D output by the feature extractor BONE is 512, the fully-connected neural network MLP for predicting the prescription vector is set to comprise two fully-connected layers, the input dimension of the first fully-connected layer is D, the output dimension is D, the input dimension of the second fully-connected layer is D, the output dimension is H, and the dimension H of the predicted prescription vector is 634 of the total number of medicinal materials in the database;
training is a process of repeated iteration, and each iteration specifically comprises: randomly extracting two tongue pictures as a pair, and standardizing the two tongue pictures according to the standardization mode of an image classification network BONE; respectively extracting the two tongue picture feature vectors by using an image classification network BONE to obtain two feature vectors V 1 ,V 2 (ii) a Will V 1 ,V 2 Inputting into a fully-connected neural network MLP to obtain a predicted prescription vector O 1 ,O 2 (ii) a Adjusting parameters of BONE and MLP by using an error back propagation algorithm through two optimization losses;
the specific optimization comprises the following steps:
1) For each tongue picture which is randomly extracted, minimizing the difference between the predicted prescription vector O of the tongue picture and the corresponding real prescription vector T;
2) For a pair of tongue pictures which are randomly extracted, the difference between the predicted feature similarity S and the true prescription similarity L of the tongue pictures is minimized;
optimizing two optimization penalties L using Adam optimizer p ,L r Its learning rate is set to 0.003; loss function L p For minimizing the error between the predicted prescription vector O and the true prescription vector T;loss function L r For minimizing the difference of the ranking results sorted using feature similarity and using prescription vector similarity, L p The calculation formula of (2) is as follows:
loss function L r The calculation formula of (2) is as follows:
L r ＝(1-L)log(1+exp(α(S+β)))+Llog(1+exp(-α(S+β)))
the hyper-parameters alpha, beta are set to 10 and 0.5 respectively, the overall optimization objective is L total ＝L p +L r ；
S3, pre-extracting a feature vector of each tongue picture in the database by using a feature extractor;
s4, submitting tongue pictures to be inquired by the client;
s5, the server side extracts the feature vector of the tongue picture by using a feature extractor;
s6, the server side calculates the feature similarity between the tongue picture and each tongue picture in the database by using a similarity calculation module; calculating the feature similarity S between the two tongue pictures and the prescription similarity L corresponding to the two tongue pictures, calculating the similarity S between feature vectors through a similarity calculation module, calculating by using the inner product of the vectors, and calculating the similarity L between the real prescription T by using the cosine similarity of the prescription vectors, wherein the calculation formula of the inner product of any two vectors x and y is as follows:
two eigenvectors V i 、V j The feature similarity S of (A) is the inner product of the two<V i ,V j >Two prescription vectors T i 、T j The similarity calculation formula is as follows:
and S7, the server uses a prescription recommendation module to sort according to the feature similarity obtained by calculation in the S6, and returns the prescription corresponding to the K tongue pictures with higher similarity to the client.
2. The tongue image retrieval-based prescription recommendation method according to claim 1, wherein the vectorized representation of the prescription of step S1 comprises the medicinal materials used by the prescription and the dosage information thereof.
3. The tongue image retrieval-based prescription recommendation method as claimed in claim 2, wherein the feature extractor takes a tongue picture as an input, and extracts a feature vector in D dimension through the feature extractor.
4. The tongue image retrieval-based prescription recommendation method according to claim 3, wherein the fully-connected neural network MLP is composed of a plurality of fully-connected layers, and the output dimension of the last fully-connected layer FCn is H, which is the same as the dimension H of the prescription vector.
5. The tongue image retrieval-based prescription recommendation method as claimed in claim 4, wherein in step S2, in training the feature extractor, iteratively optimizing two optimization losses until convergence is repeated to adjust the parameters of the deep convolutional neural network, and performing target optimization according to the predicted D-dimension feature vector.
6. The tongue image retrieval-based prescription recommendation method according to claim 5, wherein in step S7, the prescription recommendation module sorts the tongue images according to feature similarity between feature vectors of the tongue images provided by the client and feature similarities of all tongue images in the database, recommends the prescription corresponding to the K tongue images with the highest similarity, and returns the prescription to the client.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011433472.8A CN112652373B (en) | 2020-12-09 | 2020-12-09 | Prescription recommendation method based on tongue image retrieval |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011433472.8A CN112652373B (en) | 2020-12-09 | 2020-12-09 | Prescription recommendation method based on tongue image retrieval |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112652373A CN112652373A (en) | 2021-04-13 |
| CN112652373B true CN112652373B (en) | 2022-10-25 |
Family
ID=75350776
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011433472.8A Active CN112652373B (en) | 2020-12-09 | 2020-12-09 | Prescription recommendation method based on tongue image retrieval |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112652373B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116168807B (en) * | 2022-12-19 | 2024-03-19 | 广州华见智能科技有限公司 | Brain wave-based traditional Chinese medicine doctor diagnosis and treatment system |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2001357131A (en) * | 2000-06-12 | 2001-12-26 | Kanai Tokichi Shoten:Kk | Method for providing prescription of herbal medicine through communication network |
| CN108182967A (en) * | 2017-12-14 | 2018-06-19 | 华南理工大学 | A kind of traditional Chinese medical science medicinal material based on deep neural network recommends method |
| CN108877923A (en) * | 2018-05-03 | 2018-11-23 | 华南理工大学 | A method of the tongue fur based on deep learning generates prescriptions of traditional Chinese medicine |
| CN109711306A (en) * | 2018-12-19 | 2019-05-03 | 新绎健康科技有限公司 | A kind of method and apparatus obtaining facial characteristics based on depth convolutional neural networks |
| CN110223751A (en) * | 2019-05-16 | 2019-09-10 | 平安科技（深圳）有限公司 | Prescription evaluation method, system and computer equipment based on medical knowledge map |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107451407A (en) * | 2017-07-31 | 2017-12-08 | 广东工业大学 | A kind of traditional Chinese medical science intelligent diagnosing method, system and traditional medical system |
-
2020
- 2020-12-09 CN CN202011433472.8A patent/CN112652373B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2001357131A (en) * | 2000-06-12 | 2001-12-26 | Kanai Tokichi Shoten:Kk | Method for providing prescription of herbal medicine through communication network |
| CN108182967A (en) * | 2017-12-14 | 2018-06-19 | 华南理工大学 | A kind of traditional Chinese medical science medicinal material based on deep neural network recommends method |
| CN108877923A (en) * | 2018-05-03 | 2018-11-23 | 华南理工大学 | A method of the tongue fur based on deep learning generates prescriptions of traditional Chinese medicine |
| CN109711306A (en) * | 2018-12-19 | 2019-05-03 | 新绎健康科技有限公司 | A kind of method and apparatus obtaining facial characteristics based on depth convolutional neural networks |
| CN110223751A (en) * | 2019-05-16 | 2019-09-10 | 平安科技（深圳）有限公司 | Prescription evaluation method, system and computer equipment based on medical knowledge map |
Non-Patent Citations (1)
| Title |
|---|
| 基于医疗数据分析的疾病预测与处方推荐算法研究;陈亚明;《中国优秀硕士学位论文全文数据 医药卫生科技辑》;20191230(第10期);E054-26 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112652373A (en) | 2021-04-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Chen et al. | Deep feature learning for medical image analysis with convolutional autoencoder neural network | |
| Kumar et al. | Breast cancer classification of image using convolutional neural network | |
| Rodrigues et al. | Comparing convolutional neural networks and preprocessing techniques for HEp-2 cell classification in immunofluorescence images | |
| Huang et al. | Neural network classifier with entropy based feature selection on breast cancer diagnosis | |
| CN109378066A (en) | A kind of control method and control device for realizing disease forecasting based on feature vector | |
| Guo et al. | Interstitial lung disease classification using improved DenseNet | |
| Gupta et al. | Improved brain age estimation with slice-based set networks | |
| CN114023412A (en) | ICD code prediction method and system based on joint learning and denoising mechanism | |
| Gürler et al. | Foreseeing brain graph evolution over time using deep adversarial network normalizer | |
| CN114913982A (en) | End-stage renal disease complication risk prediction system based on contrast learning | |
| Akbarian et al. | Evaluating knowledge transfer in the neural network for medical images | |
| CN112652373B (en) | Prescription recommendation method based on tongue image retrieval | |
| CN113160986A (en) | Model construction method and system for predicting development of systemic inflammatory response syndrome | |
| Qian | Exploration of machine algorithms based on deep learning model and feature extraction | |
| Ayalew et al. | X-Ray image-based COVID-19 detection using deep learning | |
| Wu et al. | A pseudo label-wise attention network for automatic ICD coding | |
| CN113284627A (en) | Medication recommendation method based on patient characterization learning | |
| Alwan et al. | Diabetes diagnosis system using modified Naive Bayes classifier | |
| Khader et al. | Cascaded Cross-Attention Networks for Data-Efficient Whole-Slide Image Classification Using Transformers | |
| Francis et al. | Ensemble learning approach for multi-class classification of Alzheimer’s stages using magnetic resonance imaging | |
| Cui et al. | Intelligent recommendation for departments based on medical knowledge graph | |
| Falissard et al. | A deep artificial neural network based model for underlying cause of death prediction from death certificates | |
| CN115831356B (en) | Auxiliary prediction diagnosis method based on artificial intelligence algorithm | |
| Nasiriyan et al. | Sparse Connectivity and Activity Using Sequential Feature Selection in Supervised Learning | |
| Kaur et al. | Integrating Transfer Learning and Deep Neural Networks for Accurate Medical Disease Diagnosis from Multi-Modal Data |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |