CN112270997A - Coronary heart disease diagnosis method and system based on coronary heart disease diagnosis neural network model - Google Patents
Coronary heart disease diagnosis method and system based on coronary heart disease diagnosis neural network model Download PDFInfo
- Publication number
- CN112270997A CN112270997A CN202011386089.1A CN202011386089A CN112270997A CN 112270997 A CN112270997 A CN 112270997A CN 202011386089 A CN202011386089 A CN 202011386089A CN 112270997 A CN112270997 A CN 112270997A
- Authority
- CN
- China
- Prior art keywords
- heart disease
- coronary heart
- neural network
- network model
- disease diagnosis
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 81
- 238000003745 diagnosis Methods 0.000 title claims abstract description 72
- 208000029078 coronary artery disease Diseases 0.000 title claims abstract description 69
- 238000003062 neural network model Methods 0.000 title claims abstract description 46
- 208000024891 symptom Diseases 0.000 claims abstract description 78
- 208000011580 syndromic disease Diseases 0.000 claims abstract description 44
- 239000000463 material Substances 0.000 claims abstract description 25
- 238000012549 training Methods 0.000 claims description 30
- 239000013598 vector Substances 0.000 claims description 17
- 238000012545 processing Methods 0.000 claims description 13
- 238000004590 computer program Methods 0.000 claims description 10
- 238000012795 verification Methods 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims description 4
- 238000002560 therapeutic procedure Methods 0.000 claims description 4
- 239000000126 substance Substances 0.000 claims description 3
- 238000013507 mapping Methods 0.000 abstract description 19
- 239000003814 drug Substances 0.000 abstract description 18
- 230000008569 process Effects 0.000 description 25
- 238000013528 artificial neural network Methods 0.000 description 8
- 239000008280 blood Substances 0.000 description 6
- 210000004369 blood Anatomy 0.000 description 6
- 210000002569 neuron Anatomy 0.000 description 6
- 206010008479 Chest Pain Diseases 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 244000170916 Paeonia officinalis Species 0.000 description 3
- 235000006484 Paeonia officinalis Nutrition 0.000 description 3
- 208000002193 Pain Diseases 0.000 description 3
- 230000009471 action Effects 0.000 description 3
- 230000008901 benefit Effects 0.000 description 3
- 230000017531 blood circulation Effects 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 3
- 201000010099 disease Diseases 0.000 description 3
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 235000008216 herbs Nutrition 0.000 description 3
- 244000020518 Carthamus tinctorius Species 0.000 description 2
- 235000003255 Carthamus tinctorius Nutrition 0.000 description 2
- 206010033557 Palpitations Diseases 0.000 description 2
- 206010062519 Poor quality sleep Diseases 0.000 description 2
- 240000007164 Salvia officinalis Species 0.000 description 2
- 230000004913 activation Effects 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000003058 natural language processing Methods 0.000 description 2
- 244000138993 panchioli Species 0.000 description 2
- 230000001737 promoting effect Effects 0.000 description 2
- 235000005412 red sage Nutrition 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 239000002023 wood Substances 0.000 description 2
- 241001529821 Agastache Species 0.000 description 1
- 241000045403 Astragalus propinquus Species 0.000 description 1
- 244000037364 Cinnamomum aromaticum Species 0.000 description 1
- 235000014489 Cinnamomum aromaticum Nutrition 0.000 description 1
- 244000223760 Cinnamomum zeylanicum Species 0.000 description 1
- 244000303040 Glycyrrhiza glabra Species 0.000 description 1
- 235000006200 Glycyrrhiza glabra Nutrition 0.000 description 1
- 235000017443 Hedysarum boreale Nutrition 0.000 description 1
- 235000007858 Hedysarum occidentale Nutrition 0.000 description 1
- 241000112528 Ligusticum striatum Species 0.000 description 1
- 241000237502 Ostreidae Species 0.000 description 1
- 244000197580 Poria cocos Species 0.000 description 1
- 235000008599 Poria cocos Nutrition 0.000 description 1
- 241001162994 Rugosus Species 0.000 description 1
- 241001180876 Saposhnikovia Species 0.000 description 1
- 244000273928 Zingiber officinale Species 0.000 description 1
- 235000006886 Zingiber officinale Nutrition 0.000 description 1
- 235000006545 Ziziphus mauritiana Nutrition 0.000 description 1
- 244000126002 Ziziphus vulgaris Species 0.000 description 1
- 235000008529 Ziziphus vulgaris Nutrition 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 208000006673 asthma Diseases 0.000 description 1
- 235000006533 astragalus Nutrition 0.000 description 1
- 238000013475 authorization Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 235000017803 cinnamon Nutrition 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000002405 diagnostic procedure Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 229940079593 drug Drugs 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000008014 freezing Effects 0.000 description 1
- 238000007710 freezing Methods 0.000 description 1
- 235000008397 ginger Nutrition 0.000 description 1
- 239000001947 glycyrrhiza glabra rhizome/root Substances 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000002483 medication Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 235000020636 oyster Nutrition 0.000 description 1
- 230000001717 pathogenic effect Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000011158 quantitative evaluation Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
-
- 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
-
- 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
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/50—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for simulation or modelling of medical disorders
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Public Health (AREA)
- Medical Informatics (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Primary Health Care (AREA)
- Epidemiology (AREA)
- Pathology (AREA)
- Databases & Information Systems (AREA)
- Computational Linguistics (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biophysics (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- General Physics & Mathematics (AREA)
- Medicinal Chemistry (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Chemical & Material Sciences (AREA)
- Medical Treatment And Welfare Office Work (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
The invention discloses a coronary heart disease diagnosis method and a system based on a coronary heart disease diagnosis neural network model, wherein the method comprises the following steps: inputting one or more words regarding the symptom base and other characteristics; inputting the words into a pre-trained neural network model for diagnosing coronary heart disease, and obtaining and outputting the syndrome elements, syndromes, treatment methods and/or basic element information of medicinal materials of the patient. The embodiment is based on an improved Transformer model, which is a model applicable to complex dialectical thinking of traditional Chinese medicine, the model is an end-to-end model, a Transformer module mapping relation in the improved Transformer model can not only solve a linear mapping relation, but also has a self-adaptive module, and the mapping relation just meets the requirements of people to solve the correct medicinal material corresponding relation under the condition that the symptom input comprises various characteristics, so that the whole diagnosis result is more accurate.
Description
Technical Field
The invention relates to a disease diagnosis method, in particular to a coronary heart disease diagnosis method and system based on a coronary heart disease diagnosis neural network model.
Background
With the large-scale growth of natural language data in social aspects and professional fields on the internet, various natural language processing technologies are receiving wide attention and application. Due to the specific diagnosis and treatment mode, the traditional Chinese medicine combines the curative effect reports of patients and caregivers and laboratory inspection reports to obtain more obvious clinical curative effect and gain advices of a plurality of patients. However, since most of the effects of the traditional Chinese medicine treatment lack an objective and quantitative evaluation mode, it becomes difficult to simulate the traditional Chinese medicine diagnosis and treatment and summarize the traditional Chinese medicine experience by using deep learning or other algorithms.
The diagnosis process of TCM is slightly different according to its genres. In order to simulate the traditional Chinese medicine diagnosis thought and make a prescription, the existing natural speech recognition semantic understanding model is utilized and modified to make a prescription, the traditional Chinese medicine diagnosis process is specifically shown in figure 1, the traditional Chinese medicine comprehensively analyzes information collected by looking at, asking and cutting four diagnoses, distinguishes and clarifies the relationship between the etiology, the disease property, the disease position and the pathogenic factors of a patient into syndromes; therefore, determining a reasonable symptom and prescribing a prescription around the symptom becomes a critical step. By repeated discussion and exploration with traditional Chinese medicine, a diagnostic process as in FIG. 1 is contemplated: wherein "1" represents a process of judging the syndrome elements by the series of symptoms of the patient, "2" determines the treatment method by the syndrome elements, "3" prescribes a prescription to the patient according to the treatment method, "4" in a main deduction process ("123"), the patient may have some symptoms, the syndrome elements cannot be covered, and additional treatment is required, thus determining the treatment method. "5" finally, if there is a need for special treatment of some symptoms or signs of the patient, such as the patient having symptoms of asthma, special contraindications are required for some medications, or the patient is prescribed few or no more.
In the existing research of several types of traditional Chinese medicine diagnosis and treatment methods, a reasonable decision system is not available for realizing the process from natural language processing to prescription making of a patient; the traditional method is a decision system based on a decision tree, and the model has a hierarchical decision model structure of the following formula 1:
f(X)＝Pn(...P2(P1(x1)，x2)...，xn) Equation 1
X＝{x1，x2，x3，x4，x5...，xn}; x is input symptom formula 2
The above-described model with a hierarchy level has the following problems:
(1) the practicability problem is that: the patient needs to answer the symptoms layer by layer according to the prompt of the system, the method increases the accuracy to some extent, but as the logic of the system becomes complex, more and more questions need to be answered, and probably 20-30 questions need to be answered to obtain answers, so that the method is cumbersome and troublesome to use.
(2) Input order of symptoms problem: the first and last inputs of a symptom may have completely different output results, and the biggest problem is that the system becomes more and more complex as the symptoms increase.
In view of the above, it is desirable to provide a coronary heart disease diagnosis method that simulates the diagnosis thinking of traditional Chinese medicine and generates an effective prescription.
Disclosure of Invention
In order to solve the technical problems, the invention adopts the technical scheme that a neural network model coronary heart disease diagnosis method based on coronary heart disease diagnosis is provided, and the method comprises the following steps:
s1, inputting one or more words about the basic elements and other characteristics of the symptoms; other characteristics can be the sex, age, medical history and other accessory characteristics input according to the user requirements;
inputting the words into a pre-trained coronary heart disease diagnosis neural network model, and obtaining and outputting syndrome elements, syndromes, treatment methods and/or medicinal material basic element information of a patient;
the coronary heart disease diagnosis neural network model is an improved Transformer model and comprises an Embelling module and three Transformer modules, namely a Transformer1, a Transformer2 and a Transformer 3;
a concat + Linear module is arranged between the two Transformer modules and is concat + Linear1 and concat + Linear2 respectively;
the Embedding module is respectively connected with a first Transformer1, a concat + Linear1 and a concat + Linear2 in output;
the Transformer1 and the Transformer2 are respectively connected with a Linear module.
In the above method, the Transformer module comprises a Multihead attachment layer and a fed forward layer, wherein the Multihead attachment layer and the fed forward layer are arranged in parallel, and the multi-head attachment layer and the fed forward layer are arranged in parallel in the same layer
The MultiheadAttention layer performs the following calculations:
Q＝QWQ，K＝KWK，V＝VWV
head＝Attention(Q，K，V)
Attention(Q，K，V)＝V×Softmax(KWa×QWb)
all W are learnable variables, the initial time is initialized to be a random number which obeys normal distribution, then the random number is driven by data to carry out gradient back propagation, and Q, K and V respectively represent Queue, Key and Value vectors.
In the above method, the Embedding module includes
The system comprises an Embedding module and a segment Embedding module, wherein the Embedding module is used for processing input symptom elements, and the segment Embedding module is used for processing elements except the input symptoms; namely, it is
Embeddingall＝Embeddingseg+Embedding{features}。
In the above method, the Embedding module projects each symptom to a 64-dimensional vector.
The invention also provides a neural network model coronary heart disease diagnosis system based on coronary heart disease diagnosis, which comprises
An input unit: inputting one or more words regarding the symptom base and other characteristics;
a diagnosis unit: inputting the words into a pre-trained coronary heart disease diagnosis neural network model so as to obtain the syndrome elements, syndromes, treatment methods and/or basic element information of medicinal materials of the patient;
an output unit: outputting the syndrome element, syndrome, therapeutic method or basic element information of the medicinal materials of the patient;
wherein the content of the first and second substances,
the diagnosis and treatment neural network model comprises an improved Transformer model which comprises an Embedding module and three Transformer modules, namely Transformer1, Transformer2 and Transformer 3;
a concat + Linear module is arranged between the two Transformer modules and is concat + Linear1 and concat + Linear2 respectively;
the Embedding module is respectively connected with a first Transformer1, a concat + Linear1 and a concat + Linear2 in output;
the Transformer1 and the Transformer2 are respectively connected with a Linear module.
In the above aspect, the diagnostic unit includes a training subunit including
A data input module: for obtaining a historical symptom data set;
an initial diagnostic module: randomly dividing a symptom data set into a training set, a verification set and a test set; and respectively inputting the corresponding training set and the medical record data in the verification set into the initial coronary heart disease diagnosis neural network model for model training, and stopping training when the training conditions are met to obtain the trained coronary heart disease diagnosis neural network model.
In the above scheme, the Transformer module includes a multimedia attribute layer and a fed forward layer, wherein the multimedia attribute layer and the fed forward layer are included in the Transformer module
The Multihead Attention layer performs the following calculations:
Q＝QWQ，K＝KWK，V＝VWV
head＝Attention(Q，K，V)
Attention(Q，K，V)＝V×Softmax(KWa×QWb)
all W are learnable variables, the initial moment is initialized to be random numbers which obey normal distribution, and then the random numbers are driven by data to carry out gradient back propagation.
The invention also provides computer equipment which comprises a memory, a processor and a computer program which is stored in the memory and can run on the processor, wherein the processor executes the computer program to realize the coronary heart disease diagnosis method based on the coronary heart disease diagnosis neural network model.
The invention also provides a computer readable storage medium storing a computer program which, when executed by a processor, implements the method for diagnosing coronary heart disease based on a neural network model for diagnosing coronary heart disease as described in any one of the above.
The improved Transformer-based model is a model which can be applied to complex dialectical thinking of the traditional Chinese medicine, the model is an end-to-end model, characteristics such as symptoms, gender and medical history can be input at one end, and syndrome elements, a treatment method and medicinal materials are output in the middle process; the whole doctor deduction process is followed, and the advantage of using the neural network to carry out traditional Chinese medicine deduction in the embodiment is that repeated learning can be realized, so that the accuracy rate of the diagnosis result of the whole network model is improved; in addition, the Transformer module mapping relation in the improved Transformer model of the embodiment can not only solve the linear mapping relation, but also has a self-adaptive module, and the mapping relation just meets the requirement of people to solve the correct medicinal material corresponding relation under the condition that the symptom input comprises various characteristics, so that the whole diagnosis result is more accurate.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
Fig. 1 is a schematic diagram of a prior art deduction method for diagnosis and treatment of traditional Chinese medicine provided by the present invention;
FIG. 2 is a schematic flow chart of a method provided by the present invention;
FIG. 3 is a schematic block diagram of a neural network model for coronary heart disease diagnosis according to the present invention;
FIG. 4 is a schematic block diagram of a Transformer module structure provided in the present invention;
FIG. 5 is a schematic block diagram of a system provided by the present invention;
fig. 6 is a schematic block diagram of a computer device structure provided by the present invention.
Detailed Description
The technical solutions of the present invention will be described clearly and completely with reference to the following embodiments, and it should be understood that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In the description of the present invention, it is to be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", and the like, indicate orientations and positional relationships based on those shown in the drawings, and are used only for convenience of description and simplicity of description, and do not indicate or imply that the device or element being referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus, should not be considered as limiting the present invention.
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, features defined as "first", "second", may explicitly or implicitly include one or more of the described features. In the description of the present invention, "a plurality" means two or more unless specifically defined otherwise. Furthermore, the terms "mounted," "connected," and "connected" are to be construed broadly and may, for example, be fixedly connected, detachably connected, or integrally connected; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meanings of the above terms in the present invention can be understood in specific cases to those skilled in the art.
The invention is described in detail below with reference to specific embodiments and the accompanying drawings.
As shown in fig. 2, the invention provides a neural network model coronary heart disease diagnosis method based on coronary heart disease diagnosis, comprising the following steps:
s1, inputting one or more words about the basic elements and other characteristics of the symptoms; other characteristics may be gender, age, medical history, etc. as an adjunct to user demand input. In this embodiment, all parameters (basic elements and other features) and mapping relationships are read into a memory or a video memory together, and are continuously updated in the training process, and finally, the trained model can be packed and stored in a centralized manner.
S2, inputting words into a pre-trained neural network model for diagnosing coronary heart disease, and obtaining and outputting syndrome elements, syndromes, treatment methods and/or basic element information of medicinal materials of a patient;
as shown in fig. 3, the neural network model for coronary heart disease diagnosis is an improved Transformer model, which includes an Embedding module and three Transformer modules, namely, Transformer1, Transformer2 and Transformer 3;
a concat + Linear module is arranged between the two Transformer modules and is concat + Linear1 and concat + Linear2 respectively;
the Embedding module is respectively connected with a first Transformer1, a concat + Linear1 and a concat + Linear2 in output;
the Transformer1 and the Transformer2 are respectively connected with a Linear module;
in this embodiment, the primary function of the transform module is to group inputs, and then determine according to the grouped contents and give a result. The Linear module realizes Linear and nonlinear mapping, and the transform module is equivalent to performing logic judgment on input information, for example, the input is as follows: coronary heart disease, palpitation, chest pain, fixed position pain, chest distress, poor sleep, pale red tongue and wiry and rapid pulse, wherein a transform module is used for performing logic grouping judgment, after the judgment is finished, a space vector needs to be mapped to an onehot vector, and the mapping process of the part is realized by using a Linear module.
In this embodiment, the Linear module is a neuron structure capable of performing Linear or nonlinear processing, and each layer of neurons is formed by adding a Linear mapping to an activation function, which is specifically as follows:
y＝sigmoid(wx+b)
a plurality of such neurons are stacked to form an mlp (multi layer per term) module, which is a classifier commonly used for deep learning. This embodiment replaces MLP (i.e., Linear module) with a transform module; and the Transformer module structure is shown in fig. 4, and includes a Multihead attribute layer and a fed forward layer;
in this embodiment, the multiheadAttention layer specifically completes the following calculations:
Q＝QWQ，K＝KWK，V＝VWV
head＝Attention(Q，K，V)
Attention(Q，K，V)＝V×Softmax(KWa×QWb)
all W are learnable variables, the initial moment is initialized to be random numbers which obey normal distribution, and then the random numbers are driven by data to carry out gradient back propagation.
As shown in the model structure of fig. 4, in this embodiment, the Embedding module projects each symptom to a 64-dimensional vector, that is, each symptom can be represented by a 64-dimensional vector, or each symptom can be described by 64 features, which is advantageous in that if the number of features is set too large or too small, the accuracy of the experimental result is affected, which also indicates that only 64 are enough for the current effective symptom dimension.
The model of this embodiment trains the network by adopting a single-input and multiple-output (symptom input, syndrome element, treatment method, and medicinal material output) strategy, but incomplete input may occur in real data, and under the condition that a real medical plan is difficult to think in the training process, the data of symptom-syndrome element-treatment method-medicinal material are complete, and more cases are the corresponding relationship of symptom-medicinal material or symptom-syndrome element. In the training process, if only symptoms are taken as input and syndrome elements are taken as output, the training process only updates the parameters of the network from the symptoms to the syndrome elements, and if the input is the symptoms and the output is medicinal materials, the whole network needs to be updated from the symptoms to the syndrome elements to the treatment method to the medicinal materials. For example, some have symptoms-syndrome elements, and some have symptoms-herbs. In the training process, if only the syndrome elements are output, the network behind the syndrome elements is frozen (all parameters in the network can be updated through gradients, and the freezing of the part of the network refers to parameters which are not updated, so that the processing mode can lead the part, which is not updated, of the network to be deduced according to the previous parameters, and the updated part of the network is subjected to gradient retransmission according to each pair of data, so that the learning purpose is achieved, and the parameters are not updated).
This embodiment considers that in a real scenario, it is likely that the patient's description of the symptoms is not in any order. However, the input of the neural network is sequential in nature, and in order to solve the problem, the embodiment fixes the parameters in one step, that is, puts 600 symptoms on a 600-dimensional vector, and each dimension represents a fixed symptom.
As shown in fig. 2, the model input is 1000-dimensional vector for representing 600 different symptoms, the rest 400 is reserved as a backup for adding additional symptoms, the output of the Embedding module and each layer of transform module is 64-dimensional vector, the concat in concat + line is to output the 64-dimensional vector output by line for syndrome element output of 80-dimensional, the therapeutic method is 80-dimensional, and the medicinal material is output of 1000-dimensional. Here, the 80 dimensions, equivalent to 80-bit placeholders, each location represents a possibility of an output, which needs to correspond to the actual output. 64-dimensional characteristics are needed to carry out information transmission among the linear modules, each dimension of the 64 dimensions is a free parameter, and the meaning of each dimension is determined through the gradient back propagation process of an actual task.
In this embodiment, preferably, the spatial expansion of the input of the chinese medical science may refer to a method of processing the input by using a BERT model, that is, an Embedding module in the model includes an Embedding module and a segment Embedding module, the Embedding module is used for processing an input symptom element, and the segment Embedding module is used for processing an element other than the input symptom; namely, it is
Embeddingall＝Embeddingseg+Embedding{features}
For example, the user inputs not only symptoms to determine the last herb, but also characteristics such as sex and medical history, but they cannot be handled as symptoms and should be treated differently. Therefore, different feature types need to be projected to different spaces, and then the Embedding operation is performed.
The idea and advantages of this embodiment for diagnosing coronary heart disease using the Transformer model will be described in detail below.
The embodiment understands the improved transform model as a special Linear that can extract the relationship between symptoms, and the Linear structure is a simple Linear mapping. The following is an analysis of the real case:
case 1.
Symptom input: coronary heart disease, palpitation, chest pain, painful immobilization, chest distress, poor sleep, pale red tongue, wiry and rapid pulse;
the syndrome elements are: blood stasis, qi stagnation, qi deficiency;
the treatment method comprises the following steps: promoting blood circulation, removing blood stasis, dredging collaterals, and relieving pain;
the medicinal materials are as follows: red sage root, red peony root, safflower, ligusticum wallichii, dalbergia wood, cassia twig, oyster, keel, white paeony root, Chinese date, ginger, cortex moutan, poria cocos, rhizoma alismatis, radix angelicae, divaricate saposhnikovia root, astragalus mongholicus, rhizoma gastrodiae, agastache rugosus and honey-fried licorice root;
the patient has obvious blood stasis symptoms, such as fixed position pain and chest pain, so the two symptoms possibly indicate that the patient has blood stasis symptoms according to syndrome differentiation, the corresponding treatment method basically takes blood circulation promotion and blood stasis removal as the main part, and the common medicine for promoting blood circulation and removing blood stasis is the prescription No. 2 of the coronary heart disease, which specifically comprises the following steps: red sage root, red peony root, safflower, dalbergia wood and cinnamon twig. Other medicinal materials are used for solving other syndrome factors and symptom conditions of patients. The corresponding relationship between the symptoms and the herbs should be reflected after the main herbs are determined. This relationship is already illustrated in fig. 1. The thinking scene in the diagnosis process of the doctor can be restored through the lower model.
FThinking model(X)＝P3(P2(P1(X)，P4(X))，P5(X))
The present embodiment can completely use Linear mapping (Linear) to represent all mapping relationships P1～5But for P1～3In other words, the relationship is complicated and cannot be completely solved by Linear. There may also be a case where linear mapping cannot be completely solved, that is, there is a relationship between symptoms. For example, symptom a causes one syndrome element, but if symptom B occurs, the syndrome element is completely changed. For this relationship, Linear cannot be perfectly solved. Therefore, a new mapping relationship is needed to solve the problem, the relationship of 'mutual dependence and joint determination' can be solved by using the Transformer module provided by the embodiment, the mapping relationship of the Transformer can not only solve the linear mapping relationship, but also have a self-adaptive module, and the mapping relationship just meets the requirements of people to solve the correct medicinal material corresponding relationship under the condition that the symptom input comprises various characteristics, so that the whole diagnosis result is more accurate.
In this embodiment, the neural network model for coronary heart disease diagnosis is trained by the following method:
randomly dividing a symptom data set into a training set, a verification set and a test set; and respectively inputting the corresponding training set and the medical record data in the verification set into the initial coronary heart disease diagnosis neural network model for model training, and stopping training when the training conditions are met to obtain the trained coronary heart disease diagnosis neural network model.
In the training process, the input of the data can be that the symptom corresponds to the syndrome element and corresponds to the medicinal material; the symptoms and syndrome factors correspond to the herbs. However, in the prediction process, the model only needs to judge symptoms and give predicted syndrome elements and medicinal materials. Different input conditions are designed to meet the requirements of training networks with different data forms.
The improved Transformer-based model is a model which can be applied to complex dialectical thinking of the traditional Chinese medicine, the model is an end-to-end model, characteristics such as symptoms, gender and medical history can be input at one end, and syndrome elements, a treatment method and traditional Chinese medicinal materials are output in the middle process; the method follows the whole doctor deduction process, and the advantage of using the neural network to carry out traditional Chinese medicine deduction in the embodiment is that repeated learning can be carried out, wherein the repeated learning means that the neural network in the data can achieve 90% of effect, if new data exist, the new data can be continuously input into the neural network, and the previous neural network is taken as a network which is trained continuously. Therefore, the neural network can learn further on the basis of the previous network, like the learning process of a person, if new knowledge conflicts with old knowledge, the new knowledge is taken as the standard, meanwhile, much previous knowledge is not forgotten, the accuracy of the diagnosis result of the whole network model is improved, and the neural network is also superior to other algorithm structures.
As shown in FIG. 5, the present invention also provides a neural network model coronary heart disease diagnosis system based on coronary heart disease diagnosis, which comprises
An input unit: inputting one or more words regarding the symptom base and other characteristics;
a diagnosis unit: inputting the words into a pre-trained coronary heart disease diagnosis neural network model so as to obtain the syndrome elements, syndromes, treatment methods and/or basic element information of medicinal materials of the patient;
an output unit: outputting the syndrome element, syndrome, therapeutic method or basic element information of the medicinal materials of the patient;
wherein the content of the first and second substances,
the diagnosis and treatment neural network model comprises an improved Transformer model which comprises an Embedding module and three Transformer modules, namely Transformer1, Transformer2 and Transformer 3;
a concat + Linear module is arranged between the two Transformer modules and is concat + Linear1 and concat + Linear2 respectively;
the Embedding module is respectively connected with a first Transformer1, a concat + Linear1 and a concat + Linear2 in output;
the Transformer1 and the Transformer2 are respectively connected with a Linear module.
In this embodiment, the Linear module is a neuron structure capable of performing Linear or nonlinear processing, and each layer of neurons is formed by adding a Linear mapping to an activation function, which is specifically as follows:
y＝sigmoid(wx+b)
a plurality of such neurons are stacked to form an mlp (multi layer per term) module, which is a classifier commonly used for deep learning. This embodiment replaces MLP (i.e., Linear module) with a transform module; the Transformer module comprises a Multihead authorization layer and a fed forward layer;
in this embodiment, the multimedia attribute layer specifically completes the following calculations:
Q＝input×WQ，K＝input×WK，V＝input×WV
head＝Attention(Q，K，V)
Attention(Q，K，V)＝V×Softmax(KWa×QWb)
wherein, all W are learnable variables, the initial time is initialized to a random number which obeys normal distribution, then the random number is driven by data to carry out gradient back propagation, Q, K and V respectively represent the vector of Queue, Key and Value mentioned in the paper, WaAnd WbThe reason is to distinguish that the parameters passed by the K vector are different from the parameters passed by the Q vector.
In this embodiment, the Embedding module projects 600 symptoms to a 60-dimensional vector, that is, each symptom can be represented by a 64-dimensional vector, or each symptom can be described by 64 features, and this setting is advantageous in that if the feature setting is too large or too small, the accuracy of the experimental result is affected, which also indicates that only 64 are enough for the current effective symptom dimension.
In this embodiment, preferably, the spatial expansion of the input of the chinese medical science may refer to a method of processing the input by using a BERT model, that is, an Embedding module in the model includes an Embedding module and a segment Embedding module, the Embedding module is used for processing an input symptom element, and the segment Embedding module is used for processing an element other than the input symptom; namely, it is
Embeddingall＝Embeddingseg+Embedding{features}
In this embodiment, the diagnosis unit further includes a model training subunit, configured to train an initial coronary heart disease diagnosis neural network model to obtain a coronary heart disease diagnosis neural network model; comprises that
A data input module: for obtaining a historical symptom data set;
an initial diagnostic module: randomly dividing a symptom data set into a training set, a verification set and a test set; and respectively inputting the corresponding training set and the medical record data in the verification set into the initial coronary heart disease diagnosis neural network model for model training, and stopping training when the training conditions are met to obtain the trained coronary heart disease diagnosis neural network model.
As shown in fig. 6, the present invention further provides a computer-readable storage medium, on which a computer program is stored, and the computer program, when executed by a processor, implements the method for diagnosing coronary heart disease based on the neural network model for diagnosing coronary heart disease in the above-mentioned embodiments, or the computer program, when executed by the processor, implements the method for diagnosing coronary heart disease based on the neural network model for diagnosing coronary heart disease in the above-mentioned embodiments.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus Direct RAM (RDRAM), direct bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for apparatus or system embodiments, since they are substantially similar to method embodiments, they are described in relative terms, as long as they are described in partial descriptions of method embodiments. The above-described embodiments of the apparatus and system are merely illustrative, and the units described as separate parts may or may not be physically separate, and the parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
It is noted that, in this document, relational terms such as "first" and "second," and the like, may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
The foregoing are merely exemplary embodiments of the present invention, which enable those skilled in the art to understand or practice the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (9)
1. The coronary heart disease diagnosis method based on the neural network model for coronary heart disease diagnosis is characterized by comprising the following steps of:
s1, inputting one or more words about the basic elements and other characteristics of the symptoms; other characteristics can be the sex, age, medical history and other accessory characteristics input according to the user requirements;
inputting the words into a pre-trained coronary heart disease diagnosis neural network model, and obtaining and outputting syndrome elements, syndromes, treatment methods and/or medicinal material basic element information of a patient;
the coronary heart disease diagnosis neural network model is an improved Transformer model and comprises an Embelling module and three Transformer modules, namely a Transformer1, a Transformer2 and a Transformer 3;
a concat + Linear module is arranged between the two Transformer modules and is concat + Linear1 and concat + Linear2 respectively;
the Embedding module is respectively connected with a first Transformer1, a concat + Linear1 and a concat + Linear2 in output;
the Transformer1 and the Transformer2 are respectively connected with a Linear module.
2. The neural network model coronary heart disease diagnosis method based on coronary heart disease diagnosis of claim 1, wherein the Transformer module comprises a Multihead attachment layer and a fed forward layer, wherein
The Multihead Attention layer performs the following calculations:
0＝QWQ，K＝KWK，V＝VWV
head＝Attention(Q，K，V)
Attention(Q，K，V)＝VXSoftmax(KWaXQWb)
all W are learnable variables, the initial time is initialized to be a random number which obeys normal distribution, then the random number is driven by data to carry out gradient back propagation, and Q, K and V respectively represent Queue, Key and Value vectors.
3. The neural network model coronary heart disease diagnosing method based on coronary heart disease of claim 1, wherein the Embedding module includes
The system comprises an Embedding module and a segment Embedding module, wherein the Embedding module is used for processing input symptom elements, and the segment Embedding module is used for processing elements except the input symptoms; namely, it is
Embeddingall＝Embeddingseg+Embedding{features}。
4. The neural network model coronary heart disease diagnosis method based on coronary heart disease diagnosis according to any one of claims 1-3, wherein the Embedding module projects each symptom to a 64-dimensional vector respectively.
5. The neural network model coronary heart disease diagnosis system based on coronary heart disease diagnosis is characterized by comprising
An input unit: inputting one or more words regarding the symptom base and other characteristics;
a diagnosis unit: inputting the words into a pre-trained coronary heart disease diagnosis neural network model so as to obtain the syndrome elements, syndromes, treatment methods and/or basic element information of medicinal materials of the patient;
an output unit: outputting the syndrome element, syndrome, therapeutic method or basic element information of the medicinal materials of the patient;
wherein the content of the first and second substances,
the diagnosis and treatment neural network model comprises an improved Transformer model which comprises an Embedding module and three Transformer modules, namely Transformer1, Transformer2 and Transformer 3;
a concat + Linear module is arranged between the two Transformer modules and is concat + Linear1 and concat + Linear2 respectively;
the Embedding module is respectively connected with a first Transformer1, a concat + Linear1 and a concat + Linear2 in output;
the Transformer1 and the Transformer2 are respectively connected with a Linear module.
6. The neural network model coronary heart disease diagnostic system based on coronary heart disease diagnosis according to claim 5, wherein the diagnostic unit comprises a training subunit including
A data input module: for obtaining a historical symptom data set;
an initial diagnostic module: randomly dividing a symptom data set into a training set, a verification set and a test set; and respectively inputting the corresponding training set and the medical record data in the verification set into the initial coronary heart disease diagnosis neural network model for model training, and stopping training when the training conditions are met to obtain the trained coronary heart disease diagnosis neural network model.
7. The neural network model coronary heart disease diagnosis system based on coronary heart disease diagnosis of claim 6, wherein the Transformer module comprises a Multihead attachment layer and a fed forward layer, wherein
The Multihead Attention layer performs the following calculations:
Q＝QWQ，K＝KWK，V＝VWV
head＝Attention(Q，K，V)
Attention(Q，K，V)＝V×Softmax(KWa×QWb)
all W are learnable variables, the initial moment is initialized to be random numbers which obey normal distribution, and then the random numbers are driven by data to carry out gradient back propagation.
8. Computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the method for diagnosing coronary heart disease based on a neural network model for diagnosing coronary heart disease as claimed in any one of claims 1 to 5 when executing the computer program.
9. Computer readable storage medium, in which a computer program is stored, which, when being executed by a processor, implements the method according to any one of claims 1 to 5 for diagnosing coronary heart disease based on a neural network model for diagnosing coronary heart disease.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2020109086234 | 2020-09-02 | ||
| CN202010908623.4A CN112053785A (en) | 2020-09-02 | 2020-09-02 | Coronary heart disease diagnosis method and system based on coronary heart disease diagnosis neural network model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112270997A true CN112270997A (en) | 2021-01-26 |
Family
ID=73607770
Family Applications (2)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010908623.4A Pending CN112053785A (en) | 2020-09-02 | 2020-09-02 | Coronary heart disease diagnosis method and system based on coronary heart disease diagnosis neural network model |
| CN202011386089.1A Pending CN112270997A (en) | 2020-09-02 | 2020-12-01 | Coronary heart disease diagnosis method and system based on coronary heart disease diagnosis neural network model |
Family Applications Before (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010908623.4A Pending CN112053785A (en) | 2020-09-02 | 2020-09-02 | Coronary heart disease diagnosis method and system based on coronary heart disease diagnosis neural network model |
Country Status (1)
| Country | Link |
|---|---|
| CN (2) | CN112053785A (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1420488A (en) * | 2001-08-07 | 2003-05-28 | 陈涛 | Vedio tape picture and text data generating and coding method and picture and text data playback device |
| US20050048527A1 (en) * | 2001-11-15 | 2005-03-03 | Third Wave Technologies, Inc. | Endonuclease-substrate complexes |
| CN102580242A (en) * | 2012-04-06 | 2012-07-18 | 刘昭利 | Pulse type headache physical therapy device |
| US20130296748A1 (en) * | 2012-05-03 | 2013-11-07 | Sitkovetskiy Lev N. | Method of massage for improving of body condition, diagnosis, prevention and treatment diseases, and the massage device for these purposes. |
| CN105793280A (en) * | 2013-11-04 | 2016-07-20 | 伊玛提克斯生物技术有限公司 | Personalized immunotherapy against several neuronal and brain tumors |
| US20180137941A1 (en) * | 2015-06-02 | 2018-05-17 | Infervision Co., Ltd. | Method For Analysing Medical Treatment Data Based On Deep Learning and Intelligence Analyser Thereof |
| CN110348019A (en) * | 2019-07-17 | 2019-10-18 | 南通大学 | A kind of medical bodies vector method for transformation based on attention mechanism |
| CN111028951A (en) * | 2019-11-13 | 2020-04-17 | 上海中医药大学 | Method and equipment for Chinese medicine diagnosis retrieval display and evaluation |
-
2020
- 2020-09-02 CN CN202010908623.4A patent/CN112053785A/en active Pending
- 2020-12-01 CN CN202011386089.1A patent/CN112270997A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1420488A (en) * | 2001-08-07 | 2003-05-28 | 陈涛 | Vedio tape picture and text data generating and coding method and picture and text data playback device |
| US20050048527A1 (en) * | 2001-11-15 | 2005-03-03 | Third Wave Technologies, Inc. | Endonuclease-substrate complexes |
| CN102580242A (en) * | 2012-04-06 | 2012-07-18 | 刘昭利 | Pulse type headache physical therapy device |
| US20130296748A1 (en) * | 2012-05-03 | 2013-11-07 | Sitkovetskiy Lev N. | Method of massage for improving of body condition, diagnosis, prevention and treatment diseases, and the massage device for these purposes. |
| CN105793280A (en) * | 2013-11-04 | 2016-07-20 | 伊玛提克斯生物技术有限公司 | Personalized immunotherapy against several neuronal and brain tumors |
| US20180137941A1 (en) * | 2015-06-02 | 2018-05-17 | Infervision Co., Ltd. | Method For Analysing Medical Treatment Data Based On Deep Learning and Intelligence Analyser Thereof |
| CN110348019A (en) * | 2019-07-17 | 2019-10-18 | 南通大学 | A kind of medical bodies vector method for transformation based on attention mechanism |
| CN111028951A (en) * | 2019-11-13 | 2020-04-17 | 上海中医药大学 | Method and equipment for Chinese medicine diagnosis retrieval display and evaluation |
Non-Patent Citations (3)
| Title |
|---|
| ALEXANDER ERNST等: "Feasibility of recanalization of human coronary arteries using high-intensity ultrasound", THE AMERICAN JOURNAL OF CARDIOLOGY, vol. 73, no. 2, pages 126 - 132 * |
| 李洪峥等: "基于改良动手深度学习算法的冠心病病变全程证候要素分布与组合规律研究", 世界科学技术-中医药现代化, vol. 23, no. 9, pages 3086 - 3094 * |
| 石秀峰: "基于双向LSTM的药物相互关系提取模型", 中国优秀硕士学位论文全文数据库基础科学辑, no. 2, pages 006 - 463 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112053785A (en) | 2020-12-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Kadhim et al. | Design and implementation of fuzzy expert system for back pain diagnosis | |
| Moein | Medical diagnosis using artificial neural networks | |
| CN111798954A (en) | Drug combination recommendation method based on time attention mechanism and graph convolution network | |
| TW201040756A (en) | Chinese medicine intelligent formulary system | |
| CN108986911A (en) | A kind of differential diagnosis in tcm opinion controls data processing method | |
| Ayat et al. | A comparison of artificial neural networks learning algorithms in predicting tendency for suicide | |
| CN108231146A (en) | A kind of medical records model building method, system and device based on deep learning | |
| Hung et al. | Predicting gastrointestinal bleeding events from multimodal in-hospital electronic health records using deep fusion networks | |
| Nohria | Medical expert system-A comprehensive review | |
| Chou et al. | Extracting drug utilization knowledge using self-organizing map and rough set theory | |
| Saeed et al. | The prognosis of allergy-based diseases using pythagorean fuzzy hypersoft mapping structures and recommending medication | |
| Hebbar et al. | Web powered CT scan diagnosis for brain hemorrhage using deep learning | |
| CN112270997A (en) | Coronary heart disease diagnosis method and system based on coronary heart disease diagnosis neural network model | |
| Devi et al. | Prediction of medicines using LVQ methodology | |
| JPWO2020176476A5 (en) | ||
| Klüver | Self-Enforcing Networks (SEN) for the development of (medical) diagnosis systems | |
| Chang et al. | Using machine learning algorithms in medication for cardiac arrest early warning system construction and forecasting | |
| Singh et al. | Heart disease prediction using classification and feature selection techniques | |
| Jokiniemi | Ontologies and computational methods for traditional Chinese medicine | |
| Morris | The Rise of Medicalised Mindfulness During the 1970s and 1980s: The Attempted Convergence of Religion and Science | |
| John et al. | A framework for medical diagnosis using hybrid reasoning | |
| CN113990502A (en) | ICU heart failure prediction system based on heterogeneous graph neural network | |
| Singla | Intelligent medical diagnostic system for diabetes | |
| Torres-Alegre et al. | Application of koniocortex-like networks to cardiac arrhythmias classification | |
| CN112466436A (en) | Intelligent traditional Chinese medicine evolution model training method and device based on recurrent neural network |
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 |