CN116825382A - Epileptic drug effectiveness detection method and device based on multi-modal fusion - Google Patents
Epileptic drug effectiveness detection method and device based on multi-modal fusion Download PDFInfo
- Publication number
- CN116825382A CN116825382A CN202310203047.7A CN202310203047A CN116825382A CN 116825382 A CN116825382 A CN 116825382A CN 202310203047 A CN202310203047 A CN 202310203047A CN 116825382 A CN116825382 A CN 116825382A
- Authority
- CN
- China
- Prior art keywords
- neural network
- clinical data
- network model
- training
- epileptic
- 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
- 230000001037 epileptic effect Effects 0.000 title claims abstract description 60
- 239000003814 drug Substances 0.000 title claims abstract description 55
- 229940079593 drug Drugs 0.000 title claims abstract description 55
- 230000004927 fusion Effects 0.000 title claims abstract description 37
- 238000001514 detection method Methods 0.000 title claims abstract description 28
- 238000003062 neural network model Methods 0.000 claims abstract description 104
- 238000012549 training Methods 0.000 claims abstract description 87
- 208000026911 Tuberous sclerosis complex Diseases 0.000 claims abstract description 46
- 208000009999 tuberous sclerosis Diseases 0.000 claims abstract description 46
- 239000011159 matrix material Substances 0.000 claims abstract description 44
- 238000007781 pre-processing Methods 0.000 claims abstract description 12
- 230000006870 function Effects 0.000 claims description 42
- 238000000034 method Methods 0.000 claims description 29
- 239000002131 composite material Substances 0.000 claims description 16
- 238000011282 treatment Methods 0.000 claims description 11
- 238000000540 analysis of variance Methods 0.000 claims description 10
- 238000013528 artificial neural network Methods 0.000 claims description 8
- 238000002790 cross-validation Methods 0.000 claims description 5
- 238000010998 test method Methods 0.000 claims description 5
- 230000002068 genetic effect Effects 0.000 claims description 4
- 238000013473 artificial intelligence Methods 0.000 abstract description 4
- 238000011156 evaluation Methods 0.000 abstract description 2
- 238000002595 magnetic resonance imaging Methods 0.000 description 11
- 206010010904 Convulsion Diseases 0.000 description 10
- 238000010586 diagram Methods 0.000 description 10
- 239000001961 anticonvulsive agent Substances 0.000 description 5
- 229960003965 antiepileptics Drugs 0.000 description 4
- 238000002591 computed tomography Methods 0.000 description 4
- 238000004590 computer program Methods 0.000 description 4
- 206010059866 Drug resistance Diseases 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 230000003044 adaptive effect Effects 0.000 description 2
- 238000013459 approach Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 230000003111 delayed effect Effects 0.000 description 2
- 201000010099 disease Diseases 0.000 description 2
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 206010015037 epilepsy Diseases 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 238000001000 micrograph Methods 0.000 description 2
- 108090000623 proteins and genes Proteins 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 229940126585 therapeutic drug Drugs 0.000 description 2
- 238000003325 tomography Methods 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 208000024827 Alzheimer disease Diseases 0.000 description 1
- 208000001654 Drug Resistant Epilepsy Diseases 0.000 description 1
- 101000795659 Homo sapiens Tuberin Proteins 0.000 description 1
- 208000026350 Inborn Genetic disease Diseases 0.000 description 1
- 208000018737 Parkinson disease Diseases 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 208000016361 genetic disease Diseases 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 239000004973 liquid crystal related substance Substances 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
- 230000035772 mutation Effects 0.000 description 1
- 230000000324 neuroprotective effect Effects 0.000 description 1
- 238000011369 optimal treatment Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000002271 resection Methods 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 238000010187 selection method Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
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
- G16H70/00—ICT specially adapted for the handling or processing of medical references
- G16H70/40—ICT specially adapted for the handling or processing of medical references relating to drugs, e.g. their side effects or intended usage
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H10/00—ICT specially adapted for the handling or processing of patient-related medical or healthcare data
- G16H10/60—ICT specially adapted for the handling or processing of patient-related medical or healthcare data for patient-specific data, e.g. for electronic patient records
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Public Health (AREA)
- Primary Health Care (AREA)
- Epidemiology (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Pharmacology & Pharmacy (AREA)
- Medicinal Chemistry (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Toxicology (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Chemical & Material Sciences (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The invention discloses a multimode fusion-based epileptic drug effectiveness detection method, which comprises the following steps: collecting clinical data of a patient suffering from tuberous sclerosis; preprocessing clinical data to obtain processed clinical data; randomly selecting samples from the processed clinical data and creating a small sample graph as input characteristics to obtain a training characteristic matrix; constructing a neural network model, inputting a training feature matrix into the neural network model for training, and obtaining a trained neural network model; and inputting the clinical data of the newly acquired patient with the tuberous sclerosis into the trained neural network model to obtain a detection result. The invention fuses clinical data in multiple modes, and utilizes artificial intelligence to carry out rapid auxiliary detection, thereby having important auxiliary significance for accuracy and standardization of validity evaluation of epileptic drugs.
Description
Technical Field
The invention relates to the field of artificial intelligence, in particular to a method and a device for detecting the effectiveness of epileptic drugs based on multi-modal fusion.
Background
Tuberous Sclerosis (TSC) is an autosomal dominant genetic disease caused by mutations in the TSC1 or TSC2 genes. From 80% to 90% of TSC patients develop seizures, with nearly two-thirds of them first seizures in the first year of birth. 60% of epileptic patients are inherently resistant to epileptic therapeutic drugs, and epilepsy is currently treated only by attempts to administer different epileptic therapeutic drugs to epileptic patients, which may take a long time to determine that the patient is resistant if the patient is resistant, with an increased risk of mortality. However, if only clinical intervention is used, it is insufficient to detect whether an epileptic patient is drug resistant within a short period of time. This would be an expensive and time consuming effort to ensure that epileptic patients are resistant. Not only multiple MRI images, computerized tomography (Computed Tomography, CT) images, and Electroencephalogram (EEG) scans are required for a patient, but also an experienced imaging physician is required to diagnose by combining images of multiple modalities, which has high experience and technical requirements for the physician, resulting in low accuracy of clinical judgment.
In terms of artificial intelligence diagnostics, graph roll-up neural networks (Graph Convolutional networks, GCN) in deep learning are of great interest because of their ability to effectively fuse multimodal features and model correlations between samples. Graphs are widely used as a natural framework for capturing interactions between elements represented as nodes in the graph, and furthermore, graph-based approaches can be interpreted more strongly than CNNs. The graphic neural network model has been widely used for diagnosing various diseases, and it has been demonstrated that it is possible to diagnose and classify Alzheimer's disease, parkinson's disease on magnetic resonance imaging (Magnetic Resonance Imaging, MRI) with high accuracy. However, deep learning algorithms often require a large number of data sets to improve accuracy, but clinical data for TSC patients is difficult to collect, resulting in a smaller sample size for TSC patients. This presents challenges for the detection and classification of epileptic drug effectiveness.
Accordingly, the prior art is still in need of improvement and development.
Disclosure of Invention
The invention aims to solve the technical problems that in the prior art, the method and the device for detecting the effectiveness of the epileptic drug based on multi-mode fusion are provided, and aim to solve the problems that in the prior art, the sample size of a TSC patient is small, the accuracy of clinical judgment is low and the effectiveness detection of the clinical epileptic drug is difficult.
The technical scheme adopted for solving the technical problems is as follows:
in a first aspect, the present invention provides a method for detecting the effectiveness of an epileptic drug based on multimodal fusion, wherein the method comprises:
collecting clinical data of a patient suffering from tuberous sclerosis;
preprocessing the clinical data to obtain processed clinical data;
randomly selecting samples from the processed clinical data and creating a small sample graph as input characteristics to obtain a training characteristic matrix;
constructing a neural network model, and inputting the training feature matrix into the neural network model for training to obtain a trained neural network model;
and inputting the clinical data of the newly acquired patient with the tuberous sclerosis into the trained neural network model to obtain a detection result.
In one implementation, the collecting clinical data of a patient with tuberous sclerosis comprises:
collecting clinical data of a patient suffering from tuberous sclerosis; wherein the patient with tuberous sclerosis receives epileptic medication for more than one year and has not received an excision procedure; wherein the clinical data comprises: clinical, MRI, CT, EEG, genetic; the categories of clinical features include controlled and uncontrolled;
classifying the clinical data of the clinical feature class as control into a control group; wherein the control group comprises clinical data that reduces seizure number by 50% or more in one year over the last year;
categorizing the clinical features into uncontrolled groups of clinical data; wherein the uncontrolled group comprises clinical data that reduces seizure number by less than 50% in one year over the last year.
In one implementation, the preprocessing the clinical data to obtain processed clinical data includes:
removing redundant features in the clinical data; wherein the redundancy feature comprises: numbering, name, year and month of birth, date of examination;
filling the missing value in the clinical data by adopting a median;
and calculating an analysis of variance F value by adopting an analysis of variance F test method, arranging the F values in a descending order, and taking the F values as the clinical data after treatment according to the first 35 features.
In one implementation, the randomly selecting samples in the processed clinical data and creating a small sample graph as input features, to obtain a training feature matrix, includes:
randomly selecting samples to create a small sample graph as an input feature, and obtaining a training feature matrix X as a label of an unknown node in a graph structure through multiple layers
Where x is a known sample from the controlled and uncontrolled groups,for unknown samples, C is a label, C is the number of categories of the clinical feature, N is the number of nodes, and n=c·q+1, where q is the number of known samples per category.
In one implementation, the building a neural network model includes:
according to the graph convolution neural network, a first neural network model is obtained;
creating a loss function according to the cross entropy loss function and creating a composite loss function, wherein the composite loss function is used for rewarding edges involved in correct classification and punishing edges causing incorrect classification, and specifically comprises the following steps:
wherein L is cross-entropy L is a cross entropy loss function graph As a function of the composite loss,E(a i ) (t+1) ＝0.9*E(a i ) (t) +0.1*a i the total loss function is L AMDGM ＝L ross-entropy +L graph The method comprises the steps of carrying out a first treatment on the surface of the Wherein y is i For the ith sample predictor, z i A tag value for the i-th sample; delta (y) i ,z i ) For the bonus function, the average precision E (a i ) And the current value a i Differences between if y i ＝z i Then a i And is 1, otherwise 0,edge weight P for layer I ij (X; θ, t), and E (a) in step t+1 i ) (t+1) From E (a) i ) (t) And a i Is jointly formed;
obtaining a second neural network model according to the first neural network model, the loss function and the composite loss function;
and evaluating the second neural network model by adopting five-fold cross validation to obtain the neural network model.
In one implementation manner, the inputting the training feature matrix into the neural network model for training, to obtain a trained neural network model, includes:
inputting the training feature matrix X into a graph convolution neural network model for training to obtain a label Y of an unknown node;
according to the label Y of the unknown node, back propagation of the neural network model is carried out, and updated network parameters are obtained;
and updating the neural network model according to the updated network parameters to obtain an updated neural network model, and re-executing the steps of randomly selecting samples in the processed clinical data and creating a small sample graph as input features to obtain a training feature matrix until the iterative training times are completed to finally obtain the trained neural network model.
In one implementation, the building a neural network model and inputting the training feature matrix into the neural network model for training further includes:
setting the learning rate of the neural network model to be 0.001, setting the epoch to be 100, and setting the iterative training frequency to be 100.
In a second aspect, an embodiment of the present invention further provides an epileptic drug effectiveness detection apparatus based on multimodal fusion, where the apparatus includes:
the clinical data acquisition module is used for acquiring clinical data of patients with tuberous sclerosis;
the pretreatment module is used for carrying out pretreatment on the clinical data to obtain the treated clinical data;
the training feature matrix acquisition module is used for randomly selecting samples from the processed clinical data and creating a small sample graph as input features to obtain a training feature matrix;
the training module is used for constructing a neural network model, inputting the training feature matrix into the neural network model for training, and obtaining a trained neural network model;
and the detection module is used for inputting the clinical data of the newly acquired patient with the tuberous sclerosis into the trained neural network model to obtain a detection result.
In a third aspect, an embodiment of the present invention further provides an intelligent terminal, where the intelligent terminal includes a memory, a processor, and a multimode fusion-based epileptic drug validity detection program stored in the memory and capable of running on the processor, where the steps of the multimode fusion-based epileptic drug validity detection method described in any one of the above are implemented when the processor executes the multimode fusion-based epileptic drug validity detection program.
In a fourth aspect, an embodiment of the present invention further provides a computer readable storage medium, where the computer readable storage medium stores an epileptic drug validity detection program based on multi-modal fusion, where the epileptic drug validity detection program based on multi-modal fusion implements the steps of the epileptic drug validity detection method based on multi-modal fusion as described in any one of the above.
The beneficial effects are that: compared with the prior art, the invention provides a method for detecting the effectiveness of epileptic drugs based on multi-mode fusion. Firstly, collecting clinical data of a patient with tuberous sclerosis; the clinical data includes: clinical features, MRI features, CT features, EEG features and gene features, the multi-modal fusion of clinical data is realized, and the problem of small sample size of TSC patients is solved. The clinical data is then preprocessed to obtain standard clinical data. Then, samples are randomly selected from the processed clinical data, a small sample graph is created as input features, a training feature matrix is obtained, and the flexibility of the model is enhanced by creating the small sample graph, so that the calculation burden is reduced. The unknown samples are randomly selected, so that the problems of unbalance of positive and negative samples of our data and small sample data size are solved. And finally, constructing a neural network model, training to obtain a trained neural network model, and inputting clinical data of a new acquired patient with the tuberous sclerosis into the trained neural network model so as to accurately predict the effectiveness of the epileptic drug in an early stage.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are required to be used in the embodiments or the description of the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments described in the present invention, and other drawings may be obtained according to the drawings without inventive effort to those skilled in the art.
Fig. 1 is a schematic flow chart of a method for detecting the effectiveness of epileptic drugs based on multimodal fusion according to an embodiment of the present invention.
Fig. 2 (a) is a schematic diagram of a DGM model of a neural network according to an embodiment of the present invention.
Fig. 2 (b) is a schematic diagram of a neural network AMDGM model according to an embodiment of the invention.
FIG. 2 (c) is a schematic diagram of training and testing strategies for providing a neural network AMDGM model according to an embodiment of the present invention.
Fig. 3 is a schematic block diagram of an epileptic drug effectiveness detection apparatus based on multimodal fusion according to an embodiment of the present invention.
Fig. 4 is a schematic block diagram of an internal structure of an intelligent terminal according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and effects of the present invention clearer and more specific, the present invention will be described in further detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
As used herein, the singular forms "a", "an", "the" and "the" are intended to include the plural forms as well, unless expressly stated otherwise, as understood by those skilled in the art. It will be further understood that the terms "comprises" and/or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof. It will be understood that when an element is referred to as being "connected" or "coupled" to another element, it can be directly connected or coupled to the other element or intervening elements may also be present. Further, "connected" or "coupled" as used herein may include wirelessly connected or wirelessly coupled. The term "and/or" as used herein includes all or any element and all combination of one or more of the associated listed items.
It will be understood by those skilled in the art that all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs unless defined otherwise. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.
The main ways of judging the effectiveness of epileptic drugs clinically at present are as follows: if one anti-epileptic drug is not effective, the other anti-epileptic drug is not effective after a plurality of anti-epileptic drugs are tried, and then the patient is judged to be resistant to epilepsy, also called refractory epilepsy. If multiple medications are not available and the process of diagnosing drug-resistant seizures may take a longer time, the timing of TSC treatment may be delayed. In addition, various MRI images, computed tomography (Computed Tomography, CT) images, and Electroencephalogram (EEG) scans are performed on the patient, and diagnosis is performed by an experienced imaging physician in combination with images of multiple modalities. This approach has several drawbacks: the time cost is high. Most epileptic patients are diagnosed with drug resistance but have delayed the treatment time of TSC and have missed the optimal treatment opportunity. Therefore, if the epileptic drug resistance can be confirmed in the early or subclinical stage, the patient can be correspondingly neuroprotective treated as soon as possible. The labor cost is high, and in the multi-mode image scanning, a doctor who is trained in a professional way can diagnose the images due to the specificity of the medical images; with experience limitations, young doctors have difficulty in getting on hand quickly, and in addition, due to experience differences, different people can judge the same image differently.
In order to solve the problems, the invention provides a method for detecting the effectiveness of epileptic drugs based on multi-mode fusion. Firstly, collecting clinical data of a patient with tuberous sclerosis; the clinical data includes: clinical features, MRI features, CT features, EEG features and gene features, the multi-modal fusion of clinical data is realized, and the problem of small sample size of TSC patients is solved. The clinical data is then preprocessed to obtain standard clinical data. Then, samples are randomly selected from the processed clinical data, a small sample graph is created as input features, a training feature matrix is obtained, and the flexibility of the model is enhanced by creating the small sample graph, so that the calculation burden is reduced. The unknown samples are randomly selected, so that the problems of unbalance of positive and negative samples of our data and small sample data size are solved. And finally, constructing a neural network model, training to obtain a trained neural network model, and inputting clinical data of a new acquired patient with the tuberous sclerosis into the trained neural network model so as to accurately predict the effectiveness of the epileptic drug in an early stage.
Exemplary method
The embodiment provides an epileptic drug effectiveness detection method based on multi-mode fusion. As shown in fig. 1, the method comprises the steps of:
step S100, collecting clinical data of patients with tuberous sclerosis;
specifically, the acquisition of clinical data is performed by performing a variety of MRI images, computerized tomography (Computed Tomography, CT) images, brain wave (EEG) scans on the patient to diagnose in combination with the images of the multiple modalities. In this example, all clinical data were from Shenzhen children hospital, and written informed consent was obtained from all subjects prior to the study. The proposal is approved by the institute of advanced technology and inspection committee (IRB) of Shenzhen, national academy of sciences. The clinical data contained 103 TSC confirmed cases.
In one implementation, the step S100 in this embodiment includes the following steps:
step S101, collecting clinical data of a patient with tuberous sclerosis; wherein the patient with tuberous sclerosis receives epileptic medication for more than one year and has not received an excision procedure; wherein the clinical data comprises: clinical, MRI, CT, EEG, genetic; the categories of clinical features include controlled and uncontrolled;
step S102, classifying the clinical data with the clinical characteristics as control into a control group; wherein the control group comprises clinical data that reduces seizure number by 50% or more in one year over the last year;
step S103, classifying the clinical data with the clinical characteristics as uncontrolled clinical data into uncontrolled groups; wherein the uncontrolled group comprises clinical data that reduces seizure number by less than 50% in one year over the last year.
Specifically, in this example, 103 patients with TSC confirmed cases were aged between 50 days and 16 years at the time of visit, with a median of 3 years. All patients received at least one year of anti-epileptic drug (AED) treatment, and none received resections. The clinical data is in tabular form, containing 109 columns of clinical data information for 103 patients. The epileptic drug treatment results of patients are included in the category of clinical features of the clinical data and fall into two categories, controlled and uncontrolled. After administration, the number of seizures is reduced to less than half of the original number under at least one year of observation, and the classification of clinical features is divided into controlled groups, otherwise, uncontrolled groups. Class 0 of clinical features indicates control, and 1 indicates no control.
Step 200, preprocessing the clinical data to obtain processed clinical data;
specifically, in order to obtain effective clinical data, redundant, repeated and missing invalid data are removed, so that the training efficiency is improved, and the calculation force is saved. In this embodiment, 109 columns of clinical data information of 103 patients acquired will be subjected to data preprocessing.
In one implementation, the step S200 in this embodiment includes the following steps:
step S201, removing redundant features in the clinical data; wherein the redundancy feature comprises: numbering, name, year and month of birth, date of examination;
step S202, filling the missing value in the clinical data by adopting a median;
step S203, calculating an analysis of variance F value by adopting an analysis of variance F test method, arranging the F values in a descending order, and taking the F values as the processed clinical data according to the first 35 features.
Specifically, in this embodiment, the method for preprocessing clinical data includes removing features unrelated to experimental tasks, such as numbers, names, birth months, and inspection dates, and filling in missing values. The missing value in the data has two cases, one is omission in recording, and filling is only needed according to a default value. The other is that the patients with the disease age deficiency of 3 patients are filled in according to the median. And finally, selecting the characteristics. For the feature selection algorithm, an analysis of variance F test method is used. The specific selection method comprises the following steps: and calculating the F value of the variance analysis of the features, then arranging the F values in descending order of the results, and finally selecting the first 35 features.
Step S300, randomly selecting samples from the processed clinical data and creating a small sample graph as input characteristics to obtain a training characteristic matrix;
in one implementation, the step S300 in this embodiment includes the following steps:
step S301, randomly selecting samples to create a small sample graph as an input feature, and obtaining a training feature matrix X as a label of an unknown node in a graph structure through multiple layers
Where x is a known sample from the controlled and uncontrolled groups,for unknown samples, C is a label, C is the number of categories of the clinical feature, N is the number of nodes, and n=c·q+1, where q is the number of known samples per category.
Specifically, to solve the problem of unbalance of positive and negative samples of data and small samples of data, the present embodiment proposes an AMDGM model (Auto-Metric Differentiable Graph Module). In this embodiment, a small sample graph created from randomly selected samples is used as input to obtain the labels of unknown nodes in the graph structure through multiple layers. When a model is trained by multiple tasks, an unknown sample is randomly extracted for each training, and even if the graph is small, the information contained in the multiple graphs is enough for the model to obtain classification experience.
Specifically, the greater the number of nodes in the graph, the higher the computational resources consumed. In addition, a large amount of information of the same category is superfluous, not leading to significant performance improvement. In this embodiment, the use of the small graph enhances the flexibility of the model, thereby reducing the computational burden. In addition, the data used to construct the graph is randomly selected from the training samples so that a small number of samples can generate a large number of different graphs. For example, to construct a graph with three nodes, when the number of training samples is 10 (positive example is 5 and negative example is 5), we can randomly select one from the positive samples, one from the negative samples, and then use the remaining samples as unknown nodes. Thus we have a combination of 5×5×8=200, which means that at most 200 different graphs can be constructed. This number is well above 10. Therefore, even if the number of samples is small, a good training effect can be achieved, and the problems of unbalanced positive and negative samples and small sample data size of the data can be solved.
For example, there are two classes of classification of clinical features in the task of predicting epileptic drug effectiveness: a controlled group and an uncontrolled group. When q is equal to 20, 20 samples are randomly selected from each category, and one sample of unknown label is added, namely, the node number in the graph is c·q+1=41.
Step S400, constructing a neural network model, and inputting the training feature matrix into the neural network model for training to obtain a trained neural network model;
specifically, as shown in fig. 2, in this embodiment, the sampled feature matrix X of the micro graph is initially input into the DGM model, so as to obtain a label Y of the unknown node. The network parameters are updated through supervised learning, and the updated parameters are used as initial parameters for training the subsequent new tasks and then are subjected to iterative training.
In one implementation, the step S400 in this embodiment includes the following steps:
step S401, a neural network is rolled according to a graph to obtain a first neural network model;
step S402, creating a loss function according to the cross entropy loss function, and creating a composite loss function, wherein the composite loss function is used for rewarding edges involved in correct classification and punishing edges causing incorrect classification, and specifically comprises the following steps:
wherein L is cross-entropy L is a cross entropy loss function graph As a function of the composite loss,E(a i ) (t+1) ＝0.9*E(a i ) (t) +0.1*a i the total loss function is L AMDGM ＝L ross-entropy +L graph The method comprises the steps of carrying out a first treatment on the surface of the Wherein y is i For the ith sample predictor, z i A tag value for the i-th sample; delta (y) i ,z i ) For the bonus function, the average precision E (a i ) And the current value a i Differences between if y i ＝z i Then a i And is 1, otherwise 0,edge weight P for layer I ij (X; θ, t), and E (a) in step t+1 i ) (t+1) From E (a) i ) (t) And a i Is jointly formed;
step S403, obtaining a second neural network model according to the first neural network model, the loss function and the composite loss function;
and step S404, evaluating the second neural network model by adopting five-fold cross validation to obtain the neural network model.
Step S405, inputting the training feature matrix X into a graph convolution neural network model for training to obtain a label Y of an unknown node;
step S406, according to the label Y of the unknown node, carrying out back propagation of the neural network model to obtain updated network parameters;
step S407, setting the learning rate of the neural network model to be 0.001, setting epoch to be 100, and setting the iterative training frequency to be 100.
Step S408, updating the neural network model according to the updated network parameters to obtain an updated neural network model, and re-executing the step of randomly selecting samples in the processed clinical data and creating a small sample graph as input features to obtain a training feature matrix until the iterative training times are completed to finally obtain a trained neural network model.
Specifically, in this embodiment, after the neural network model is built, the neural network model is trained. Five-fold cross-validation was used to evaluate the model. In the example experiment, an adaptive motion estimation algorithm (Adaptive Moment Estimation, adam) is adopted to train a network, the learning rate is set to 0.001, the epoch is set to 100, the training iteration number of the micro-graph is 100, and a cross entropy loss function (L cross-entropy ). In addition to this, a composite loss function (L graph ) Edges involved in the correct classification are rewarded and edges that lead to incorrect classification are penalized.
For example, let the unknown sample be a, the label predicted by the DGM model be Y, which is the label Y predicted by the model, if the label of a of the original sample is S, comparing Y with S, if Y and S are the same, the prediction is correct, if Y and S are different, the prediction is not correct, and the network needs to be updated. The network performs back propagation to update network parameters, which is one kind of supervised learning. Firstly, constructing a small diagram, wherein the small diagram comprises an unknown sample a and q known samples, obtaining a prediction label Y of the unknown sample a after model training, and updating network parameters of the model through back propagation. Then we randomly extract the samples to reconstruct a small graph, and then back-propagate, i.e. iterate the training continuously. Finally, a trained neural network model is obtained and used for predicting the test set.
And S500, inputting the clinical data of the newly acquired patient with the tuberous sclerosis into the trained neural network model to obtain a detection result.
Specifically, in this embodiment, clinical data of multiple modes fused with a patient suffering from tuberous sclerosis is newly collected, and then the clinical data is input into the trained neural network model after being processed to obtain clinical characteristics classified as controlled and uncontrolled, if the clinical characteristics are classified as controlled, it is indicated that the epileptic drug is effective, and if the clinical characteristics are classified as uncontrolled, it is indicated that the patient suffering from tuberous sclerosis has developed drug resistance to the epileptic drug, that is, the epileptic drug is ineffective. So as to accurately predict the effectiveness of epileptic drugs in early stage.
Exemplary apparatus
As shown in fig. 3, the present embodiment further provides an apparatus, including:
a clinical data acquisition module 10 for acquiring clinical data of a patient with tuberous sclerosis;
a preprocessing module 20, configured to preprocess the clinical data to obtain processed clinical data;
a training feature matrix acquisition module 30, configured to randomly select samples from the processed clinical data and create a small sample graph as input features, so as to obtain a training feature matrix;
the training module 40 is configured to construct a neural network model, and input the training feature matrix into the neural network model for training, so as to obtain a trained neural network model;
the detection module 50 is used for inputting the clinical data of the new acquired patient with the tuberous sclerosis into the trained neural network model to obtain a detection result.
In one implementation, the acquisition module 10 includes:
a clinical data acquisition unit for acquiring clinical data of a patient with tuberous sclerosis; wherein the patient with tuberous sclerosis receives epileptic medication for more than one year and has not received an excision procedure; wherein the clinical data comprises: clinical, MRI, CT, EEG, genetic; the categories of clinical features include controlled and uncontrolled;
a first grouping unit for grouping clinical data of which the category of the clinical feature is control into a control group; wherein the control group comprises clinical data that reduces seizure number by 50% or more in one year over the last year;
a second grouping unit for grouping the clinical data of the clinical feature category as uncontrolled into uncontrolled groups; wherein the uncontrolled group comprises clinical data that reduces seizure number by less than 50% in one year over the last year.
In one implementation, the preprocessing module 20 includes:
a redundant feature removal unit configured to remove redundant features in the clinical data; wherein the redundancy feature comprises: numbering, name, year and month of birth, date of examination;
the missing value filling unit is used for filling missing values in the clinical data by adopting a median;
and the analysis of variance unit is used for calculating an analysis of variance F value by adopting an analysis of variance F test method, arranging the F values in a descending order and taking the F values as the clinical data after the treatment according to the first 35 characteristics.
In one implementation, the training feature matrix acquisition module 30 includes:
the training feature matrix acquisition unit is used for randomly selecting samples to create a small sample graph as an input feature, and obtaining the training feature matrix X as an input feature by obtaining labels of unknown nodes in the graph structure through multiple layers
Where x is a known sample from the controlled and uncontrolled groups,for unknown samples, C is a label, C is the number of categories of the clinical feature, N is the number of nodes, and n=c·q+1, where q is the number of known samples per category.
In one implementation, the training module 40 includes:
the first neural network model acquisition unit is used for rolling the neural network according to the graph to obtain a first neural network model;
a loss function creation unit for creating a loss function from the cross entropy loss function and creating a composite loss function for rewarding edges involved in correct classification and punishing edges causing incorrect classification, specifically:
wherein L is cross-entropy L is a cross entropy loss function graph As a function of the composite loss,E(a i ) (t+1) ＝0.9*E(a i ) (t) +0.1*a i the total loss function is L AMDGM ＝L ross-entropy +L graph The method comprises the steps of carrying out a first treatment on the surface of the Wherein y is i For the ith sample predictor, z i A tag value for the i-th sample; delta (y) i ,z i ) For the bonus function, the average precision E (a i ) And the current value a i Differences between if y i ＝z i Then a i And is 1, otherwise 0,edge weight P for layer I ij (X; θ, t), and E (a) in step t+1 i ) (t+1) From E (ai) ) (t) And a i Is jointly formed;
the second neural network model acquisition unit is used for acquiring a second neural network model according to the first neural network model, the loss function and the composite loss function;
and the neural network model acquisition unit is used for evaluating the second neural network model by adopting five-fold cross validation to obtain the neural network model.
The training unit is used for inputting the training feature matrix X into the graph convolution neural network model for training to obtain a label Y of an unknown node;
the network parameter acquisition unit is used for carrying out back propagation of the neural network model according to the label Y of the unknown node to obtain updated network parameters;
and the parameter setting unit is used for setting the learning rate of the neural network model to be 0.001, the epoch to be 100, and the iterative training frequency to be 100.
And the iterative training unit is used for updating the neural network model according to the updated network parameters to obtain an updated neural network model, and re-executing the steps of randomly selecting samples in the processed clinical data and creating a small sample graph as input characteristics to obtain a training characteristic matrix until the iterative training times are completed to finally obtain the trained neural network model.
Based on the above embodiment, the present invention further provides an intelligent terminal, and a functional block diagram thereof may be shown in fig. 4. The intelligent terminal comprises a processor, a memory, a network interface, a display screen and a temperature sensor which are connected through a system bus. The processor of the intelligent terminal is used for providing computing and control capabilities. The memory of the intelligent terminal comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface of the intelligent terminal is used for communicating with an external terminal through network connection. The computer program, when executed by a processor, implements a method for detecting the effectiveness of epileptic drugs based on multimodal fusion. The display screen of the intelligent terminal can be a liquid crystal display screen or an electronic ink display screen, and a temperature sensor of the intelligent terminal is arranged in the intelligent terminal in advance and used for detecting the running temperature of internal equipment.
It will be appreciated by those skilled in the art that the schematic block diagram shown in fig. 4 is merely a block diagram of a portion of the structure associated with the present inventive arrangements and is not limiting of the intelligent terminal to which the present inventive arrangements are applied, and that a particular intelligent terminal may include more or less components than those shown, or may combine some of the components, or may have a different arrangement of components.
In one embodiment, an intelligent terminal is provided, the intelligent terminal includes a memory, a processor, and a multimode fusion-based epileptic drug effectiveness detection program stored in the memory and capable of running on the processor, and when the processor executes the multimode fusion-based epileptic drug effectiveness detection program, the following operation instructions are implemented:
collecting clinical data of a patient suffering from tuberous sclerosis;
preprocessing the clinical data to obtain processed clinical data;
randomly selecting samples from the processed clinical data and creating a small sample graph as input characteristics to obtain a training characteristic matrix;
constructing a neural network model, and inputting the training feature matrix into the neural network model for training to obtain a trained neural network model;
and inputting the clinical data of the newly acquired patient with the tuberous sclerosis into the trained neural network model to obtain a detection result.
Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, storage, operational database, or other medium used in embodiments provided herein may include non-volatile and/or volatile memory. The nonvolatile 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), dual operation data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous Link DRAM (SLDRAM), memory bus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), among others.
In summary, the invention discloses a multimode fusion-based epileptic drug effectiveness detection method, which comprises the following steps: collecting clinical data of a patient suffering from tuberous sclerosis; preprocessing clinical data to obtain processed clinical data; randomly selecting samples from the processed clinical data and creating a small sample graph as input characteristics to obtain a training characteristic matrix; constructing a neural network model, inputting a training feature matrix into the neural network model for training, and obtaining a trained neural network model; and inputting the clinical data of the newly acquired patient with the tuberous sclerosis into the trained neural network model to obtain a detection result. The invention fuses clinical data in multiple modes, and utilizes artificial intelligence to carry out rapid auxiliary detection, thereby having important auxiliary significance for accuracy and standardization of validity evaluation of epileptic drugs.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (10)
1. A method for detecting the effectiveness of epileptic drugs based on multimodal fusion, which is characterized by comprising the following steps:
collecting clinical data of a patient suffering from tuberous sclerosis;
preprocessing the clinical data to obtain processed clinical data;
randomly selecting samples from the processed clinical data and creating a small sample graph as input characteristics to obtain a training characteristic matrix;
constructing a neural network model, and inputting the training feature matrix into the neural network model for training to obtain a trained neural network model;
and inputting the clinical data of the newly acquired patient with the tuberous sclerosis into the trained neural network model to obtain a detection result.
2. The method for detecting the effectiveness of epileptic drugs based on multimodal fusion according to claim 1, wherein said collecting clinical data of patients with tuberous sclerosis comprises:
collecting clinical data of a patient suffering from tuberous sclerosis; wherein the patient with tuberous sclerosis receives epileptic medication for more than one year and has not received an excision procedure; wherein the clinical data comprises: clinical, MRI, CT, EEG, genetic; the categories of clinical features include controlled and uncontrolled;
classifying the clinical data of the clinical feature class as control into a control group; wherein the control group comprises clinical data that reduces seizure number by 50% or more in one year over the last year;
categorizing the clinical features into uncontrolled groups of clinical data; wherein the uncontrolled group comprises clinical data that reduces seizure number by less than 50% in one year over the last year.
3. The method for detecting the validity of epileptic drugs based on multimodal fusion according to claim 1, wherein the preprocessing of the clinical data to obtain the processed clinical data comprises the following steps:
removing redundant features in the clinical data; wherein the redundancy feature comprises: numbering, name, year and month of birth, date of examination;
filling the missing value in the clinical data by adopting a median;
and calculating an analysis of variance F value by adopting an analysis of variance F test method, arranging the F values in a descending order, and taking the F values as the clinical data after treatment according to the first 35 features.
4. The method for detecting the validity of epileptic drugs based on multi-modal fusion according to claim 2, wherein the steps of randomly selecting samples from the processed clinical data and creating a small sample graph as input features to obtain a training feature matrix include:
randomly selecting samples to create a small sample graph as an input feature, and obtaining a training feature matrix X as a label of an unknown node in a graph structure through multiple layers
Where x is a known sample from the controlled and uncontrolled groups,for unknown samples, C is a label, C is the number of categories of the clinical feature, N is the number of nodes, and n=c·q+1, where q is the number of known samples per category.
5. The method for detecting the validity of epileptic drugs based on multi-modal fusion according to claim 4, wherein the constructing a neural network model comprises:
according to the graph convolution neural network, a first neural network model is obtained;
creating a loss function according to the cross entropy loss function and creating a composite loss function, wherein the composite loss function is used for rewarding edges involved in correct classification and punishing edges causing incorrect classification, and specifically comprises the following steps:
wherein L is cross-entropy L is a cross entropy loss function graph As a function of the composite loss,E(a i ) (t+1) ＝0.9*E(a i ) (t) +0.1*a i the total loss function is L AMDGM ＝L ross-entropy +L graph The method comprises the steps of carrying out a first treatment on the surface of the Wherein y is i For the ith sample predictor, z i A tag value for the i-th sample; delta (y) i ,z i ) For the bonus function, the average precision E (a i ) And the current value a i Differences between if y i ＝z i Then a i 1, otherwise 0, < >>Edge weight P for layer I ij (X; θ, t), and E (a) in step t+1 i ) (t+1) From E (a) i ) (t) And a i Is jointly formed;
obtaining a second neural network model according to the first neural network model, the loss function and the composite loss function;
and evaluating the second neural network model by adopting five-fold cross validation to obtain the neural network model.
6. The method for detecting epileptic drug effectiveness based on multimodal fusion according to claim 5, wherein the inputting the training feature matrix into the neural network model for training to obtain a trained neural network model comprises:
inputting the training feature matrix X into a graph convolution neural network model for training to obtain a label Y of an unknown node;
according to the label Y of the unknown node, back propagation of the neural network model is carried out, and updated network parameters are obtained;
and updating the neural network model according to the updated network parameters to obtain an updated neural network model, and re-executing the steps of randomly selecting samples in the processed clinical data and creating a small sample graph as input features to obtain a training feature matrix until the iterative training times are completed to finally obtain the trained neural network model.
7. The method for detecting the validity of epileptic drugs based on multimodal fusion according to claim 6, wherein the constructing a neural network model and inputting the training feature matrix into the neural network model for training further comprises:
setting the learning rate of the neural network model to be 0.001, setting the epoch to be 100, and setting the iterative training frequency to be 100.
8. An epileptic drug effectiveness detection device based on multimodal fusion, the device comprising:
the clinical data acquisition module is used for acquiring clinical data of patients with tuberous sclerosis;
the pretreatment module is used for carrying out pretreatment on the clinical data to obtain the treated clinical data;
the training feature matrix acquisition module is used for randomly selecting samples from the processed clinical data and creating a small sample graph as input features to obtain a training feature matrix;
the training module is used for constructing a neural network model, inputting the training feature matrix into the neural network model for training, and obtaining a trained neural network model;
and the detection module is used for inputting the clinical data of the newly acquired patient with the tuberous sclerosis into the trained neural network model to obtain a detection result.
9. An intelligent terminal, characterized in that the intelligent terminal comprises a memory, a processor and a multimode fusion-based epileptic drug effectiveness detection program stored in the memory and capable of running on the processor, wherein the steps of the multimode fusion-based epileptic drug effectiveness detection method according to any one of claims 1-7 are realized when the multimode fusion-based epileptic drug effectiveness detection program is executed by the processor.
10. A computer readable storage medium, wherein the computer readable storage medium has stored thereon an epileptic drug effectiveness detection program based on multi-modal fusion, which when executed by a processor, implements the steps of the epileptic drug effectiveness detection method based on multi-modal fusion as defined in any one of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310203047.7A CN116825382A (en) | 2023-02-23 | 2023-02-23 | Epileptic drug effectiveness detection method and device based on multi-modal fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310203047.7A CN116825382A (en) | 2023-02-23 | 2023-02-23 | Epileptic drug effectiveness detection method and device based on multi-modal fusion |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116825382A true CN116825382A (en) | 2023-09-29 |
Family
ID=88139992
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310203047.7A Pending CN116825382A (en) | 2023-02-23 | 2023-02-23 | Epileptic drug effectiveness detection method and device based on multi-modal fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116825382A (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180211010A1 (en) * | 2017-01-23 | 2018-07-26 | Ucb Biopharma Sprl | Method and system for predicting refractory epilepsy status |
| CN112992377A (en) * | 2021-03-03 | 2021-06-18 | 深圳市儿童医院 | Method, device, terminal and storage medium for generating drug treatment result prediction model |
| CN113516641A (en) * | 2021-07-06 | 2021-10-19 | 西安电子科技大学 | End-to-end brain image data processing method and device based on deep learning |
| CN114711717A (en) * | 2022-03-10 | 2022-07-08 | 中国科学院深圳先进技术研究院 | Epilepsy drug treatment outcome prediction method and device based on multi-modal imaging omics |
| CN115249539A (en) * | 2022-01-27 | 2022-10-28 | 云南师范大学 | Multi-modal small sample depression prediction model construction method |
-
2023
- 2023-02-23 CN CN202310203047.7A patent/CN116825382A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180211010A1 (en) * | 2017-01-23 | 2018-07-26 | Ucb Biopharma Sprl | Method and system for predicting refractory epilepsy status |
| CN112992377A (en) * | 2021-03-03 | 2021-06-18 | 深圳市儿童医院 | Method, device, terminal and storage medium for generating drug treatment result prediction model |
| CN113516641A (en) * | 2021-07-06 | 2021-10-19 | 西安电子科技大学 | End-to-end brain image data processing method and device based on deep learning |
| CN115249539A (en) * | 2022-01-27 | 2022-10-28 | 云南师范大学 | Multi-modal small sample depression prediction model construction method |
| CN114711717A (en) * | 2022-03-10 | 2022-07-08 | 中国科学院深圳先进技术研究院 | Epilepsy drug treatment outcome prediction method and device based on multi-modal imaging omics |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Brand et al. | Joint multi-modal longitudinal regression and classification for alzheimer’s disease prediction | |
| EP4005498A1 (en) | Information processing device, program, learned model, diagnostic assistance device, learning device, and method for generating prediction model | |
| WO2018204311A1 (en) | Deep learning architecture for cognitive examination subscore trajectory prediction in alzheimer's disease | |
| Nisha et al. | Applications of deep learning in biomedical engineering | |
| US20200380339A1 (en) | Integrated neural networks for determining protocol configurations | |
| CN111063442B (en) | Brain disease process prediction method and system based on weak supervision multitask matrix completion | |
| WO2012149380A1 (en) | Apparatus, system and methods for assessing drug efficacy using holistic analysis and visualization of pharmacological data | |
| Al-Adhaileh | Diagnosis and classification of Alzheimer's disease by using a convolution neural network algorithm | |
| CN111951265A (en) | Cerebral infarction scoring method and device based on cerebral CT image, computer equipment and storage medium | |
| Gleichgerrcht et al. | Radiological identification of temporal lobe epilepsy using artificial intelligence: a feasibility study | |
| Kang et al. | Study design features that improve effect sizes in cross-sectional and longitudinal brain-wide association studies | |
| Baskar et al. | An Accurate Prediction and Diagnosis of Alzheimer’s Disease using Deep Learning | |
| Bi et al. | Explainable and programmable hypergraph convolutional network for imaging genetics data fusion | |
| CN111160441B (en) | Classification method, computer device, and storage medium | |
| Zhou et al. | Attentive deep canonical correlation analysis for diagnosing alzheimer’s disease using multimodal imaging genetics | |
| Bhandarkar et al. | Deep learning based computer aided diagnosis of Alzheimer’s disease: a snapshot of last 5 years, gaps, and future directions | |
| CN116825382A (en) | Epileptic drug effectiveness detection method and device based on multi-modal fusion | |
| US20220223231A1 (en) | Systems and Methods for Improved Prognostics in Medical Imaging | |
| CN114628034A (en) | Alzheimer's disease assessment method, system, device and storage medium | |
| Subasi et al. | Alzheimer’s disease detection using artificial intelligence | |
| Sharma et al. | Predicting the Cerebral Blood Flow Change Condition during Brain Strokes using Feature Fusion of FMRI Images and Clinical Features | |
| CN112568872A (en) | Brain age fusion prediction method based on MRI (magnetic resonance imaging) image and blood biochemical indexes | |
| Sharma et al. | Early detection and diagnosis using deep learning | |
| Bretones et al. | Automatic identification of schizophrenia employing EEG records analyzed with deep learning algorithms | |
| Olaniyan et al. | Clinical applications of deep learning in neurology and its enhancements with future predictions |
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 |