CN107863147B - Medical diagnosis method based on deep convolutional neural network - Google Patents
Medical diagnosis method based on deep convolutional neural network Download PDFInfo
- Publication number
- CN107863147B CN107863147B CN201711002335.7A CN201711002335A CN107863147B CN 107863147 B CN107863147 B CN 107863147B CN 201711002335 A CN201711002335 A CN 201711002335A CN 107863147 B CN107863147 B CN 107863147B
- Authority
- CN
- China
- Prior art keywords
- diagnosed
- medical
- electronic medical
- medical record
- word
- 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
- 238000000034 method Methods 0.000 title claims abstract description 92
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 90
- 238000003745 diagnosis Methods 0.000 title claims abstract description 59
- 239000013598 vector Substances 0.000 claims abstract description 155
- 201000010099 disease Diseases 0.000 claims abstract description 61
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 claims abstract description 61
- 239000011159 matrix material Substances 0.000 claims abstract description 35
- 238000012549 training Methods 0.000 claims description 27
- 238000012216 screening Methods 0.000 claims description 15
- 238000001914 filtration Methods 0.000 claims description 14
- 238000011176 pooling Methods 0.000 claims description 10
- 230000011218 segmentation Effects 0.000 claims description 8
- 238000004458 analytical method Methods 0.000 claims description 4
- 238000005516 engineering process Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 abstract description 7
- 230000007547 defect Effects 0.000 abstract description 3
- 230000008569 process Effects 0.000 description 9
- 230000000694 effects Effects 0.000 description 8
- 238000013135 deep learning Methods 0.000 description 7
- 230000006870 function Effects 0.000 description 7
- 238000013528 artificial neural network Methods 0.000 description 6
- 238000012423 maintenance Methods 0.000 description 6
- 238000012545 processing Methods 0.000 description 6
- 238000013461 design Methods 0.000 description 5
- 206010019233 Headaches Diseases 0.000 description 3
- 230000008901 benefit Effects 0.000 description 3
- 238000010276 construction Methods 0.000 description 3
- 231100000869 headache Toxicity 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 230000000474 nursing effect Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 229940079593 drug Drugs 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 230000003862 health status Effects 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000012886 linear function Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 210000000653 nervous system Anatomy 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000000844 transformation Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/334—Query execution
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Databases & Information Systems (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The invention discloses a medical diagnosis method based on a deep convolutional neural network, which comprises the following steps: acquiring a word vector matrix corresponding to the electronic medical record to be diagnosed; inputting a word vector matrix corresponding to the electronic medical record to be diagnosed into a pre-constructed deep convolutional neural network model to obtain a feature vector of the electronic medical record to be diagnosed; and classifying the feature vectors of the electronic medical record to be diagnosed by using the classifier to obtain the disease probability of each disease corresponding to the electronic medical record to be diagnosed. The method applies the convolutional neural network to the text semantic understanding of the medical electronic medical record and performs auxiliary medical diagnosis, and can effectively overcome the defects of the rule extraction and matching based method.
Description
Technical Field
The invention relates to the technical field of medical informatization, in particular to a medical diagnosis method based on a deep convolutional neural network.
Background
Clinical decision support (CDSS) refers to the use of relevant and systematic Clinical knowledge and patient basic information and disease information to enhance medically relevant decisions and actions and improve medical quality and medical service levels. The CDSS is an important means for improving medical quality, and the primary purposes of the CDSS are to evaluate and improve medical quality, reduce medical errors and control medical expenses.
At present, most CDSS in the world is realized based on rule extraction and matching, and the following technical scheme is adopted: based on a clinical database, establishing logic association knowledge points by collecting, sorting, classifying, filtering, processing and the like information; adopting warning reminding, information buttons, group medical advice (medical advice package), document management and related data expression forms; decision support in aspects of diagnosis, treatment, nursing, operation, reasonable medication and the like is carried out on diseases; decision support in the aspects of suggestion, reminding, alarming, calculation and prediction is provided for the diagnosis and treatment of a clinician;
the rule-based extraction and matching method has its inherent drawbacks, in particular the following:
1. there is a problem of semantic ambiguity. For example, there may be a plurality of different words such as "headache" and "headache" when describing headache, but the description method with the same semantics needs to contain all possible descriptions as much as possible when constructing the knowledge base, which causes redundancy and inefficiency of the knowledge base, and also causes a decrease in matching accuracy because all possible cases cannot be contained.
2. The types of cases encountered by departments of hospitals are various and have large differences, a knowledge base needs to be established for hundreds of diseases of the departments, and the departments are extremely complicated, are not beneficial to management and maintenance, and cause extremely low efficiency;
3. the construction based on the knowledge base is used for assisting medical diagnosis, so that the suspicion of 'talking about on paper' is inevitable, the medical diagnosis is required to be known to depend on clinical practice and experience extremely, the construction rules are described on a book and then the clinical practice is guided, the effect of inversion at this end can be achieved, and the purpose of assisting diagnosis is difficult to achieve.
Therefore, a medical diagnosis method based on a deep convolutional neural network becomes an urgent technical problem to be solved.
Disclosure of Invention
The object of the present invention is to solve at least to some extent one of the above mentioned technical problems.
Therefore, the first purpose of the invention is to provide a medical diagnosis method based on a deep convolutional neural network, which applies the convolutional neural network to the semantic understanding of the text of the medical electronic medical record and performs auxiliary medical diagnosis, and can effectively overcome the defects of the rule extraction and matching-based method. The method can effectively solve the problem of semantic gap and effectively eliminate the influence caused by synonymy different words in electronic medical records such as writing; the method can be used for the conditions of various disease types and large differences of each department, a unified model frame can be constructed without constructing corresponding rules and matching algorithms for each disease type, then the model is trained by using historical data of each disease type, the effect that a plurality of disease types can be pre-diagnosed by only one model can be achieved, the method is very suitable for management and maintenance, and the expandability is very strong; the method does not need to artificially design rules and characteristics, all the characteristics and the rules learned by the model come from a large amount of clinical historical data, clinical historical data are used for guiding clinical decision, and compared with a rule-based auxiliary diagnosis method, the method has very strong practical guiding significance.
In order to achieve the above object, a method for medical diagnosis based on a deep convolutional neural network according to an embodiment of the first aspect of the present invention includes:
acquiring a word vector matrix corresponding to the electronic medical record to be diagnosed;
inputting the word vector matrix corresponding to the electronic medical record to be diagnosed into a pre-constructed deep convolutional neural network model to obtain a characteristic vector corresponding to the electronic medical record to be diagnosed;
and classifying the feature vectors of the electronic medical record to be diagnosed by using a classifier to obtain the disease probability of each disease corresponding to the electronic medical record to be diagnosed.
The method for acquiring the word vector matrix corresponding to the electronic medical record to be diagnosed comprises the following steps:
performing at least one operation of information filtering, screening, word segmentation and statistics on the electronic medical record to be diagnosed to obtain each medical vocabulary of the medical record to be diagnosed;
acquiring a word vector corresponding to the medical vocabulary of the medical record to be diagnosed in the preset word vector database, wherein the preset word vector database stores the corresponding relation between the medical vocabulary and the word vector;
and generating a word vector matrix corresponding to the electronic disease to be diagnosed according to the word vector corresponding to the medical vocabulary of each electronic medical record to be diagnosed.
The method, before the obtaining of the word vector matrix corresponding to the electronic medical record to be diagnosed, includes:
acquiring each medical vocabulary in a medical word bank;
inputting the medical vocabulary in the medical lexicon into a Word2Vec model established in advance, and acquiring Word vectors corresponding to the medical vocabulary;
and forming a word vector sample by using the word vectors corresponding to the medical vocabulary, and storing the word vector sample in a preset word vector database.
The method as described above, before the obtaining of each medical vocabulary in the medical lexicon, includes:
acquiring a plurality of diagnosed electronic medical records;
information filtering is carried out on each diagnosed electronic medical record by utilizing an information filtering technology, and a medical vocabulary set is obtained;
and counting the word frequency of each medical word in the medical word set, screening each medical word according to a set screening rule, and establishing the medical word bank according to a screening result.
The method, before the obtaining of the word vector matrix corresponding to the electronic medical record to be diagnosed, includes:
acquiring word vector matrixes corresponding to a plurality of diagnosed electronic medical records, and taking the word vector matrixes corresponding to the diagnosed electronic medical records as training samples;
and training the training sample to construct the deep convolutional neural network model.
The method as described above, after the constructing the deep convolutional neural network model, comprising:
acquiring a doctor diagnosis result corresponding to each diagnosed electronic medical record;
acquiring a feature vector corresponding to the diagnosed electronic medical record output by the deep convolutional neural network model aiming at each diagnosed electronic medical record;
classifying the feature vectors of the diagnosed electronic medical record by using the classifier to obtain the illness probability of each illness of the diagnosed electronic medical record;
and analyzing the disease probability of each disease of the diagnosed electronic medical record and a doctor diagnosis result corresponding to the diagnosed electronic medical record by using a preset algorithm, and correcting the parameters of the deep convolutional neural network model and the parameters of the classifier according to the analysis result.
In the method, the preset algorithm is a reverse iteration algorithm.
In the method described above, the classifier is a softmax classifier.
The method as described above, the deep convolutional neural network model comprising: input layer, convolution layer, pooling layer, full connection layer.
As with the method described above, the convolutional layer includes a plurality of convolutional kernels of different sizes.
Additional aspects and advantages of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.
Drawings
The above and/or additional aspects and advantages of the present invention will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings of which,
FIG. 1 is a schematic flow chart of a method for medical diagnosis based on deep convolutional neural network according to an embodiment of the present invention;
FIG. 2 is a schematic flow chart of a method for deep convolutional neural network-based medical diagnosis according to another embodiment of the present invention;
FIG. 3 is a schematic flow chart of a method for medical diagnosis based on deep convolutional neural network according to another embodiment of the present invention;
FIG. 4 is a schematic flow chart of a method for deep convolutional neural network-based medical diagnosis according to yet another embodiment of the present invention;
FIG. 5 is a flow chart illustrating a method for deep convolutional neural network-based medical diagnosis according to an embodiment of the present invention;
FIG. 6 is a schematic flow chart of a method for deep convolutional neural network-based medical diagnosis according to another embodiment of the present invention;
FIG. 7 is an architectural diagram of an exemplary deep convolutional neural network model in accordance with an embodiment of the present invention.
Detailed Description
Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the drawings are illustrative and intended to be illustrative of the invention and are not to be construed as limiting the invention.
A method for medical diagnosis based on a deep convolutional neural network according to an embodiment of the present invention is described below with reference to the accompanying drawings.
The deep learning related art is briefly described here. Deep learning (deep learning) is a branch of machine learning, which is a method for learning data based on characterization, and it attempts to abstract data at a high level using multiple processing layers containing complex structures or consisting of multiple nonlinear transformations. The benefit of deep learning is to replace the manual feature acquisition with unsupervised or semi-supervised feature learning and hierarchical feature extraction efficient algorithms. The goal of token learning is to seek better representations and create better models to learn these representations from large-scale unlabeled data. Expressing ways that resemble advances in neuroscience and loosely create an understanding of information processing and communication patterns in similar nervous systems, such as neural coding, in an attempt to define relationships between responses that pull neurons and relationships between electrical activity of neurons in the brain. Several deep learning frameworks such as deep neural networks, convolutional neural networks, and deep belief networks and recurrent neural networks have been applied to the fields of computer vision, speech recognition, natural language processing, audio recognition, and bioinformatics and have achieved excellent results.
Convolutional Neural Network (CNN) is a deep learning framework, which is a feed-forward Neural Network, and is composed of one or more Convolutional layers and top fully connected layers (corresponding to classical Neural networks), and also includes associated weights and pooling layers (pooling layers). This structure enables the convolutional neural network to utilize a two-dimensional structure of the input data. Compared with other deep and feedforward neural networks, the convolutional neural network needs fewer estimated parameters, so that the convolutional neural network becomes an attractive deep learning structure. A great deal of research and application has already demonstrated that convolutional neural networks have very strong feature extraction, expression, semantic understanding capabilities on images and text, do not need to artificially design features, and learn various features from a large amount of data by themselves.
Fig. 1 is a flowchart illustrating a method for medical diagnosis based on a deep convolutional neural network according to an embodiment of the present invention. As shown in fig. 1, the method for medical diagnosis based on a deep convolutional neural network provided in this embodiment includes:
s101, obtaining a word vector matrix corresponding to the electronic medical record to be diagnosed.
Specifically, the medical record is the original record of the whole process of the diagnosis and treatment of the patient in the hospital, and comprises a first page, a course record, an examination and examination result, a medical order, an operation record, a nursing record and the like. The electronic medical record not only refers to static medical record information, but also comprises provided related services, namely information which is managed in an electronic mode and related to the health status of the whole life of a person and medical care behaviors, and all process information related to the acquisition, storage, transmission, processing and utilization of patient information. It should be noted that the electronic medical record not only includes medical vocabulary but also includes a lot of non-medical vocabulary such as personal sensitive information, and in the automated medical diagnosis, it is necessary to remove interference items such as non-medical vocabulary and obtain medical vocabulary in the electronic medical record to ensure the reliability of the medical diagnosis.
In this embodiment, first, after unstructured and deleted non-medical vocabularies such as personal sensitive information and the like, the plain text content of the electronic medical record to be diagnosed is obtained; secondly, performing text recognition on the plain text content of the electronic medical record to be diagnosed, and extracting medical words included in the electronic medical record to be diagnosed; and then, obtaining the word vector of each medical vocabulary, and obtaining the word vector matrix corresponding to the electronic medical record to be diagnosed according to the word vector of each medical vocabulary.
In this embodiment, the word vector may be understood as a form in which words in a language are expressed in a mathematical manner, and the word vector can effectively solve the problem of "semantic gap" and effectively eliminate semantic ambiguity. The words with similar semantics have similar distances in the vector space, so that the influence caused by synonymy different words in the writing of the electronic medical record can be effectively eliminated.
A brief description is given here of how a word vector matrix is obtained from word vectors. For example, a word vector matrix represented by L × F for each electronic medical record is defined in advance, the first row represents a word vector of the first word of the text, the second row represents a word vector of the second word, and so on. For the electronic medical record with the word length exceeding L words, intercepting the front L words; for an electronic medical record with less than L words, the subsequent word vectors are filled with the number 0. Thus, each electronic medical record is uniformly expressed as a word vector matrix of L × F, which is denoted as M and can be expressed as:
wherein,
denotes a connection symbol, by Xi:jRepresenting a matrix of the ith to jth word vectors, each row representing a word vector, X1:LAnd representing a matrix formed by the 1 st word vector to the L < th > word vector, wherein i, j, L and F are positive integers.
S102, inputting a word vector matrix corresponding to the electronic medical record to be diagnosed into a pre-constructed deep convolutional neural network model, and acquiring a feature vector corresponding to the electronic medical record to be diagnosed.
In the embodiment, the word vector can effectively solve the problem of 'semantic gap', so that semantic ambiguity is effectively eliminated, the word vector matrix corresponding to the electronic medical record to be diagnosed is input to the pre-constructed deep convolutional neural network model, and the acquired feature vector can effectively eliminate the influence caused by synonymy different words in writing in the electronic medical record.
The deep convolutional neural network model in the embodiment is obtained by training massive diagnosed electronic medical records from clinical historical data, so that the feature vector output by the deep convolutional neural network model has very strong practical guiding significance.
In this embodiment, the electronic medical records do not need to be classified according to disease types, a deep convolutional neural network model is established according to disease types, and one deep convolutional neural network model can predict the effect of a plurality of disease types, so that the deep convolutional neural network model is very suitable for management and maintenance and has very strong expandability.
S103, classifying the feature vectors of the electronic medical record to be diagnosed by using the classifier, and acquiring the disease probability of each disease corresponding to the electronic medical record to be diagnosed.
Specifically, in the embodiment, the classifier is used for performing data mining on the feature vectors output by the deep convolutional neural network model, and the disease probability of each disease in the feature vectors is analyzed for reference by medical personnel, so that a better medical auxiliary diagnosis effect is realized. Optionally, the classifier is a softmax classifier, which can better analyze the prevalence probability of each condition.
According to the medical diagnosis method based on the deep convolutional neural network, a word vector matrix corresponding to an electronic medical record to be diagnosed is obtained; inputting a word vector matrix corresponding to the electronic medical record to be diagnosed into a pre-constructed deep convolutional neural network model, and acquiring a feature vector corresponding to the electronic medical record to be diagnosed; and classifying the feature vectors of the electronic medical record to be diagnosed by using the classifier to obtain the disease probability of each disease corresponding to the electronic medical record to be diagnosed. The method applies the convolutional neural network to the text semantic understanding of the medical electronic medical record and performs auxiliary medical diagnosis, and can effectively overcome the defects of the rule extraction and matching based method. The method can effectively solve the problem of semantic gap and effectively eliminate the influence caused by synonymy different words in electronic medical records such as writing; the method can be used for the conditions of various disease types and large differences of each department, a unified model frame can be constructed without constructing corresponding rules and matching algorithms for each disease type, then the model is trained by using historical data of each disease type, the effect that a plurality of disease types can be pre-diagnosed by only one model can be achieved, the method is very suitable for management and maintenance, and the expandability is very strong; the method does not need to artificially design rules and characteristics, all the characteristics and the rules learned by the model come from a large amount of clinical historical data, clinical historical data are used for guiding clinical decision, and compared with a rule-based auxiliary diagnosis method, the method has very strong practical guiding significance.
Fig. 2 is a flowchart illustrating a method for medical diagnosis based on a deep convolutional neural network according to another embodiment of the present invention. On the basis of the above embodiment, this embodiment optimizes "obtaining a word vector matrix corresponding to the electronic medical record to be diagnosed" to "perform at least one operation of information filtering, screening, word segmentation, and statistics on the electronic medical record to be diagnosed, so as to obtain each medical vocabulary of the medical record to be diagnosed; acquiring a word vector corresponding to a medical vocabulary of a medical record to be diagnosed in a preset word vector database, wherein the preset word vector database stores the corresponding relation between the medical vocabulary and the word vector; and generating a word vector matrix corresponding to the electronic disease to be diagnosed according to the word vector corresponding to the medical vocabulary of each electronic medical record to be diagnosed. "
As shown in fig. 2, the method for medical diagnosis based on a deep convolutional neural network provided in this embodiment includes:
s201, performing at least one operation of information filtering, screening, word segmentation and statistics on the electronic medical record to be diagnosed to obtain each medical vocabulary of the medical record to be diagnosed.
Specifically, the medical vocabulary is reserved and the non-medical vocabulary is filtered by performing operations such as information filtering, screening, word segmentation, statistics and the like on the electronic medical record, so that the processing speed of the whole system is improved.
S202, word vectors corresponding to medical vocabularies of the medical record to be diagnosed are obtained from a preset word vector database, wherein the preset word vector database stores corresponding relations between the medical vocabularies and the word vectors.
Specifically, the preset word vector database stores word vectors corresponding to a large number of medical vocabularies, and also stores corresponding relationships between the medical vocabularies and the word vectors. After each medical vocabulary of the medical record to be diagnosed is obtained, the word vector corresponding to the medical vocabulary of the medical record to be diagnosed can be efficiently inquired in the preset word vector database based on the corresponding relation between the medical vocabulary and the word vector. In the embodiment, the word vector corresponding to the medical vocabulary of the medical record to be diagnosed is efficiently acquired by using the preset word vector database storing the word vectors corresponding to a large number of medical vocabularies.
S203, generating a word vector matrix corresponding to the electronic medical record to be diagnosed according to the word vectors corresponding to the medical vocabularies of the electronic medical record to be diagnosed.
Specifically, how to generate the word vector matrix corresponding to the electronic disease to be diagnosed can refer to the brief introduction of how to obtain the word vector matrix according to the word vectors, and details are not described herein again.
S204, inputting the word vector matrix corresponding to the electronic medical record to be diagnosed into a pre-constructed deep convolutional neural network model, and acquiring the feature vector corresponding to the electronic medical record to be diagnosed.
S205, classifying the feature vectors of the electronic medical record to be diagnosed by using the classifier, and acquiring the disease probability of each disease corresponding to the electronic medical record to be diagnosed.
The implementation manners of steps S204 and S205 in this embodiment are the same as the implementation manners of S102 and S103 in the above embodiment, and are not described herein again.
According to the medical diagnosis method based on the deep convolutional neural network, the medical vocabulary is reserved and the non-medical vocabulary is filtered by performing information filtering operation on the electronic medical record, so that the processing speed of the whole system is improved; the word vector corresponding to the medical vocabulary of the medical record to be diagnosed is efficiently acquired by utilizing the preset word vector database which stores the word vectors corresponding to a large number of medical vocabularies.
Fig. 3 is a flowchart illustrating a method for medical diagnosis based on a deep convolutional neural network according to another embodiment of the present invention. On the basis of the above embodiments, the present embodiment optimizes and explains the establishment of the preset word vector database.
As shown in fig. 3, the method for medical diagnosis based on a deep convolutional neural network provided in this embodiment includes:
s301, acquiring each medical vocabulary in the medical lexicon.
Specifically, a medical word bank is established in advance, and a large amount of medical words are recorded in the medical word bank.
S302, inputting the medical vocabulary in the medical vocabulary bank into a Word2Vec model which is established in advance, and obtaining Word vectors corresponding to the medical vocabulary.
The Word2Vec model is an efficient tool for expressing words as real-valued vectors, each Word can be mapped into a K-dimensional real vector through training (K is generally a hyper-parameter in the model), and the similarity in vector space can be used for representing the semantic similarity of texts.
And S303, forming word vector samples from the word vectors corresponding to the medical vocabularies, and storing the word vector samples in a preset word vector database.
Specifically, the preset word vector database stores word vectors corresponding to a large number of medical vocabularies, and also stores corresponding relationships between the medical vocabularies and the word vectors. For example, after the medical vocabulary is acquired, the word vector corresponding to the medical vocabulary can be efficiently searched in the preset word vector database based on the corresponding relationship between the medical vocabulary and the word vector.
According to the medical diagnosis method based on the deep convolutional neural network, each medical vocabulary in a medical lexicon is obtained; inputting medical words in a medical Word bank into a Word2Vec model which is established in advance, and acquiring Word vectors corresponding to the medical words; and forming word vector samples by the word vectors corresponding to the medical vocabularies, and storing the word vector samples in a preset word vector database. The medical word bank records massive medical words, and accordingly, the established preset word vector database stores word vectors corresponding to the massive medical words, so that the efficient operation of the whole system is facilitated. In addition, Word vectors corresponding to medical vocabularies can be efficiently acquired by using the Word2Vec model.
Fig. 4 is a flowchart illustrating a method for medical diagnosis based on a deep convolutional neural network according to another embodiment of the present invention. On the basis of the above embodiments, the present embodiment explains the establishment of a medical thesaurus.
As shown in fig. 4, the method for medical diagnosis based on a deep convolutional neural network provided in this embodiment includes:
s401, obtaining a plurality of diagnosed electronic medical records.
For example, the diagnosed electronic medical records in the present embodiment may be electronic medical records extracted from medical information management systems of hospitals for main diseases of each department over the past several years. After acquiring a large amount of diagnosed electronic medical records, the diagnosed electronic medical records are unstructured, and information such as personal sensitive information is deleted, so that the plain text content of the electronic medical records is obtained. The plurality of diagnosed electronic medical records may be a preset number of diagnosed electronic medical records.
S402, information filtering is carried out on each diagnosed electronic medical record by utilizing an information filtering technology, and a medical vocabulary set is obtained.
For example, in the embodiment, semantic understanding is performed on the obtained plain text content of the diagnosed electronic medical record first, and medical terms included in each diagnosed electronic medical record are obtained. If the electronic medical record text is Chinese, word segmentation operation is required to be carried out to obtain a segmentation word segment of each electronic medical record text, and the segmentation word segment corresponds to each word in the English text.
In this embodiment, a large number of medical vocabularies are obtained by analyzing a large number of diagnosed electronic medical records, and accordingly, the sample capacity of a medical vocabulary set composed of the large number of medical vocabularies is large.
And S403, counting the word frequency of each medical word in the medical word set, screening each medical word according to a set screening rule, and establishing a medical word bank according to a screening result.
For example, a medical thesaurus is defined, initially empty. And counting the word frequency of each medical vocabulary in the medical vocabulary set, wherein the set screening rule can be that the medical vocabulary with the word frequency larger than a set threshold value is added into a medical vocabulary bank. In the embodiment, the statistical medical vocabularies are screened, so that the established medical word bank is closer to the medical industry, and the authority is higher.
According to the medical diagnosis method based on the deep convolutional neural network, a large-volume medical vocabulary set of a sample is obtained by obtaining a large number of diagnosed electronic medical records, and the statistical medical vocabularies are screened, so that the established medical lexicon is closer to the medical industry, and the authority is higher.
Fig. 5 is a flowchart illustrating a method for medical diagnosis based on a deep convolutional neural network according to an embodiment of the present invention. On the basis of the above embodiments, the present embodiment explains the construction of a deep convolutional neural network model.
As shown in fig. 5, the method for medical diagnosis based on a deep convolutional neural network provided in this embodiment includes:
s501, obtaining word vector matrixes corresponding to a plurality of diagnosed electronic medical records, and taking the word vector matrixes corresponding to the diagnosed electronic medical records as training samples.
For example, the diagnosed electronic medical records may be electronic medical records that extract major conditions of each department over the past few years from a medical information management system of a hospital. After acquiring a large amount of diagnosed electronic medical records, the diagnosed electronic medical records are unstructured, information such as personal sensitive information is deleted, pure text content of the electronic medical records is obtained, information filtering is performed, and medical vocabularies corresponding to the diagnosed electronic medical records are acquired. After obtaining each medical vocabulary of the diagnosed medical record, based on the corresponding relation between the medical vocabulary and the word vector
And finally, generating a word vector matrix corresponding to the diagnosed electronic medical record according to the word vector corresponding to the medical vocabulary of the diagnosed medical record.
Specifically, for the deep convolutional neural network model in this embodiment, the diagnosed electronic medical record can be regarded as a training sample. In order to train a deep convolutional neural network model for predicting a plurality of disease types, the more training samples in the embodiment, the better.
S502, training the training samples and constructing a deep convolution neural network model.
Specifically, the deep convolutional neural network model in the present embodiment is a model created based on a deep neural network structure.
For example, a deep convolutional neural network structure such as that shown in fig. 7 is defined, divided into an input layer, a convolutional layer, a pooling layer, and a fully-connected layer.
Wherein, the input layer is a text matrix M. For example, individual training samples are input into the input layer.
The number of convolutional layers and the number of convolutional kernels in each layer are determined according to actual conditions, the width of each convolutional kernel is the same as the width of the input matrix of the layer, and is set to be F, and the height is determined according to actual conditions. If the height of the convolution kernel is h, the convolution kernel is expressed as w epsilon RhFR represents a real number, then the convolution kernel is at Xi:jTo the extracted feature ciComprises the following steps:
ci＝f(w*Xi:j+bi)
wherein f is a non-linear function, biIs an offset. And the convolution kernel slides on the text matrix from top to bottom to obtain the layer of feature spectrum c:
c＝[c1,c1,…,cL-h+1]
it should be noted that the above process describes a process in which one convolution kernel generates one feature, the actual model has multiple convolution layers in calculation, each layer has multiple convolution kernels, and the generation process of each feature is as above.
Wherein the pooling layer is used to reduce the eigenvectors output by the convolutional layer by pooling while improving the result (overfitting is not easy to occur). In this embodiment, pooling may be a maximum poolPooling may also be average pooling, or maximum pooling
Wherein the feature vectors generated by the fully-connected layer are represented as
The output is then:
y＝w*z+bo
the dimension of the output vector y is N, and N is the number of disease types to be predicted. w is the weight to be learned, boIs an offset.
The method for medical diagnosis based on the deep convolutional neural network obtains word vector matrixes corresponding to a plurality of diagnosed electronic medical records, trains the training samples by using the word vector matrixes corresponding to the diagnosed electronic medical records as the training samples, and constructs the deep convolutional neural network model. According to the method, mass diagnosed electronic medical records are obtained and used as historical data, the constructed deep convolutional neural network model can be used for predicting the effect of multiple disease types, the method is very suitable for management and maintenance, and the expandability is very strong; the method does not need to artificially design rules and characteristics, all the characteristics and the rules learned by the model come from a large amount of clinical historical data, clinical historical data are used for guiding clinical decision, and compared with a rule-based auxiliary diagnosis method, the method has very strong practical guiding significance.
Fig. 6 is a flowchart illustrating a method for medical diagnosis based on a deep convolutional neural network according to another embodiment of the present invention. On the basis of the above embodiments, the present embodiment optimizes and explains the deep convolutional neural network model and the classifier.
As shown in fig. 6, the method for medical diagnosis based on a deep convolutional neural network provided in this embodiment includes:
s601, obtaining word vector matrixes corresponding to a plurality of diagnosed electronic medical records, and taking the word vector matrixes corresponding to the diagnosed electronic medical records as training samples.
And S602, training the training samples to construct a deep convolutional neural network model.
The implementation manners of steps S601 and S602 in this embodiment are the same as the implementation manners of S501 and S502 in the above embodiment, and are not described herein again.
And S603, acquiring the diagnosis result of the doctor corresponding to each diagnosed electronic medical record.
In this embodiment, the diagnosis result of the diagnosed electronic medical record corresponding to the doctor can be understood as the medical diagnosis made by the doctor according to various medical indexes in the electronic medical record.
S604, acquiring a feature vector corresponding to the diagnosed electronic medical record output by the deep convolutional neural network model aiming at each diagnosed electronic medical record.
In this embodiment, the deep convolutional neural network model is obtained by training a large number of diagnosed electronic medical records as training samples, each training sample is input into the deep convolutional neural network model, and the deep convolutional neural network model outputs a corresponding training result.
S605, classifying the feature vectors of the diagnosed electronic medical record by using a classifier, and acquiring the disease probability of each disease of the diagnosed electronic medical record.
Specifically, taking the classifier as a softmax classifier as an example, referring to fig. 7, the full connection layer of the deep convolutional neural network model may be connected to the softmax classifier, and the softmax classifier classifies the feature vectors of the deep convolutional neural network model to obtain the disease probability of each disease:
wherein p isiIndicating the disease probability of the ith disease of the diagnosed electronic medical record, yiThe ith element for representing y is a feature vector corresponding to the ith disease in the electronic medical record, yjAnd j element representing y, namely a characteristic vector corresponding to j disease in the electronic medical record.
S606, analyzing the disease probability of each disease of the diagnosed electronic medical record and the doctor diagnosis result corresponding to the diagnosed electronic medical record by using a preset algorithm, and correcting the parameters of the deep convolutional neural network model and the parameters of the classifier according to the analysis result.
Specifically, in the training process, the parameters of the deep convolutional neural network model and the parameters of the classifier are updated through a preset algorithm such as a Back Propagation (BP) algorithm.
For example, the objective function of the output of the entire network is represented as:
it should be noted that the whole network can be understood as a network composed of a deep convolutional neural network model and a classifier. Where P denotes the output of the classifier, each PiThe element represents the probability of the disease being the ith disease. T is the true value, namely T is the diagnosis result of the doctor of the electronic medical record, namely if the electronic medical record corresponds to the tth disease, the tth element value in the T vector is 1, and the rest values are 0. num represents the number of samples per training. w represents the parameters of the deep convolutional neural network model. The training process minimizes the LOSS function through a BP (Back Propagation) algorithm until the network converges and the LOSS does not drop any more, at which time the training is completed and all parameters in the whole network are reserved.
It is noted that the T-vector is generated based on the probability of illness for each condition of the diagnosed electronic medical record.
For example, in the test phase, all the parameters in the whole network which are reserved are continuously read to update the parameters of the deep convolutional neural network model and the parameters of the classifier. Inputting the electronic medical record into the deep convolutional neural network model, and obtaining an output result corresponding to the electronic medical record by using a classifier as a P vector, wherein the P vector is a disease prediction result of the electronic medical record, and if the jth element value in the P vector is the maximum, the most possible disease of the patient is predicted to be the jth disease.
The method for medical diagnosis based on the deep convolutional neural network provided by the embodiment is characterized in that a classifier is used for classifying feature vectors of a diagnosed electronic medical record to obtain the disease probability of each disease of the diagnosed electronic medical record, a reverse iterative algorithm is used for analyzing the disease probability of each disease of the diagnosed electronic medical record and a doctor diagnosis result corresponding to the diagnosed electronic medical record, and parameters of a deep convolutional neural network model and parameters of the classifier are corrected according to the analysis result, so that the constructed deep convolutional neural network model can be used for predicting the effect of a plurality of disease types, and is very suitable for management and maintenance and very strong in expandability; the method does not need to artificially design rules and characteristics, all the characteristics and the rules learned by the model come from a large amount of clinical historical data, clinical historical data are used for guiding clinical decision, and compared with a rule-based auxiliary diagnosis method, the method has very strong practical guiding significance.
In the description herein, references to the description of the term "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., mean that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above are not necessarily intended to refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, various embodiments or examples and features of different embodiments or examples described in this specification can be combined and combined by one skilled in the art without contradiction.
Furthermore, the terms "first", "second" and "first" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present invention, "a plurality" means at least two, e.g., two, three, etc., unless specifically limited otherwise.
Any process or method descriptions in flow charts or otherwise described herein may be understood as representing modules, segments, or portions of code which include one or more executable instructions for implementing steps of a custom logic function or process, and alternate implementations are included within the scope of the preferred embodiment of the present invention in which functions may be executed out of order from that shown or discussed, including substantially concurrently or in reverse order, depending on the functionality involved, as would be understood by those reasonably skilled in the art of the present invention.
The logic and/or steps represented in the flowcharts or otherwise described herein, e.g., an ordered listing of executable instructions that can be considered to implement logical functions, can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For the purposes of this description, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of the computer-readable medium would include the following: an electrical connection (electronic device) having one or more wires, a portable computer diskette (magnetic device), a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium could even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, via for instance optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner if necessary, and then stored in a computer memory.
It should be understood that portions of the present invention may be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, the various steps or methods may be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. If implemented in hardware, as in another embodiment, any one or combination of the following techniques, which are known in the art, may be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application specific integrated circuit having an appropriate combinational logic gate circuit, a Programmable Gate Array (PGA), a Field Programmable Gate Array (FPGA), or the like.
It will be understood by those skilled in the art that all or part of the steps carried by the method for implementing the above embodiments may be implemented by hardware that is related to instructions of a program, and the program may be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.
In addition, functional units in the embodiments of the present invention may be integrated into one processing module, or each unit may exist alone physically, or two or more units are integrated into one module. The integrated module can be realized in a hardware mode, and can also be realized in a software functional module mode. The integrated module, if implemented in the form of a software functional module and sold or used as a separate product, may also be stored in a computer readable storage medium.
The storage medium mentioned above may be a read-only memory, a magnetic or optical disk, etc. Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention, and that variations, modifications, substitutions and alterations can be made to the above embodiments by those of ordinary skill in the art within the scope of the present invention.
Claims (5)
1. A method for medical diagnosis based on a deep convolutional neural network, comprising:
acquiring a word vector matrix corresponding to the electronic medical record to be diagnosed;
inputting the word vector matrix corresponding to the electronic medical record to be diagnosed into a pre-constructed deep convolutional neural network model to obtain a characteristic vector corresponding to the electronic medical record to be diagnosed;
classifying the feature vectors of the electronic medical record to be diagnosed by using a classifier to obtain the disease probability of each disease corresponding to the electronic medical record to be diagnosed;
the acquiring of the word vector matrix corresponding to the electronic medical record to be diagnosed includes:
performing at least one operation of information filtering, screening, word segmentation and statistics on the electronic medical record to be diagnosed to obtain each medical vocabulary of the medical record to be diagnosed;
acquiring a word vector corresponding to a medical vocabulary of the medical record to be diagnosed in a preset word vector database, wherein the preset word vector database stores the corresponding relation between the medical vocabulary and the word vector;
generating a word vector matrix corresponding to the electronic disease to be diagnosed according to the word vector corresponding to the medical vocabulary of each electronic medical record to be diagnosed;
before the obtaining of the word vector matrix corresponding to the electronic medical record to be diagnosed, the method includes:
acquiring each medical vocabulary in a medical word bank;
inputting the medical vocabulary in the medical lexicon into a Word2Vec model established in advance, and acquiring Word vectors corresponding to the medical vocabulary;
forming word vector samples by word vectors corresponding to the medical vocabularies, and storing the word vector samples in a preset word vector database;
before the obtaining of the word vector matrix corresponding to the electronic medical record to be diagnosed, the method includes:
acquiring word vector matrixes corresponding to a plurality of diagnosed electronic medical records, and taking the word vector matrixes corresponding to the diagnosed electronic medical records as training samples;
training the training sample to construct the deep convolutional neural network model;
after the constructing the deep convolutional neural network model, the method comprises the following steps:
acquiring a doctor diagnosis result corresponding to each diagnosed electronic medical record;
acquiring a feature vector corresponding to the diagnosed electronic medical record output by the deep convolutional neural network model aiming at each diagnosed electronic medical record;
classifying the feature vectors of the diagnosed electronic medical record by using the classifier to obtain the illness probability of each illness of the diagnosed electronic medical record;
analyzing the disease probability of each disease of the diagnosed electronic medical record and a doctor diagnosis result corresponding to the diagnosed electronic medical record by using a preset algorithm, and correcting the parameters of the deep convolutional neural network model and the parameters of the classifier according to the analysis result;
before the obtaining of each medical vocabulary in the medical lexicon, the method comprises the following steps:
acquiring a plurality of diagnosed electronic medical records;
information filtering is carried out on each diagnosed electronic medical record by utilizing an information filtering technology, and a medical vocabulary set is obtained; and counting the word frequency of each medical word in the medical word set, screening each medical word according to a set screening rule, and establishing the medical word bank according to a screening result.
2. The method of claim 1, wherein the pre-set algorithm is a back-propagation algorithm.
3. The method of any one of claims 1 to 2, wherein the classifier is a softmax classifier.
4. The method of any one of claims 1 to 2, wherein the deep convolutional neural network model comprises: input layer, convolution layer, pooling layer, full connection layer.
5. The method of claim 4, wherein the convolutional layer comprises a plurality of convolutional kernels of different sizes.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711002335.7A CN107863147B (en) | 2017-10-24 | 2017-10-24 | Medical diagnosis method based on deep convolutional neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711002335.7A CN107863147B (en) | 2017-10-24 | 2017-10-24 | Medical diagnosis method based on deep convolutional neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107863147A CN107863147A (en) | 2018-03-30 |
| CN107863147B true CN107863147B (en) | 2021-03-16 |
Family
ID=61696344
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711002335.7A Active CN107863147B (en) | 2017-10-24 | 2017-10-24 | Medical diagnosis method based on deep convolutional neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107863147B (en) |
Families Citing this family (40)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| CN108877929A (en) * | 2018-05-31 | 2018-11-23 | 平安医疗科技有限公司 | It is neural network based to intend examining recommendation process method, apparatus and computer equipment |
| CN108766561B (en) * | 2018-05-31 | 2023-12-08 | 平安医疗科技有限公司 | Method, apparatus, computer device and storage medium for processing disease information |
| CN109003677B (en) * | 2018-06-11 | 2021-11-05 | 清华大学 | Structured analysis processing method for medical record data |
| CN109192299A (en) * | 2018-08-13 | 2019-01-11 | 中国科学院计算技术研究所 | A kind of medical analysis auxiliary system based on convolutional neural networks |
| CN110929511B (en) * | 2018-09-04 | 2021-12-17 | 清华大学 | Intelligent matching method for personalized traditional Chinese medicine diagnosis and treatment information and traditional Chinese medicine information based on semantic similarity |
| CN109243618B (en) * | 2018-09-12 | 2020-06-16 | 腾讯科技（深圳）有限公司 | Medical model construction method, disease label construction method and intelligent device |
| CN109390058A (en) * | 2018-09-28 | 2019-02-26 | 湖南智腾安控科技有限公司 | A kind of method for building up of case history Computer Aided Analysis System and the system |
| CN109411062A (en) * | 2018-09-30 | 2019-03-01 | 天津科技大学 | A method of diabetes and its complication are judged automatically using physiological parameter based on depth learning technology |
| CN109493931B (en) * | 2018-10-25 | 2024-06-04 | 平安科技（深圳）有限公司 | Medical record file encoding method, server and computer readable storage medium |
| CN109599185B (en) * | 2018-11-14 | 2021-05-25 | 金色熊猫有限公司 | Disease data processing method and device, electronic equipment and computer readable medium |
| CN111199801B (en) * | 2018-11-19 | 2023-08-08 | 零氪医疗智能科技(广州)有限公司 | Construction method and application of model for identifying disease types of medical records |
| CN109698016A (en) * | 2018-12-11 | 2019-04-30 | 中国科学院深圳先进技术研究院 | Disease automatic coding and device |
| CN109801706B (en) * | 2018-12-12 | 2020-11-10 | 清华大学 | Psychological stress problem sensing method and device |
| CN109378066A (en) * | 2018-12-20 | 2019-02-22 | 翼健(上海)信息科技有限公司 | A kind of control method and control device for realizing disease forecasting based on feature vector |
| CN109920501B (en) * | 2019-01-24 | 2021-04-20 | 西安交通大学 | Electronic medical record classification method and system based on convolutional neural network and active learning |
| CN109949927A (en) * | 2019-02-18 | 2019-06-28 | 四川拾智联兴科技有限公司 | A kind of intelligent diagnosing method and its system based on deep neural network |
| CN110021439B (en) * | 2019-03-07 | 2023-01-24 | 平安科技（深圳）有限公司 | Medical data classification method and device based on machine learning and computer equipment |
| CN109919243A (en) * | 2019-03-15 | 2019-06-21 | 天津拾起卖科技有限公司 | A kind of scrap iron and steel type automatic identifying method and device based on CNN |
| CN109949929A (en) * | 2019-03-19 | 2019-06-28 | 挂号网（杭州）科技有限公司 | A kind of assistant diagnosis system based on the extensive case history of deep learning |
| CN110136788B (en) * | 2019-05-14 | 2021-08-17 | 清华大学 | Medical record quality inspection method, device, equipment and storage medium based on automatic detection |
| CN110349668A (en) * | 2019-07-11 | 2019-10-18 | 四川大学 | A kind of therapeutic scheme aid decision-making method and its system based on BP neural network |
| CN112347776B (en) * | 2019-08-09 | 2024-05-24 | 金色熊猫有限公司 | Medical data processing method and device, storage medium and electronic equipment |
| CN110471941B (en) * | 2019-08-12 | 2020-10-02 | 贵州医渡云技术有限公司 | Method and device for automatically positioning judgment basis and electronic equipment |
| CN112446201B (en) * | 2019-08-12 | 2024-08-02 | 北京国双科技有限公司 | Method and device for determining comment properties of text |
| CN110755045B (en) * | 2019-10-30 | 2022-06-07 | 湖南财政经济学院 | Skin disease comprehensive data analysis and diagnosis auxiliary system |
| CN110911009A (en) * | 2019-11-14 | 2020-03-24 | 南京医科大学 | Clinical diagnosis aid decision-making system and medical knowledge map accumulation method |
| CN111081379B (en) * | 2019-12-02 | 2023-05-16 | 中南大学 | Disease probability decision method and system thereof |
| CN111199797A (en) * | 2019-12-31 | 2020-05-26 | 中国中医科学院中医药信息研究所 | Auxiliary diagnosis model establishing and auxiliary diagnosis method and device |
| CN111259116A (en) * | 2020-01-16 | 2020-06-09 | 北京珞安科技有限责任公司 | Sensitive file detection method based on convolutional neural network |
| CN111462895B (en) * | 2020-03-30 | 2024-04-05 | 讯飞医疗科技股份有限公司 | Auxiliary diagnosis method and system |
| CN111599463B (en) * | 2020-05-09 | 2023-07-14 | 吾征智能技术(北京)有限公司 | Intelligent auxiliary diagnosis system based on sound cognition model |
| CN111599462B (en) * | 2020-05-09 | 2023-11-21 | 吾征智能技术(北京)有限公司 | Intelligent body abnormal odor screening system based on cognitive learning |
| CN111667917A (en) * | 2020-06-10 | 2020-09-15 | 北京小白世纪网络科技有限公司 | Method, system and equipment for realizing traditional Chinese medicine diagnosis and treatment based on neural network |
| CN112001066B (en) * | 2020-07-30 | 2022-11-04 | 四川大学 | Deep learning-based method for calculating limit transmission capacity |
| CN112466462B (en) * | 2020-11-26 | 2023-03-07 | 华侨大学 | EMR information association and evolution method based on deep learning of image |
| CN112951432A (en) * | 2021-03-30 | 2021-06-11 | 太原理工大学 | Method and system for predicting heart failure complications based on CNN and electronic medical record, early warning equipment and storage medium |
| CN113658688B (en) * | 2021-08-23 | 2024-02-13 | 上海赴源科技服务有限公司 | Clinical decision support method based on word segmentation-free deep learning |
| CN114203298A (en) * | 2021-12-15 | 2022-03-18 | 山东众阳健康科技集团有限公司 | Auxiliary diagnosis method and system based on deep neural network |
| CN118248328B (en) * | 2024-05-28 | 2024-07-30 | 中国人民解放军空军军医大学 | Disease prediction model training method and disease prediction system |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106446526A (en) * | 2016-08-31 | 2017-02-22 | 北京千安哲信息技术有限公司 | Electronic medical record entity relation extraction method and apparatus |
| CN106778014A (en) * | 2016-12-29 | 2017-05-31 | 浙江大学 | A kind of risk Forecasting Methodology based on Recognition with Recurrent Neural Network |
| CN106874643A (en) * | 2016-12-27 | 2017-06-20 | 中国科学院自动化研究所 | Build the method and system that knowledge base realizes assisting in diagnosis and treatment automatically based on term vector |
| CN107247881A (en) * | 2017-06-20 | 2017-10-13 | 北京大数医达科技有限公司 | A kind of multi-modal intelligent analysis method and system |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9430829B2 (en) * | 2014-01-30 | 2016-08-30 | Case Western Reserve University | Automatic detection of mitosis using handcrafted and convolutional neural network features |
| CN107220506A (en) * | 2017-06-05 | 2017-09-29 | 东华大学 | Breast cancer risk assessment analysis system based on deep convolutional neural network |
-
2017
- 2017-10-24 CN CN201711002335.7A patent/CN107863147B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106446526A (en) * | 2016-08-31 | 2017-02-22 | 北京千安哲信息技术有限公司 | Electronic medical record entity relation extraction method and apparatus |
| CN106874643A (en) * | 2016-12-27 | 2017-06-20 | 中国科学院自动化研究所 | Build the method and system that knowledge base realizes assisting in diagnosis and treatment automatically based on term vector |
| CN106778014A (en) * | 2016-12-29 | 2017-05-31 | 浙江大学 | A kind of risk Forecasting Methodology based on Recognition with Recurrent Neural Network |
| CN107247881A (en) * | 2017-06-20 | 2017-10-13 | 北京大数医达科技有限公司 | A kind of multi-modal intelligent analysis method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107863147A (en) | 2018-03-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107863147B (en) | Medical diagnosis method based on deep convolutional neural network | |
| Sidey-Gibbons et al. | Machine learning in medicine: a practical introduction | |
| Srinivasu et al. | From blackbox to explainable AI in healthcare: existing tools and case studies | |
| RU2703679C2 (en) | Method and system for supporting medical decision making using mathematical models of presenting patients | |
| Pham et al. | Deepcare: A deep dynamic memory model for predictive medicine | |
| CN113421652A (en) | Method for analyzing medical data, method for training model and analyzer | |
| Batch et al. | Developing a cancer digital twin: supervised metastases detection from consecutive structured radiology reports | |
| CN108563626A (en) | Medical text name entity recognition method and device | |
| CN109036577A (en) | Diabetic complication analysis method and device | |
| Liu et al. | Learning hierarchical representations of electronic health records for clinical outcome prediction | |
| Dligach et al. | Toward a clinical text encoder: pretraining for clinical natural language processing with applications to substance misuse | |
| Kaswan et al. | AI-based natural language processing for the generation of meaningful information electronic health record (EHR) data | |
| Jnawali et al. | Automatic classification of radiological report for intracranial hemorrhage | |
| Peng et al. | BiteNet: bidirectional temporal encoder network to predict medical outcomes | |
| Teo et al. | Current trends in readmission prediction: an overview of approaches | |
| Wu et al. | A pseudo label-wise attention network for automatic ICD coding | |
| Fadhil et al. | Multiple efficient data mining algorithms with genetic selection for prediction of SARS-CoV2 | |
| CN113658688A (en) | Clinical decision support method based on word segmentation-free deep learning | |
| Sudharson et al. | Performance analysis of enhanced adaboost framework in multifacet medical dataset | |
| Abut et al. | Deep Neural Networks and Applications in Medical Research | |
| Srithaja et al. | Machine Learning Framework for Stagewise Classification of Alzheimer’s Disease | |
| Theodorou et al. | Synthesize extremely high-dimensional longitudinal electronic health records via hierarchical autoregressive language model | |
| CN114420232A (en) | Method and system for generating health education data based on electronic medical record data | |
| Subashini et al. | Multimodal deep learning for chronic kidney disease prediction: leveraging feature selection algorithms and ensemble models | |
| Kanchinadam et al. | Predicting clinical events via graph neural networks |
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 |