CN117894475A - Prediction device for cognitive function result of patient suffering from atrial fibrillation based on neural network - Google Patents
Prediction device for cognitive function result of patient suffering from atrial fibrillation based on neural network Download PDFInfo
- Publication number
- CN117894475A CN117894475A CN202410108284.XA CN202410108284A CN117894475A CN 117894475 A CN117894475 A CN 117894475A CN 202410108284 A CN202410108284 A CN 202410108284A CN 117894475 A CN117894475 A CN 117894475A
- Authority
- CN
- China
- Prior art keywords
- feature
- characteristic
- data
- feature extraction
- atrial fibrillation
- 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
- 206010003658 Atrial Fibrillation Diseases 0.000 title claims abstract description 85
- 230000003920 cognitive function Effects 0.000 title claims abstract description 79
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 48
- 238000000605 extraction Methods 0.000 claims abstract description 127
- 238000012545 processing Methods 0.000 claims abstract description 56
- 230000000007 visual effect Effects 0.000 claims abstract description 50
- 238000012549 training Methods 0.000 claims description 57
- 239000013598 vector Substances 0.000 claims description 38
- 230000015654 memory Effects 0.000 claims description 29
- 238000012360 testing method Methods 0.000 claims description 18
- 238000013507 mapping Methods 0.000 claims description 12
- 206010019280 Heart failures Diseases 0.000 claims description 8
- 201000002859 sleep apnea Diseases 0.000 claims description 8
- 230000035488 systolic blood pressure Effects 0.000 claims description 8
- 238000004891 communication Methods 0.000 claims description 4
- 238000012216 screening Methods 0.000 abstract description 8
- 238000013075 data extraction Methods 0.000 abstract 1
- 238000000034 method Methods 0.000 description 23
- 208000010877 cognitive disease Diseases 0.000 description 13
- 238000011156 evaluation Methods 0.000 description 9
- 230000001149 cognitive effect Effects 0.000 description 7
- 238000010586 diagram Methods 0.000 description 7
- 230000006870 function Effects 0.000 description 7
- 230000008569 process Effects 0.000 description 7
- 230000009286 beneficial effect Effects 0.000 description 4
- 238000013527 convolutional neural network Methods 0.000 description 3
- 102100036475 Alanine aminotransferase 1 Human genes 0.000 description 2
- 108010082126 Alanine transaminase Proteins 0.000 description 2
- 108010003415 Aspartate Aminotransferases Proteins 0.000 description 2
- 102000004625 Aspartate Aminotransferases Human genes 0.000 description 2
- 208000037273 Pathologic Processes Diseases 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 2
- 210000004556 brain Anatomy 0.000 description 2
- DDRJAANPRJIHGJ-UHFFFAOYSA-N creatinine Chemical compound CN1CC(=O)NC1=N DDRJAANPRJIHGJ-UHFFFAOYSA-N 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 230000009054 pathological process Effects 0.000 description 2
- 239000007787 solid Substances 0.000 description 2
- 238000004148 unit process Methods 0.000 description 2
- 208000000044 Amnesia Diseases 0.000 description 1
- 208000028698 Cognitive impairment Diseases 0.000 description 1
- 208000031226 Hyperlipidaemia Diseases 0.000 description 1
- 208000026139 Memory disease Diseases 0.000 description 1
- 208000005764 Peripheral Arterial Disease Diseases 0.000 description 1
- 208000030831 Peripheral arterial occlusive disease Diseases 0.000 description 1
- 230000007488 abnormal function Effects 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 239000003146 anticoagulant agent Substances 0.000 description 1
- 229940127219 anticoagulant drug Drugs 0.000 description 1
- 201000007201 aphasia Diseases 0.000 description 1
- 210000004204 blood vessel Anatomy 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 208000026106 cerebrovascular disease Diseases 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 208000029078 coronary artery disease Diseases 0.000 description 1
- 229940109239 creatinine Drugs 0.000 description 1
- 230000006735 deficit Effects 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 206010012601 diabetes mellitus Diseases 0.000 description 1
- 230000035487 diastolic blood pressure Effects 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 230000035622 drinking Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000003902 lesion Effects 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 230000006984 memory degeneration Effects 0.000 description 1
- 230000007074 memory dysfunction Effects 0.000 description 1
- 206010027175 memory impairment Diseases 0.000 description 1
- 208000023060 memory loss Diseases 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 210000002966 serum Anatomy 0.000 description 1
- 230000000391 smoking effect Effects 0.000 description 1
- 230000002861 ventricular Effects 0.000 description 1
Landscapes
- Image Analysis (AREA)
Abstract
The invention relates to the technical field of data processing, and discloses a prediction device, equipment and medium for cognitive function results of patients suffering from atrial fibrillation based on a neural network. The device comprises: the acquisition module is used for acquiring fundus pictures and characteristic data of the target object; the first feature extraction module is used for carrying out feature extraction on the fundus image by utilizing a pre-trained visual feature extraction model to obtain first feature information; the second feature extraction module is used for extracting the feature data by utilizing a pre-trained feature data extraction model to obtain second feature information; and the prediction module is used for predicting the cognitive function of the patient suffering from atrial fibrillation based on the first characteristic information and the second characteristic information. According to the invention, the cognitive function result of the patient suffering from atrial fibrillation can be predicted only by the fundus picture and the characteristic data of the target object, a special doctor and a special instrument are not needed, and the purpose of large-scale screening can be achieved while a large amount of time is saved.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to a prediction device, equipment and medium of cognitive function results of patients suffering from atrial fibrillation based on a neural network.
Background
Cognitive dysfunction refers to a pathological process that involves abnormal functioning in advanced intelligent processes of the brain, such as learning, memory, and thinking judgment, causing severe learning and memory impairment, accompanied by changes such as loss of speech, disuse, disrecognition, or disuse. Atrial fibrillation, however, can cause a higher probability of cognitive dysfunction.
In order to screen the cognitive dysfunction of the patient suffering from atrial fibrillation, currently, the patient is often required to fill in a cognitive evaluation scale under the guidance of a professional doctor, and then the professional doctor evaluates the cognitive function state of the patient by calculating the score of the evaluation scale to obtain a cognitive dysfunction result.
However, the method that the professional doctor guides the patient to fill in the cognitive evaluation scale and then calculates to determine the cognitive dysfunction of the atrial fibrillation patient consumes long time, and the professional doctor needs to participate in the whole course, so that the method is not suitable for large-scale screening.
Disclosure of Invention
In view of the above, the invention provides a prediction device, equipment and medium for cognitive dysfunction of patients suffering from atrial fibrillation based on a neural network, so as to solve the problems that a mode of determining the cognitive dysfunction of the patients suffering from atrial fibrillation through a cognitive evaluation scale consumes long time, and doctors needing professional training participate in the whole process, and the prediction device, the equipment and the medium are not suitable for large-scale screening.
In a first aspect, the present invention provides a neural network-based prediction apparatus for cognitive function results in patients with atrial fibrillation, the apparatus comprising: the acquisition module is used for acquiring fundus pictures and characteristic data of the target object; the first feature extraction module is used for carrying out feature extraction on the fundus image by utilizing a pre-trained visual feature extraction model to obtain first feature information; the second feature extraction module is used for carrying out feature extraction on the feature data by utilizing a pre-trained feature extraction model to obtain second feature information; and the prediction module is used for predicting the cognitive function result of the patient suffering from atrial fibrillation based on the first characteristic information and the second characteristic information.
According to the prediction device for the cognitive function result of the patient suffering from atrial fibrillation based on the neural network, the fundus picture and the feature data of the target object are obtained, the feature data are subjected to feature extraction through the feature extraction model of the feature data trained in advance to obtain second feature information, the fundus picture is subjected to feature extraction through the visual feature extraction model trained in advance to obtain first feature information, and then the cognitive function result of the patient suffering from atrial fibrillation is obtained through the first feature information and the second feature information. Compared with the mode of determining the cognitive function result of the patient suffering from atrial fibrillation through the scanning of an electrocardiogram, the cognitive function result of the patient suffering from atrial fibrillation can be predicted only by the fundus picture and the characteristic data of the target object, a special doctor and a special instrument are not needed, and the aim of large-scale screening is fulfilled while a large amount of time is saved.
In an alternative embodiment, the prediction module includes: the splicing unit is used for splicing the first characteristic information and the second characteristic information into target characteristic information; and the prediction unit is used for predicting the cognitive function result of the patient suffering from atrial fibrillation based on the target characteristic information and the full-connection layer.
According to the prediction device for the cognitive function result of the patient with atrial fibrillation based on the neural network, the first characteristic information and the second characteristic information are spliced to form the target characteristic information, so that on one hand, the existing characteristic information can be fully utilized, the influence of characteristic deletion on the performance of the model is reduced, and on the other hand, different characteristics are spliced, more information sources can be introduced, and therefore robustness of the model is improved, and tolerance of the model to noise and abnormal values is improved. In addition, the accuracy of the cognitive function result of the patients suffering from atrial fibrillation is further improved by utilizing the strong characteristic representation capability of the fully connected layer.
In an alternative embodiment, the splicing unit comprises: the first acquisition subunit is used for acquiring each first feature vector in the first feature information and the position corresponding to the first feature vector; the second obtaining subunit is used for obtaining each second feature vector in the second feature information and the position corresponding to the second feature vector; and the splicing subunit is used for respectively splicing the first characteristic vector and the second characteristic vector which correspond to the positions to obtain target characteristic information.
According to the prediction device for the cognitive function result of the patient with atrial fibrillation based on the neural network, the two characteristics possibly have different characteristics and information, and the two characteristics can be fused together through splicing to form more comprehensive characteristic information, so that the prediction device is beneficial to capturing more data characteristics and modes and improving generalization capability.
In an alternative embodiment, the second feature extraction module includes: an input unit for receiving the characteristic data; the first processing unit is used for carrying out batch standardization processing on the characteristic data to obtain first characteristic data; the second processing unit is used for carrying out weighting processing on the first characteristic data to obtain second characteristic data; the mapping unit is used for carrying out mapping processing on the second characteristic data to obtain third characteristic data; the generalization unit is used for processing the third characteristic data to obtain fourth characteristic data; the third processing unit is used for carrying out batch standardization processing on the fourth characteristic data to obtain fifth characteristic data; and the fourth processing unit is used for carrying out weighting processing on the fifth characteristic data to obtain second characteristic information.
According to the prediction device for the cognitive function result of the patient with atrial fibrillation based on the neural network, the scale difference between the feature data can be eliminated through batch standardization processing of the first processing unit and the third processing unit, the adaptability of the model to the features of different scales is improved, the representation capacity of the feature data can be enhanced through weighting processing of the second processing unit and the fourth processing unit, the generalization performance is improved, mapping processing can be carried out on the second feature data through the mapping unit, and therefore the device is beneficial to extracting more representative features and improving the classification accuracy of the data; the generalization unit processes the third characteristic data, so that generalization capability of the model to unknown data can be enhanced, and robustness of the model can be improved.
In an alternative embodiment, the apparatus further comprises: the test data set acquisition module is used for acquiring a test data set; the first feature extraction module is used for detecting the visual feature extraction model based on the test data set and generating an accuracy result; the second feature extraction module is used for detecting whether the accuracy result meets a preset accuracy result or not; the first judging module is used for judging that the visual feature extraction model is qualified if the accuracy result meets the preset accuracy result; and the second judging module is used for judging that the visual characteristic extraction model is unqualified if the accuracy rate result does not meet the preset accuracy rate result.
According to the prediction device for the cognitive function result of the patient with atrial fibrillation based on the neural network, which is provided by the embodiment, the accuracy of the visual characteristic extraction model can be detected through the test data set, so that the accuracy of the visual characteristic extraction model is improved.
In an alternative embodiment, the apparatus further comprises: the first training data set acquisition module is used for acquiring a first training data set; wherein the first training dataset comprises a historical fundus picture; the first model training module is used for carrying out model training on a preset first model based on the first training data, and constructing a visual feature extraction model.
According to the prediction device for the cognitive function result of the patient suffering from atrial fibrillation based on the neural network, the first model can be subjected to model training through the first training data set, and the visual feature extraction model is constructed, so that feature extraction can be carried out on the bottom image data through the visual feature extraction model.
In an alternative embodiment, the apparatus further comprises: the second training data set acquisition module is used for acquiring a second training data set; wherein the second training data set comprises at least: education level, age, systolic blood pressure, history of heart failure, body mass index, and presence or absence of sleep apnea syndrome; the second model training module is used for carrying out model training on a preset second model based on second training data and constructing a characteristic data characteristic extraction model.
According to the prediction device for the cognitive function result of the patient suffering from atrial fibrillation based on the neural network, which is provided by the embodiment, model training is performed on the preset second model at least through education degree, age, systolic pressure, heart failure history, body quality index and whether sleep apnea syndrome exists, and the feature data feature extraction model is constructed, so that the richness of the feature data feature extraction model can be improved, and the feature data feature extraction model can be guaranteed to perform feature extraction on feature data.
In an alternative embodiment, the second feature information and the first feature information are each 512-dimensional feature vectors.
According to the prediction device for the cognitive function result of the patient suffering from atrial fibrillation based on the neural network, which is provided by the embodiment, the characteristic of data can be better represented through the 512-dimensional characteristic vector, and more abundant information is provided for the model, so that the classification accuracy and performance of the model are improved.
In a second aspect, the present invention provides a computer device comprising: the memory and the processor are in communication connection, the memory stores computer instructions, and the processor executes the computer instructions to execute the following steps:
acquiring fundus pictures and characteristic data of a target object;
Performing feature extraction on the fundus image by using a pre-trained visual feature extraction model to obtain first feature information;
performing feature extraction on the feature data by using a pre-trained feature data feature extraction model to obtain second feature information;
The cognitive function result of the patient with atrial fibrillation is predicted based on the first characteristic information and the second characteristic information.
In a fourth aspect, the present invention provides a computer-readable storage medium having stored thereon computer instructions for causing a computer to perform the steps of:
acquiring fundus pictures and characteristic data of a target object;
Performing feature extraction on the fundus image by using a pre-trained visual feature extraction model to obtain first feature information;
performing feature extraction on the feature data by using a pre-trained feature data feature extraction model to obtain second feature information;
The cognitive function result of the patient with atrial fibrillation is predicted based on the first characteristic information and the second characteristic information.
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 needed in the description of the embodiments or the prior art will be briefly described, and it is obvious that the drawings in the description below are some embodiments of the present invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a block diagram of a neural network-based prediction apparatus for cognitive function results in patients with atrial fibrillation in accordance with an embodiment of the present invention;
FIG. 2 is a schematic diagram of a neural network-based prediction apparatus for cognitive function results in patients with atrial fibrillation according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a neural network-based prediction apparatus for cognitive function results in patients with atrial fibrillation according to an embodiment of the present invention;
FIG. 4 is a flow chart of a method for predicting cognitive function outcome in an atrial fibrillation patient based on a neural network in accordance with an embodiment of the present invention;
Fig. 5 is a schematic diagram of a hardware structure of a computer device according to an embodiment of the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Based on the related art, cognitive dysfunction refers to the pathological process of abnormal functions related to advanced intelligent brain processing processes such as learning, memory and thinking judgment, so that serious learning and memory dysfunction is caused, and changes such as aphasia, disuse or disuse are accompanied. Atrial fibrillation, however, can cause a higher probability of cognitive dysfunction.
In order to screen the cognitive dysfunction of the patient suffering from atrial fibrillation, currently, the patient is often required to fill in a cognitive evaluation scale under the guidance of a professional doctor, and then the professional doctor evaluates the cognitive function state of the patient by calculating the score of the evaluation scale to obtain a cognitive dysfunction result.
However, the method that the professional doctor guides the patient to fill in the cognitive evaluation scale and then calculates to determine the cognitive dysfunction of the atrial fibrillation patient consumes long time, and the professional doctor needs to participate in the whole course, so that the method is not suitable for large-scale screening.
Based on this, the prediction apparatus for a cognitive function result of an atrial fibrillation patient based on a neural network provided in this embodiment obtains second feature information by obtaining a fundus image and feature data of a target object, performing feature extraction on the feature data through a feature extraction model of the feature data trained in advance, performing feature extraction on the fundus image through a visual feature extraction model trained in advance, obtaining first feature information, and then performing cognitive function result on the atrial fibrillation patient through the first feature information and the second feature information. Compared with the mode of determining the cognitive function result of the patient suffering from atrial fibrillation through the scanning of an electrocardiogram, the cognitive function result of the patient suffering from atrial fibrillation can be predicted only by the fundus picture and the characteristic data of the target object, a special doctor and a special instrument are not needed, and the aim of large-scale screening is fulfilled while a large amount of time is saved.
In accordance with an embodiment of the present invention, there is provided an embodiment of a prediction apparatus for cognitive function results of an atrial fibrillation patient based on a neural network, and it should be noted that the blocks shown in the structural block diagram of the drawings may be stored in a readable storage medium such as a set of computers, and the functions of the respective blocks are performed by a computer system.
In this embodiment, a prediction apparatus for a cognitive function result of an atrial fibrillation patient based on a neural network is provided, and fig. 1 is a block diagram of a prediction apparatus for a cognitive function result of an atrial fibrillation patient based on a neural network according to an embodiment of the present invention, as shown in fig. 1, the apparatus includes:
An acquisition module 101 for acquiring fundus images and feature data of the target object.
The target object may be used to characterize a user who detects the outcome of cognitive function in patients with atrial fibrillation. The target object may be a pregnant woman, an elderly person, a young person, or the like, and is not particularly limited herein. The fundus picture may be used to characterize a picture including the fundus of the target object. The fundus image may be a fundus blood vessel image, a fundus photographic image, or the like, and is not particularly limited herein.
The feature data may be used to characterize clinical data, body information, etc. of the target object. Wherein, the characteristic data may include smoking history, drinking history, education level, age, sex, body Mass Index (BMI), systolic pressure, diastolic pressure, cerebrovascular disease history, coronary heart disease history, peripheral arterial disease history, high pressure disease history, diabetes history, heart failure history, hyperlipidemia sleep apnea syndrome history, anticoagulant, ventricular rate, glutamic-oxaloacetic transaminase (ASPARTATE AMINOTRANSFERASE, AST), glutamic-pyruvic transaminase (Alanine Aminotransferase, ALT), serum creatinine level, etc., without specific limitation.
Specifically, the fundus image of the target object may be input to the prediction device of the cognitive function result of the patient with atrial fibrillation based on the neural network by means of a mouse click, a keyboard selection or a touch screen, etc., which is not limited herein. The feature data may be obtained from a preset feature database, or may be input to a prediction device of cognitive function results of the patient with atrial fibrillation based on a neural network by means of mouse clicking, keyboard selecting or touch screen, etc., which is not limited herein.
The first feature extraction module 102 is configured to perform feature extraction on the fundus image by using a pre-trained visual feature extraction model, so as to obtain first feature information.
The first characteristic information may be used to characterize characteristic information related to cognitive impairment. Wherein the first characteristic information may include: memory loss, distraction, language impairment, etc., are not particularly limited herein. According to different characteristic information of the target object, the corresponding first characteristic information can be obtained through a pre-trained characteristic data characteristic extraction model. For example: the fundus image of the target object a and the fundus image of the target object B are input to the feature data feature extraction model trained in advance, and then output as the first feature information A1 of the target object a and the first feature information B1 of the target object B.
And the second feature extraction module 103 is configured to perform feature extraction on the feature data by using a feature extraction model of the feature data trained in advance, so as to obtain second feature information.
The second characteristic information may be used to characterize clinical data and body information corresponding to the target object. Wherein different target objects may correspond to different second characteristic information. Specifically, according to different feature data of the target object, the corresponding second feature information can be obtained through a feature extraction model of the feature data trained in advance. For example: the feature data of the target object a and the feature data of the target object B are input to the pre-trained visual feature extraction model, and then the second feature information A2 of the target object a and the second feature information B2 of the target object B are output. Wherein the second characteristic information A2 may include hypomnesis, inattention; the second characteristic information B2 may include: attention is not focused on language handicaps.
And the prediction module is used for predicting the cognitive function result of the patient suffering from atrial fibrillation based on the first characteristic information and the second characteristic information.
After the first characteristic information and the second characteristic information are obtained, the cognitive function result of the patient suffering from atrial fibrillation can be obtained by carrying out a weighting summation on the first characteristic information and the second characteristic information. The manner in which the outcome of cognitive function in an atrial fibrillation patient is specifically determined is described in detail below.
It should be noted that, the visual feature extraction model and the feature data feature extraction model may be constructed with reference to a convolutional neural network (Convolutional Neural Networks, CNN).
Referring to fig. 2, in fig. 2, the fundus image is feature-extracted through a visual feature extraction model (i.e., the visual model in fig. 2) and a series of feature data is feature-extracted through a feature data feature extraction model (i.e., the meta model in fig. 2) by using a fundus image and a series of feature data, so as to obtain a cognitive function result of the patient suffering from atrial fibrillation.
According to the prediction device for the cognitive function result of the patient suffering from atrial fibrillation based on the neural network, the fundus picture and the feature data of the target object are obtained, the feature extraction model of the feature data is trained in advance to conduct feature extraction on the data, second feature information is obtained, the feature extraction model of the visual feature extraction model is trained in advance to conduct feature extraction on the fundus picture, first feature information is obtained, and then the cognitive function result of the patient suffering from atrial fibrillation is obtained through the first feature information and the second feature information. Compared with the mode of determining the cognitive function result of the patient suffering from atrial fibrillation through the scanning of an electrocardiogram, the cognitive function result of the patient suffering from atrial fibrillation can be predicted only by the fundus picture and the characteristic data of the target object, a special doctor and a special instrument are not needed, and the aim of large-scale screening is fulfilled while a large amount of time is saved.
In some preferred embodiments, the prediction module includes:
and the splicing unit is used for splicing the first characteristic information and the second characteristic information into target characteristic information.
The first feature information and the second feature information are 512-dimensional feature vectors. Specifically, operations such as data cleaning, data conversion, data standardization and the like can be performed on the first feature information and the second feature information, so that subsequent feature stitching is facilitated. The text features and the image features may then be stitched, for example, according to the manner in which the data types are stitched. The specific manner of stitching is described in detail below.
Specifically, the above-mentioned concatenation unit includes:
The first obtaining subunit is configured to obtain each first feature vector in the first feature information and a position corresponding to the first feature vector.
The second obtaining subunit is configured to obtain each second feature vector in the second feature information and a position corresponding to the second feature vector.
After the first feature information and the second feature information are acquired, each first feature vector in the first feature information and a position corresponding to the first feature vector may be acquired, and each second feature vector in the second feature information and a position corresponding to the second feature vector may be acquired. Since the first feature information and the second feature information are 512-dimensional feature vectors, feature dimensions of the first feature information and the second feature information are consistent.
And the splicing subunit is used for respectively splicing the first characteristic vector and the second characteristic vector which correspond to the positions to obtain target characteristic information.
Because the number of the first characteristic information and the second characteristic information are consistent, the characteristic vectors of the corresponding positions can be cross multiplied to obtain new vectors corresponding to the positions, and after the characteristic vectors of all the positions are multiplied, the target characteristic information can be obtained. The target feature information is 1024-dimensional feature vectors. For example: the first characteristic information includes: a1, a2, a3...an, the second characteristic information comprises: b1, b2, b3...bn, then the target characteristic information may include: a1×b1, a2×b2, a3× b3..
And the prediction unit is used for predicting the cognitive function result of the patient suffering from atrial fibrillation based on the target characteristic information and the full-connection layer.
And then, carrying out weight summation processing on the target characteristic information through the full-connection layer to obtain the cognitive function result of the patient suffering from atrial fibrillation. For example: the target feature information may include: a1×b1, a2×b2, a3× b3...an×bn, the corresponding weights are m1, m2, m 3..mn, then the cognitive function results of patients with atrial fibrillation can be obtained by cross multiplying the vectors and the corresponding weights and summing the results after all the multiplication.
According to the prediction device for the cognitive function result of the patient with atrial fibrillation based on the neural network, the two characteristics possibly have different characteristics and information of each corresponding mode, and the two characteristics can be fused together through splicing to form more comprehensive characteristic information, so that the prediction device is beneficial to capturing more abundant data characteristics and wider algorithm modes and improving the generalization capability of a model.
In an alternative embodiment, the second feature extraction module 103 includes:
And an input unit for receiving the characteristic data.
And the first processing unit is used for carrying out batch standardization processing on the characteristic data to obtain first characteristic data.
And the second processing unit is used for carrying out weighting processing on the first characteristic data to obtain second characteristic data.
And the mapping unit is used for carrying out mapping processing on the second characteristic data to obtain third characteristic data.
And the generalization unit is used for processing the third characteristic data to obtain fourth characteristic data.
And the third processing unit is used for carrying out batch standardization processing on the fourth characteristic data to obtain fifth characteristic data.
And the fourth processing unit is used for carrying out weighting processing on the fifth characteristic data to obtain second characteristic information.
As shown in fig. 3, after the feature data is acquired, the first feature data may be obtained by inputting the feature data (i.e., metadata in fig. 3) through an input unit and then performing batch normalization processing on the feature data. Wherein, since feature scales of feature data may be different, direct input of a model may cause instability in the learning process. The batch normalization process helps to improve the stability and convergence speed of the model by normalizing the features of each sample so that they are uniform in scale. And then carrying out weighted summation processing on the first characteristic data through the full connection layer to obtain second characteristic data, carrying out mapping processing on the second characteristic data through the ReLU activation function layer to obtain third characteristic data, carrying out batch standardization processing on the third characteristic data to obtain fourth characteristic data, and carrying out weighted processing on the fourth characteristic data through the full connection layer to obtain second characteristic information (namely 512-dimensional characteristics in fig. 3).
According to the prediction device for the cognitive function result of the patient with atrial fibrillation based on the neural network, the scale difference between the feature data can be eliminated through batch standardization processing of the first processing unit and the third processing unit, the adaptability of the model to the features of different scales is improved, the representation capacity of the feature data can be enhanced through weighting processing of the second processing unit and the fourth processing unit, the generalization performance is improved, mapping processing can be carried out on the second feature data through the mapping unit, and therefore the device is beneficial to extracting more representative features and improving the classification accuracy of the data; the generalization unit processes the third characteristic data, so that generalization capability of the model to unknown data can be enhanced, and robustness of the model can be improved.
In a preferred embodiment, the apparatus further comprises:
and the test data set acquisition module is used for acquiring the test data set.
The test dataset may be used to characterize a dataset that verifies the accuracy of the visual feature extraction model. Wherein the test dataset may comprise: the history fundus picture set and the like are not particularly limited herein. Specifically, the test data set may be obtained by querying a history of medical visits, or may be obtained by performing an actual test on site on the user, which is not particularly limited herein.
The first feature extraction module 102 is configured to perform feature extraction on the visual feature extraction model based on the test dataset, and generate an accuracy result.
The accuracy results may be used to characterize the predictive capabilities of the visual feature extraction model. Specifically, in the visual feature extraction model, the test dataset may be taken as input and the accuracy result as output.
The second feature extraction module 103 is configured to detect whether the accuracy result meets a preset accuracy result.
And the first judging module is used for judging that the visual characteristic extraction model is qualified if the accuracy result meets the preset accuracy result.
And the second judging module is used for judging that the visual characteristic extraction model is unqualified if the accuracy rate result does not meet the preset accuracy rate result.
The preset accuracy result can accurately extract the first feature information for the visual features. Specifically, after the accuracy result is obtained, whether the visual feature extraction model is qualified can be determined by detecting whether the accuracy result meets a preset accuracy result. And if the accuracy rate result is that the first characteristic information is matched with the preset characteristic information, judging that the visual characteristic extraction model is qualified. And if the accuracy result is that the first characteristic information is not matched with the preset characteristic information, judging that the visual characteristic extraction model is unqualified.
Preferably ResNet-50 are used as visual feature extraction models for lesion feature extraction. ResNet (Residual Neural Network) by KAIMING HE et al of microsoft institute, 152 layers of neural networks were successfully trained by using ResNet Unit and champions were taken in ILSVRC in 2015 competition, while the reference number was low as compared to VGGNet, and the effect was very prominent. ResNet can accelerate the training of the neural network very quickly, and the accuracy of the model is improved greatly.
Preferably, the loss function used by the multimodal model is as follows:
Wherein N cls is the number of samples, and L cls is a cross entropy loss function, which is used to calculate a loss value for the predicted class probability and the true class label p i, and the specific calculation mode of the cross entropy loss is as follows:
Wherein L i is a loss value.
According to the prediction device for the cognitive function result of the patient with atrial fibrillation based on the neural network, which is provided by the embodiment, the accuracy of the visual characteristic extraction model can be detected through the test data set, so that the accuracy of the visual characteristic extraction model is improved.
In a preferred embodiment, the apparatus further comprises:
the first training data set acquisition module is used for acquiring a first training data set; wherein the first training data set comprises a historical fundus picture.
The first model training module is used for carrying out model training on a preset first model based on the first training data, and constructing a visual feature extraction model.
The training dataset may be used to characterize historical fundus pictures of different users from the historical collection. Specifically, model training is carried out on a preset first model through historical fundus pictures of each user, a visual characteristic extraction model is constructed,
According to the prediction device for the cognitive function result of the patient suffering from atrial fibrillation based on the neural network, the first model can be subjected to model training through the first training data set, and the visual feature extraction model is constructed, so that feature extraction can be carried out on the bottom image data through the visual feature extraction model.
In a preferred embodiment, the apparatus further comprises:
The second training data set acquisition module is used for acquiring a second training data set; wherein the second training data set comprises at least: education level, age, systolic blood pressure, history of heart failure, body mass index, and presence or absence of sleep apnea syndrome.
The second model training module is used for carrying out model training on a preset second model based on second training data and constructing a characteristic data characteristic extraction model.
The second training data set comprises: education levels, ages, systolic blood pressure, history of heart failure, body mass index, and presence or absence of sleep apnea syndrome for different users. Specifically, model training can be performed on a preset second model through the second training data set, and a feature data feature extraction model is constructed, so that feature extraction can be performed on the bottom-of-eye image data through the feature data feature extraction model.
According to the prediction device for the cognitive function result of the patient suffering from atrial fibrillation based on the neural network, which is provided by the embodiment, model training is performed on the preset second model at least through education degree, age, systolic pressure, heart failure history, body quality index and whether sleep apnea syndrome exists, and the feature data feature extraction model is constructed, so that the richness of the feature data feature extraction model can be improved, and the feature data feature extraction model can be guaranteed to perform feature extraction on feature data.
The prediction device of cognitive function results of patients with atrial fibrillation based on the neural network in this embodiment is presented in the form of functional units, where the functional units refer to ASIC (Application SPECIFIC INTEGRATED Circuit) circuits, processors and memories that execute one or more software or fixed programs, and/or other devices that can provide the above functions.
In accordance with an embodiment of the present invention, there is provided a neural network-based prediction method embodiment of cognitive function results in patients with atrial fibrillation, it being noted that the steps illustrated in the flowchart of the figures may be performed in a computer system, such as a set of computer-executable instructions, and that, although a logical sequence is illustrated in the flowchart, in some cases, the steps illustrated or described may be performed in a different order than that illustrated herein.
In this embodiment, a method for predicting a cognitive function result of an atrial fibrillation patient based on a neural network is provided, which may be used in the device for predicting a cognitive function result of an atrial fibrillation patient based on a neural network, and fig. 4 is a schematic flowchart of a method for predicting a cognitive function result of an atrial fibrillation patient based on a neural network according to an embodiment of the present invention, as shown in fig. 4, where the flowchart includes the following steps:
step S401, acquiring fundus images and feature data of a target object.
Step S402, feature extraction is performed on the fundus image by using a pre-trained visual feature extraction model, so as to obtain first feature information.
Step S403, feature extraction is carried out on the feature data by utilizing a pre-trained feature data feature extraction model, so as to obtain second feature information.
Step S404, predicting the cognitive function result of the patient suffering from atrial fibrillation based on the first characteristic information and the second characteristic information.
In an alternative embodiment, the step S404 includes:
And a step a1, splicing the first characteristic information and the second characteristic information into target characteristic information.
And a step a2, predicting the cognitive function result of the patient suffering from atrial fibrillation based on the target characteristic information and the fully-connected layer.
In an alternative embodiment, the step a1 includes:
step a11, each first feature vector in the first feature information and a position corresponding to the first feature vector are acquired.
Step a12, each second feature vector in the second feature information and the position corresponding to the second feature vector are acquired.
And a step a13, respectively splicing the first feature vector and the second feature vector which correspond to the positions to obtain target feature information.
In an alternative embodiment, the step S402 includes:
And b1, receiving the characteristic data.
And b2, carrying out batch standardization processing on the characteristic data to obtain first characteristic data.
And b3, weighting the first characteristic data to obtain second characteristic data.
And b4, mapping the second characteristic data to obtain third characteristic data.
And b5, processing the third characteristic data to obtain fourth characteristic data.
And b6, carrying out batch standardization processing on the fourth characteristic data to obtain fifth characteristic data.
And b7, weighting the fifth characteristic data to obtain second characteristic information.
In an alternative embodiment, the method further comprises:
step c1, acquiring a test data set.
And c2, detecting the visual characteristic extraction model based on the test data set, and generating an accuracy result.
And c3, detecting whether the accuracy result meets the preset accuracy result.
And c4, if the accuracy rate result meets the preset accuracy rate result, judging that the visual feature extraction model is qualified.
And c5, if the accuracy rate result does not meet the preset accuracy rate result, judging that the visual feature extraction model is unqualified.
In an alternative embodiment, the method further comprises:
Step d1, acquiring a first training data set; wherein the first training data set comprises a historical fundus picture.
And d2, performing model training on a preset first model based on the first training data, and constructing a visual feature extraction model.
In an alternative embodiment, the method further comprises:
Step e1, acquiring a second training data set; wherein the second training data set comprises at least: education level, age, systolic blood pressure, history of heart failure, body mass index, and presence or absence of sleep apnea syndrome.
And e2, performing model training on a preset second model based on the second training data, and constructing a feature data feature extraction model.
The above corresponding method embodiments are the same as further functional descriptions of the respective modules and units, and are not repeated here.
According to the prediction method of the cognitive function result of the patient suffering from the atrial fibrillation based on the neural network, the fundus picture and the feature data of the target object are obtained, the feature data are subjected to feature extraction through the feature data feature extraction model trained in advance to obtain second feature information, the fundus picture is subjected to feature extraction through the visual feature extraction model trained in advance to obtain first feature information, and then the cognitive function result of the patient suffering from the atrial fibrillation is predicted through the first feature information and the second feature information. Compared with the mode that a patient fills in a cognitive evaluation scale under the guidance of a professional doctor, and then the professional doctor scores the cognitive evaluation scale to finally determine the cognitive function result of the patient with atrial fibrillation, the cognitive function result of the patient with atrial fibrillation can be predicted only by the fundus picture and the characteristic data of a target object, the professional doctor and a special instrument are not needed, and the aim of large-scale screening is fulfilled while a large amount of time is saved.
The embodiment of the invention also provides computer equipment, which is provided with the prediction device of the cognitive function result of the patient with atrial fibrillation based on the neural network shown in the figure 1.
Referring to fig. 5, fig. 5 is a schematic structural diagram of a computer device according to an alternative embodiment of the present invention, as shown in fig. 5, the computer device includes: one or more processors 10, memory 20, and interfaces for connecting the various components, including high-speed interfaces and low-speed interfaces. The various components are communicatively coupled to each other using different buses and may be mounted on a common motherboard or in other manners as desired. The processor may process instructions executing within the computer device, including instructions stored in or on memory to display graphical information of the GUI on an external input/output device, such as a display device coupled to the interface. In some alternative embodiments, multiple processors and/or multiple buses may be used, if desired, along with multiple memories and multiple memories. Also, multiple computer devices may be connected, each providing a portion of the necessary operations (e.g., as a server array, a set of blade servers, or a multiprocessor system). One processor 10 is illustrated in fig. 5.
The processor 10 may be a central processor, a network processor, or a combination thereof. The processor 10 may further include a hardware chip, among others. The hardware chip may be an application specific integrated circuit, a programmable logic device, or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general-purpose array logic, or any combination thereof.
Wherein the memory 20 stores instructions executable by the at least one processor 10 to cause the at least one processor 10 to perform the methods shown in implementing the above embodiments.
The memory 20 may include a storage program area that may store an operating system, at least one application program required for functions, and a storage data area; the storage data area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include high-speed random access memory, and may also include non-transitory memory, such as at least one magnetic disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, memory 20 may optionally include memory located remotely from processor 10, which may be connected to the computer device via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
Memory 20 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk, or solid state disk; the memory 20 may also comprise a combination of the above types of memories.
The computer device further comprises input means 30 and output means 40. The processor 10, memory 20, input device 30, and output device 40 may be connected by a bus or other means, for example in fig. 5.
The input device 30 may receive input numeric or character information and generate key signal inputs related to user settings and function control of the computer apparatus, such as a touch screen, a keypad, a mouse, a trackpad, a touchpad, a pointer stick, one or more mouse buttons, a trackball, a joystick, and the like. The output means 40 may include a display device, auxiliary lighting means (e.g., LEDs), tactile feedback means (e.g., vibration motors), and the like. Such display devices include, but are not limited to, liquid crystal displays, light emitting diodes, displays and plasma displays. In some alternative implementations, the display device may be a touch screen.
The computer device also includes a communication interface for the computer device to communicate with other devices or communication networks.
The embodiments of the present invention also provide a computer readable storage medium, and the method according to the embodiments of the present invention described above may be implemented in hardware, firmware, or as a computer code which may be recorded on a storage medium, or as original stored in a remote storage medium or a non-transitory machine readable storage medium downloaded through a network and to be stored in a local storage medium, so that the method described herein may be stored on such software process on a storage medium using a general purpose computer, a special purpose processor, or programmable or special purpose hardware. The storage medium can be a magnetic disk, an optical disk, a read-only memory, a random access memory, a flash memory, a hard disk, a solid state disk or the like; further, the storage medium may also comprise a combination of memories of the kind described above. It will be appreciated that a computer, processor, microprocessor controller or programmable hardware includes a storage element that can store or receive software or computer code that, when accessed and executed by the computer, processor or hardware, implements the methods illustrated by the above embodiments.
Although embodiments of the present invention have been described in connection with the accompanying drawings, various modifications and variations may be made by those skilled in the art without departing from the spirit and scope of the invention, and such modifications and variations fall within the scope of the invention as defined by the appended claims.
Claims (10)
1. A neural network-based prediction apparatus for cognitive function results in patients with atrial fibrillation, comprising:
The acquisition module is used for acquiring fundus pictures and characteristic data of the target object;
the first feature extraction module is used for carrying out feature extraction on the fundus picture by utilizing a pre-trained visual feature extraction model to obtain first feature information;
the second feature extraction module is used for carrying out feature extraction on the feature data by utilizing a pre-trained feature data feature extraction model to obtain second feature information;
And the prediction module is used for predicting the cognitive function result of the patient suffering from atrial fibrillation based on the first characteristic information and the second characteristic information.
2. The neural network-based prediction apparatus of cognitive function outcome in patients with atrial fibrillation of claim 1, wherein the prediction module comprises:
the splicing unit is used for splicing the first characteristic information and the second characteristic information into target characteristic information;
And the prediction unit is used for predicting the cognitive function result of the atrial fibrillation patient based on the target characteristic information and the full-connection layer.
3. The neural network-based prediction apparatus of cognitive function results in patients with atrial fibrillation according to claim 2, wherein the stitching unit comprises:
The first acquisition subunit is used for acquiring each first feature vector in the first feature information and the position corresponding to the first feature vector;
The second obtaining subunit is used for obtaining each second feature vector in the second feature information and the position corresponding to the second feature vector;
And the splicing subunit is used for respectively splicing the first characteristic vector and the second characteristic vector which correspond to each other in position to obtain target characteristic information.
4. The neural network-based prediction apparatus of cognitive function outcome in an atrial fibrillation patient of claim 1, wherein the second feature extraction module comprises:
an input unit for receiving the characteristic data;
the first processing unit is used for carrying out batch standardization processing on the characteristic data to obtain first characteristic data;
the second processing unit is used for carrying out weighting processing on the first characteristic data to obtain second characteristic data;
the mapping unit is used for carrying out mapping processing on the second characteristic data to obtain third characteristic data;
The generalization unit is used for processing the third characteristic data to obtain fourth characteristic data;
The third processing unit is used for carrying out batch standardization processing on the fourth characteristic data to obtain fifth characteristic data;
and the fourth processing unit is used for carrying out weighting processing on the fifth characteristic data to obtain the second characteristic information.
5. The neural network-based prediction apparatus of cognitive function outcome in an atrial fibrillation patient of claim 1, further comprising:
the test data set acquisition module is used for acquiring a test data set;
The first feature extraction module is used for detecting the visual feature extraction model based on the test data set and generating an accuracy result;
The second feature extraction module is used for detecting whether the accuracy result meets a preset accuracy result or not;
The first judging module is used for judging that the visual feature extraction model is qualified if the accuracy result meets a preset accuracy result;
And the second judging module is used for judging that the visual feature extraction model is unqualified if the accuracy rate result does not meet the preset accuracy rate result.
6. The neural network-based prediction apparatus of cognitive function outcome in an atrial fibrillation patient of claim 1, further comprising:
the first training data set acquisition module is used for acquiring a first training data set; wherein the first training dataset comprises a historical fundus picture;
And the first model training module is used for carrying out model training on a preset first model based on the first training data and constructing the visual characteristic extraction model.
7. The neural network-based prediction apparatus of cognitive function outcome in an atrial fibrillation patient of claim 1, further comprising:
The second training data set acquisition module is used for acquiring a second training data set; wherein the second training data set comprises at least: education level, age, systolic blood pressure, history of heart failure, body mass index, and presence or absence of sleep apnea syndrome;
And the second model training module is used for carrying out model training on a preset second model based on the second training data and constructing the characteristic data characteristic extraction model.
8. The neural network-based prediction apparatus of cognitive function outcome in an atrial fibrillation patient of any one of claims 1-7, wherein the second feature information and the first feature information are each 512-dimensional feature vectors.
9. A computer device, comprising:
the device comprises a memory and a processor, wherein the memory and the processor are in communication connection, the memory stores computer instructions, and the processor executes the computer instructions so as to execute the following steps:
acquiring fundus pictures and characteristic data of a target object;
Performing feature extraction on the fundus picture by using a pre-trained visual feature extraction model to obtain first feature information;
Performing feature extraction on the feature data by using a pre-trained feature data feature extraction model to obtain second feature information;
The cognitive function result of the patient with atrial fibrillation is predicted based on the first characteristic information and the second characteristic information.
10. A computer-readable storage medium having stored thereon computer instructions for causing a computer to perform the steps of:
acquiring fundus pictures and characteristic data of a target object;
Performing feature extraction on the fundus picture by using a pre-trained visual feature extraction model to obtain first feature information;
Performing feature extraction on the feature data by using a pre-trained feature data feature extraction model to obtain second feature information;
The cognitive function result of the patient with atrial fibrillation is predicted based on the first characteristic information and the second characteristic information.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410108284.XA CN117894475A (en) | 2024-01-25 | 2024-01-25 | Prediction device for cognitive function result of patient suffering from atrial fibrillation based on neural network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410108284.XA CN117894475A (en) | 2024-01-25 | 2024-01-25 | Prediction device for cognitive function result of patient suffering from atrial fibrillation based on neural network |
Publications (1)
Publication Number | Publication Date |
---|---|
CN117894475A true CN117894475A (en) | 2024-04-16 |
Family
ID=90646646
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202410108284.XA Pending CN117894475A (en) | 2024-01-25 | 2024-01-25 | Prediction device for cognitive function result of patient suffering from atrial fibrillation based on neural network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117894475A (en) |
-
2024
- 2024-01-25 CN CN202410108284.XA patent/CN117894475A/en active Pending
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2019150813A1 (en) | Data processing device and method, recognition device, learning data storage device, machine learning device, and program | |
CN112329964B (en) | Method, device, equipment and storage medium for pushing information | |
CN111325745B (en) | Fracture region analysis method and device, electronic equipment and readable storage medium | |
CN108091372B (en) | Medical field mapping verification method and device | |
KR20170061222A (en) | The method for prediction health data value through generation of health data pattern and the apparatus thereof | |
JP2021529061A (en) | Image processing methods, electronic devices and storage media | |
Osman et al. | SOM-LWL method for identification of COVID-19 on chest X-rays | |
CN109191451B (en) | Abnormality detection method, apparatus, device, and medium | |
CN110838366A (en) | Method and device for predicting disease risk | |
Srivastava et al. | Medi-Assist: A Decision Tree based Chronic Diseases Detection Model | |
JP2014029644A (en) | Similar case retrieval device and similar case retrieval method | |
JP2020518050A (en) | Learning and applying contextual similarity between entities | |
CN114724716A (en) | Method, model training and apparatus for risk prediction of progression to type 2 diabetes | |
WO2018233520A1 (en) | Method and device for generating predicted image | |
CN111340794B (en) | Quantification method and device for coronary artery stenosis | |
US20240112329A1 (en) | Distinguishing a Disease State from a Non-Disease State in an Image | |
CN117271804A (en) | Method, device, equipment and medium for generating common disease feature knowledge base | |
CN111275121A (en) | Medical image processing method and device and electronic equipment | |
CN113220895A (en) | Information processing method and device based on reinforcement learning and terminal equipment | |
CN110414562B (en) | X-ray film classification method, device, terminal and storage medium | |
CN117894475A (en) | Prediction device for cognitive function result of patient suffering from atrial fibrillation based on neural network | |
CN115662619A (en) | Chronic disease health change prediction method and system based on big data | |
CN112397195A (en) | Method, apparatus, electronic device, and medium for generating physical examination model | |
TWI790479B (en) | Physiological status evaluation method and physiological status evaluation device | |
CN109192306A (en) | A kind of judgment means of diabetes, equipment and computer readable storage medium |
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 |