CN111161883A - Disease prediction system based on variational self-encoder and electronic equipment thereof - Google Patents
Disease prediction system based on variational self-encoder and electronic equipment thereof Download PDFInfo
- Publication number
- CN111161883A CN111161883A CN201911404700.6A CN201911404700A CN111161883A CN 111161883 A CN111161883 A CN 111161883A CN 201911404700 A CN201911404700 A CN 201911404700A CN 111161883 A CN111161883 A CN 111161883A
- Authority
- CN
- China
- Prior art keywords
- disease prediction
- encoder
- diagnosis
- layer
- prediction system
- 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
- 201000010099 disease Diseases 0.000 title claims abstract description 98
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 title claims abstract description 98
- 238000003745 diagnosis Methods 0.000 claims abstract description 99
- 238000007781 pre-processing Methods 0.000 claims abstract description 13
- 238000012545 processing Methods 0.000 claims abstract description 12
- 230000006870 function Effects 0.000 claims description 46
- 238000005070 sampling Methods 0.000 claims description 20
- 238000012549 training Methods 0.000 claims description 15
- 238000013507 mapping Methods 0.000 claims description 11
- 238000004422 calculation algorithm Methods 0.000 claims description 10
- 238000005457 optimization Methods 0.000 claims description 8
- 238000013528 artificial neural network Methods 0.000 claims description 7
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 5
- 230000004913 activation Effects 0.000 claims description 3
- 238000000034 method Methods 0.000 description 25
- 238000004891 communication Methods 0.000 description 10
- 238000003860 storage Methods 0.000 description 9
- 238000010586 diagram Methods 0.000 description 8
- 238000009826 distribution Methods 0.000 description 7
- 238000001914 filtration Methods 0.000 description 7
- 230000008569 process Effects 0.000 description 7
- 238000004590 computer program Methods 0.000 description 6
- 230000000694 effects Effects 0.000 description 6
- 208000020084 Bone disease Diseases 0.000 description 5
- 206010012601 diabetes mellitus Diseases 0.000 description 5
- 238000005065 mining Methods 0.000 description 4
- 238000012986 modification Methods 0.000 description 4
- 230000004048 modification Effects 0.000 description 4
- 230000003287 optical effect Effects 0.000 description 4
- 238000011160 research Methods 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- 239000002131 composite material Substances 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 239000004065 semiconductor Substances 0.000 description 3
- 206010019280 Heart failures Diseases 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 230000033001 locomotion Effects 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 101000822695 Clostridium perfringens (strain 13 / Type A) Small, acid-soluble spore protein C1 Proteins 0.000 description 1
- 101000655262 Clostridium perfringens (strain 13 / Type A) Small, acid-soluble spore protein C2 Proteins 0.000 description 1
- 206010061818 Disease progression Diseases 0.000 description 1
- 101000655256 Paraclostridium bifermentans Small, acid-soluble spore protein alpha Proteins 0.000 description 1
- 101000655264 Paraclostridium bifermentans Small, acid-soluble spore protein beta Proteins 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000011263 association rule mining technique Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 230000007177 brain activity Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 206010008118 cerebral infarction Diseases 0.000 description 1
- 208000026106 cerebrovascular disease Diseases 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000005750 disease progression Effects 0.000 description 1
- 230000005684 electric field Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000000474 nursing effect Effects 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 230000000644 propagated effect Effects 0.000 description 1
- 230000006798 recombination Effects 0.000 description 1
- 238000005215 recombination Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 235000019640 taste Nutrition 0.000 description 1
- 230000026683 transduction Effects 0.000 description 1
- 238000010361 transduction Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients
-
- 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
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Public Health (AREA)
- Medical Informatics (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Pathology (AREA)
- Epidemiology (AREA)
- Primary Health Care (AREA)
- Databases & Information Systems (AREA)
- Biophysics (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
The invention provides a disease prediction system based on a variational self-encoder and an electronic device thereof, wherein the disease prediction system comprises: the data acquisition module is used for acquiring historical diagnosis data related to a user; the preprocessing module is used for extracting a diagnosis result from historical diagnosis data; and the disease prediction module is used for processing the diagnosis result based on the trained variational self-encoder to generate a disease prediction result. Therefore, the possible diseases can be predicted according to the historical diagnosis result of the user.
Description
[ technical field ] A method for producing a semiconductor device
The invention belongs to the field of disease prediction, and particularly relates to a disease prediction system based on a variational self-encoder and electronic equipment thereof.
[ background of the invention ]
With the development of information technology and artificial intelligence in recent years, research on predicting a specific disease has been widely applied, and in particular, deep learning technology with a neural network as a core has been rapidly developed. Due to its efficient feature extraction capability and nonlinear learning capability, more and more studies are applying deep learning to the diagnostic prediction of various diseases and showing very satisfactory results.
Many studies on prediction of a single disease have been disclosed in these years, and for example, for prediction of cerebral infarction, there has been a technique of starting prediction by constructing a Convolutional Neural Network (CNN) model using data suitable for structured and unstructured. As another example, the "deep patient" model representation studied and derived by Miotto, r. As another example, Nagrecha et al use a "diagnostic map" to predict heart failure in an elderly patient in an attempt to mine an important disease progression track to help one predict heart failure. For another example, Qingyu Zhao et al uses a general regression model based on a variational self-encoding framework and applies it to the brain aging prediction problem of structured nuclear magnetic images. These problems all involve the prediction of a single disease, and the probability or index of risk of future disease occurrence for other patients similar to the patient cannot be predicted based on the patient's diagnosis history.
In recent years, research has been undertaken to develop collaborative filtering algorithms that can be applied in the field of disease prediction. The collaborative filtering algorithm is an algorithm based on association rules, finds the preference of users by mining historical behavior data of the users, divides the groups of the users based on different preferences and recommends commodities with similar tastes, and is successfully applied to a recommendation system in the entertainment industry and the electronic retail industry. These systems predict item dependencies not found in an entity history by exploring the entity's current item history. Based on this idea, if the disease is taken as a project and the current medical history of the subject is taken as a project history, theoretically, the collaborative filtering method should also be adaptable to disease prediction. For example, Davis et al first proposed and discussed the use of a mechanism of collaborative filtering as a disease predictor. They adopt the method of user preference vector similarity to solve the problem. They have created a system called CARE that uses patient history as input to predict future diagnostic risk based on characteristics of other similar patients. Folino et al employ association rule analysis and Markov models to predict disease risk, which uses a combination of mining models to extract continuous disease patterns. In this regard, one of the most recent studies is the work of Liang et al on collaborative filtering using variational auto-encoders (VAEs), which discusses many modifications to the original VAE loss function to improve recommendation accuracy.
Although the research and application for disease prediction are numerous at present, most of the research is directed to the analysis and prediction of a certain specific disease, but the methods for predicting the risk of various diseases which may occur in the future based on the collaborative filtering algorithm based on the correlation among various diseases of patients are very few, and there is no report on the disease prediction aspect of the application of the collaborative filtering algorithm based on VAE in the related technical literature. For some patients, if the risk probability or index of future disease occurrence of other patients similar to the patient can be predicted by modeling the patient diagnosis history data set, the prevention and prompting of the disease in future life of the patient can be facilitated.
[ summary of the invention ]
The invention aims to solve the defects of the prior art and provide a disease prediction system based on a variational self-encoder and a device thereof so as to provide a more excellent disease risk prediction for a patient.
In order to achieve the above object, the present invention provides a disease prediction system based on a variational self-encoder, comprising:
the data acquisition module is used for acquiring historical diagnosis data related to a user;
the preprocessing module is used for extracting a diagnosis result from historical diagnosis data; and
and the disease prediction module is used for processing the diagnosis result based on the trained variational self-encoder to generate a disease prediction result.
In an embodiment of the present invention, wherein the preprocessing module is further configured to:
preprocessing the historical diagnostic data to obtain a diagnostic result in the historical diagnostic data; and
and mapping the diagnosis result based on the ICD-10 code to obtain the mapped diagnosis result.
In an embodiment of the present invention, wherein mapping the diagnostic result based on the ICD-10 code to obtain the mapped diagnostic result includes:
and mapping the diagnosis result based on the ICD-10 code to obtain a multidimensional binary vector, wherein the multidimensional binary vector is the mapped diagnosis result.
In an embodiment of the present invention, the variational self encoder includes an input layer, an encoding layer, a sampling layer, a decoding layer and an output layer;
wherein the input layer is configured to receive the mapped diagnostic result;
the coding layer is used for coding the mapped diagnosis result;
wherein the sampling layer is used for sampling the diagnosis result after encoding;
wherein the decoding layer is used for decoding the sampled diagnosis result to generate a disease prediction result;
wherein the output layer comprises a first output layer for outputting the disease prediction result.
In an embodiment of the invention, the first output layer is a Softmax layer.
In an embodiment of the present invention, the output layer further includes a second output layer, configured to output the diagnosis result reconstructed based on the Sigmoid-activated function.
In an embodiment of the present invention, in the process of training the variational self-encoder, a loss function of the variational self-encoder is formed by weighting a first loss function and a second loss function, wherein the first loss function is a negative log-likelihood function; the second loss function is a binary cross entropy between the input and the reconstructed output.
In an embodiment of the present invention, during the training of the variational auto-encoder, a neural network Adam optimization algorithm is used to optimize the variational auto-encoder.
In one embodiment of the present invention, in the training of the variational autoencoder, the variational autoencoder is evaluated by using prediction accuracy, coverage and average ranking.
To achieve the above object, the present invention also provides an electronic device, comprising a memory for storing information including program instructions and a processor for controlling the execution of the program instructions, wherein the program instructions are loaded and executed by the processor to implement the variational auto-encoder based disease prediction system as described above.
Compared with the prior art, the disease prediction system based on the variational self-encoder provided by the invention has the advantages that the historical diagnosis data of a user is preprocessed in a mapping coding mode, the running speed and efficiency of the disease prediction system can be improved, and the disease prediction system has good interpretability.
In addition, compared with the prior art, the variational self-encoder training process of the disease prediction system based on the variational self-encoder provided by the invention adopts the weighted composite loss function, so that the fitting property and the prediction accuracy of the variational self-encoder can be improved.
[ description of the drawings ]
Fig. 1 is a block diagram of a disease prediction system based on a variational self-encoder according to an embodiment of the present invention.
Fig. 2 is a block diagram of the variational self-encoder in the above embodiment provided by the present invention.
Fig. 3 is a network architecture diagram for the variational self-encoder in the above embodiment provided by the present invention.
Fig. 4 is a schematic diagram of network reparameterization of the sampling layer for the variational self-encoder in the above embodiment provided by the present invention.
Fig. 5 is a network architecture diagram of the output layer for the variational self-encoder in the above embodiment provided by the present invention.
Fig. 6 is a comparison graph of the optimization using the complex loss function and the optimization using the single loss function for the variational self-encoder in the above embodiment provided by the present invention.
FIG. 7 is a block diagram illustrating an electronic device according to the above embodiment of the present invention.
[ detailed description ] embodiments
The following description is presented to disclose the invention so as to enable any person skilled in the art to practice the invention. The preferred embodiments in the following description are given by way of example only, and other obvious variations will occur to those skilled in the art. The basic principles of the invention, as defined in the following description, may be applied to other embodiments, variations, modifications, equivalents, and other technical solutions without departing from the spirit and scope of the invention.
In the present invention, the terms "a" and "an" in the claims and the description should be understood as meaning "one or more", that is, one element may be one in number in one embodiment, and the element may be more than one in number in another embodiment. The terms "a" and "an" should not be construed as limiting the number unless the number of such elements is explicitly recited as one in the present disclosure, but rather the terms "a" and "an" should not be construed as being limited to only one of the number.
In the description of the present invention, it is to be understood that the terms "first," "second," and the like are used for descriptive purposes only and are not to be construed as indicating or implying relative importance. In the description of the present invention, it should be noted that, unless explicitly stated or limited otherwise, the terms "connected" and "connected" are to be interpreted broadly, and may be, for example, fixedly connected, detachably connected, or integrally connected; can be mechanically or electrically connected; they may be directly connected or indirectly connected through an intermediate. The specific meanings of the above terms in the present invention can be understood by those skilled in the art according to specific situations.
In the description herein, references to the description of the term "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., mean that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above are not necessarily intended to refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, various embodiments or examples and features of different embodiments or examples described in this specification can be combined and combined by one skilled in the art without contradiction.
The embodiment of the invention provides a disease prediction system based on a variational self-encoder, which is used for analyzing historical diagnosis data of a user through a variational self-encoder to obtain the probability that the user suffers from a certain disease in the future, as shown in fig. 1. The disease prediction system comprises a data acquisition module 10, a preprocessing module 20 and a disease prediction module 30.
A data acquisition module 10 for acquiring historical diagnostic data associated with a user.
And the preprocessing module 20 is used for extracting a diagnosis result from the historical diagnosis data.
And the disease prediction module 30 is used for processing the diagnosis result based on the trained variational self-encoder to generate a disease prediction result.
The historical diagnostic data associated with the user includes the year of diagnosis, the hospital of diagnosis, and the results of the diagnosis.
In a preprocessing module 20, preprocessing the historical diagnosis data to obtain diagnosis results in the historical diagnosis data, and mapping the diagnosis results based on the ICD-10 code to obtain the mapped diagnosis results; the preprocessing process is mainly to combine and integrate the historical diagnosis data of the user into a data table through data grouping, wherein the content of the data table only retains the diagnosis results of the diagnosis data diagnosed by the user in different historical periods. And mapping the diagnosis result of the obtained historical diagnosis data based on the ICD-10 code, so that each diagnosed disease in the diagnosis result of the historical diagnosis data has a diagnosis code with a unique identification. For example, if a diabetic bone disease is diagnosed in the historical diagnosis results of the user, the diagnosis code mapped based on the ICD-10 code is changed to E14.602, and accordingly, the diagnosis code is the code of E14.602, which refers to that the diagnosis result is the diabetic bone disease.
Further, mapping the diagnosis result based on the ICD-10 code to obtain the mapped diagnosis result includes:
and mapping the diagnosis result based on the ICD-10 code to obtain a multidimensional binary vector, wherein the multidimensional binary vector is the mapped diagnosis result.
Here, the ICD-10 codes correspond to diagnosis codes of unique identifications of 6353 common diseases, diagnosis results in historical diagnosis data of the user are mapped based on the ICD-10 codes, and whether the user suffers from a certain disease is characterized one by one through binary codes (i.e., 0 or 1) to form a 6353-dimensional disease diagnosis table. In the present invention, a binary vector is defined as a binary code (0 or 1) used to characterize whether the user has suffered from a certain disease. When the value of the historical diagnosis result is 1, the user is indicated to have a disease corresponding to the current diagnosis code in the historical diagnosis data; and when the value of the historical diagnosis result is 0, the user does not suffer from the disease corresponding to the current diagnosis code in the historical diagnosis data. For example, E14.602 corresponding to ICD-10 code is denoted as diabetic bone disease, and when the user suffers from diabetic bone disease in the real diagnostic data, it is denoted as 1; when the user has not suffered from diabetic bone disease in his historical diagnostic data, it is indicated as 0.
As shown in fig. 2, a Variational self-encoder (VAE) in the disease prediction module 30 includes an input layer 31, an encoding layer 32, a sampling layer 33, a decoding layer 34 and an output layer 35, and is configured to perform data processing on a diagnosis result and output a predicted diagnosis result, where the input layer 31 is configured to receive the mapped diagnosis result; the coding layer 32 is configured to code the mapped diagnosis result; the sampling layer 33 is used for sampling the diagnosis result after encoding; the decoding layer 34 is configured to decode the adopted diagnosis result to generate a disease prediction result, and the output layer 35 includes a first output layer 351 configured to output the disease prediction result.
Further, the variational self-encoder is used for receiving the mapped diagnosis result from the input layer 31 at the encoding layer 32, wherein the encoding layer 32 is an encoder of the variational self-encoder, and the encoder of the variational self-encoder outputs variance and mean after encoding; the sampling variables are determined from the variance and mean values at the sampling layer 33. The decoding layer 34 is a decoder of the variational self-encoder, and outputs the sampling variables in the sampling layer 33 to the decoding layer 34 to generate a disease prediction result by decoding and output the disease prediction result through the first output layer 351 of the output layer 35. Here, the disease prediction result refers to a probability that a patient may suffer from a certain disease in the future.
Further, as shown in fig. 3, in the encoding layer 32, the encoder of the variational self-encoder includes two neural networks, which respectively output μ and σ, and the sampling variables of the sampling layer 33 can be obtained by μ and σ respectively output by the two neural networks, specifically, according to μ and σ, random numbers of corresponding gaussian distributions are generated, the variance of the gaussian distributions is σ, and the mean value of the gaussian distributions is μ, a standard normal distribution is randomly sampled to obtain sampling values ε, and a sampling variable z ═ u + σ ⊙ ε is determined according to the sampling values ε and the random numbers of the gaussian distributions.
As shown in fig. 2 and 5, the output layer 35 includes a first output layer 351 and a second output layer 352, wherein the first output layer is a Softmax layer for outputting the disease prediction result; the second output layer 352 is configured to output the diagnosis result reconstructed based on the Sigmoid activation function. It is worth mentioning that in the embodiment of the present invention, the variational self-encoder uses the input vector of the diagnosis result with the dimension of 6353, wherein the intermediate hidden layers (including the encoding layer 32 and the decoding layer 34) are 500-dimensional, the potential spatial layer (the sampling layer 33) is 100-dimensional, and all the hidden layers use the Relu activation function to realize the disease prediction for the historical diagnosis result. In the embodiment of the invention, when the variational self-encoder is trained, 37000 patients are trained, and 4000 patients are verified, so that the accuracy of disease prediction is improved through a large amount of training and optimization.
In order to maximize the specific goals of the first output layer 351 and the second output layer 352, each output layer 35 has a specific loss function optimized during the training of the variational self-coder. Specifically, the composite loss function of the variational self-encoder is formed by weighting a first loss function and a second loss function, wherein the first loss function is a negative log-likelihood function to replace a reconstruction error so as to optimize the accurate probability distribution of a prediction result; the second loss function is a binary cross entropy between the input and the reconstructed output to optimize output of an accurate reconstructed diagnostic result. Here, the goal of the variational self-editor is to maximize the reconstruction probability with one regularization parameter, the loss function (first or second) is a negative log-likelihood function with regularization, but since there is no global expression shared by all data points, here the loss function is decomposed into terms that depend only on a single data point li, then the total loss is the sum of the losses of N data points as the sum of the losses of N data points
Where li is formulated as:
It is worth mentioning that the first loss function and the second loss function have the same function in intuition, but each of them is more suitable for its own output, then different weights are set for the two loss functions (the first loss function and the second loss function), and then the two loss functions are weighted to obtain a composite loss function, which has smaller loss than a single loss function, better fitting of a model, and higher accuracy. For example, the weighted complex loss function is compared to the single loss function for training, and experiments have shown that the weighted complex loss function is less lost during training and is best at the 8 th epochs, as shown in fig. 6.
Meanwhile, in order to balance the fitting effect, parameters β can be introduced to control the strength of the regularization, and then the optimization objective is converted into a function of minimizing loss:
from the perspective of a variational self-encoder, the distribution of clustering categories is prevented from being excessively balanced, which accords with the actual condition that the diagnosis result of a patient is multi-label and can not be completely summarized to a single category, and has good effect of regular term parameters.
Further, in the process of training the variational auto-encoder, in order to make the parameters relatively stable, in an embodiment of the present invention, a neural network Adam optimization algorithm is used to optimize the variational auto-encoder. Specifically, the neural network Adam optimization algorithm is obtained by fusing an algorithm called RMSprop (accelerated gradient descent) on the basis of a gradient descent method with momentum. It dynamically adjusts the learning rate of each parameter using first moment estimates and second moment estimates of the gradient.
It is specifically represented as:
mt＝μ*mt-1+(1-μ)*gt
in the above formula, mtExpressed as an updated biased first moment estimate; v. oftExpressed as updating the biased second moment estimate;
expressed as a deviation correcting the first moment;
expressed as a deviation of the corrected second moment; delta thetatRepresented as an update of the calculation parameters.
The Adam has the advantages that after offset correction, each iteration learning rate has a certain range, so that parameters are stable, and all diagnosis results can be sequenced through a trained variational self-encoder for historical diagnosis data of a patient.
To verify the training effect of the variational autoencoder, in embodiments of the present invention, the variational autoencoder is evaluated with prediction accuracy, coverage, and average ranking. Here, to simulate the challenge of predicting unknown diseases in the real world to measure the performance of a disease prediction system, we tested the performance of the disease prediction system by the hold-out method. The leave-out method is that for each patient, half of diagnosis results are reserved from historical diagnosis data of the patient, diagnosis codes of the reserved diagnosis results are changed to 0, then the reserved patient diagnosis test data are transmitted to a disease prediction system to obtain the probability of disease diagnosis, and the coverage rate and the average ranking index are calculated according to the reserved diagnosis probability. Coverage measures the importance of the system to assign to an unseen diagnosis, while average ranking measures the system's coverage of all unseen diagnoses.
In the actual verification process, the prediction result of the disease prediction system (abbreviated as VAE in the table) of the present invention is compared with the Mining technology of fuzzy association rule (abbreviated as AR Mining in the table) and the nursing system (abbreviated as CARE in the table), and the prediction accuracy, the unseen diagnosis coverage rate and the average ranking are respectively compared, which are respectively shown in table 1, table 2 and table 3.
TABLE 1 accuracy of Top-k prediction by three methods
TABLE 2 three methods for predicting Top-k coverage
TABLE 3 three methods to predict Top-k average ranking
The conclusions can be drawn from tables 1 and 2: the variational-autocoder-based disease prediction system, which has significantly more excellent performance, is robust to diagnoses not found in training data. The mean over a range of the variational autocoder-based disease prediction system of table 3 observed that the mean over a range of the variational autocoder-based disease prediction system ranked better than the CARE system (CARE). The average ranking is calculated for those unknown diagnoses that occur within range, and the fuzzy association rule mining techniques and care systems would perform worse for many diagnoses that may be present that are not within range. The disease prediction system based on the variational self-encoder can be displayed by calculating the average ranking of all unknown diagnoses, so that the performance is improved.
In summary, compared with the fuzzy association rule Mining (AR Mining) and the CARE system (CARE), the disease prediction system based on the variational autocoder of the present invention has significantly better performance.
In the training process of the variational self-encoder, the anti-noise capability of the variational self-encoder is detected or trained so as to improve the robustness of the disease prediction system to noise input. In the data preprocessing performed by the disease prediction system of the present invention, which is usually due to human error or noise included in the historical diagnosis data of the patient, in order to check whether the prediction performance of the disease prediction system of the present invention varies with the introduction of noise, assuming that a test in which noise diagnosis is manually inputted into a diagnosis test set is used, the average number of diagnoses per patient is 12, there is a noise diagnosis per patient, and the result of the prediction accuracy variation of the noise diagnosis per patient from 1 to 10 by the test is shown in table 4:
TABLE 4 accuracy of Top-20 diagnosis by noise introduction
As can be seen from the table, the anti-noise capability of the variational self-encoder is detected, and the disease prediction system has stronger robustness on noise input.
To achieve the above object, the present invention also provides an electronic device, comprising a memory for storing information including program instructions and a processor for controlling the execution of the program instructions, wherein the program instructions are loaded and executed by the processor to implement the variational auto-encoder based disease prediction system as described above.
Next, an electronic apparatus according to an embodiment of the present invention is described with reference to the drawings. As shown in fig. 7, electronic device 40 includes one or more processors 41 and memory 42.
The processor 41 may be a Central Processing Unit (CPU) or other form of processing unit having data processing capabilities and/or instruction execution capabilities, and may control other components in the electronic device 40 to perform desired functions. In other words, the processor 41 includes one or more physical devices configured to execute instructions. For example, the processor 41 may be configured to execute instructions that are part of: one or more applications, services, programs, routines, libraries, objects, components, data structures, or other logical constructs. Such instructions may be implemented to perform a task, implement a data type, transform the state of one or more components, implement a technical effect, or otherwise arrive at a desired result.
The processor 41 may include one or more processors configured to execute software instructions. Additionally or alternatively, the processor 41 may include one or more hardware or firmware logic machines configured to execute hardware or firmware instructions. The processors of the processor 41 may be single-core or multi-core, and the instructions executed thereon may be configured for serial, parallel, and/or distributed processing. The various components of the processor 41 may optionally be distributed over two or more separate devices, which may be remotely located and/or configured for coordinated processing. Aspects of the processor 41 may be virtualized and executed by remotely accessible networked computing devices configured in a cloud computing configuration.
The memory 42 may include one or more computer program products that may include various forms of computer-readable storage media, such as volatile memory and/or non-volatile memory. The volatile memory may include, for example, Random Access Memory (RAM), cache memory (cache), and/or the like. The non-volatile memory may include, for example, Read Only Memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer readable storage medium and executed by the processor 41 to implement some or all of the steps of the above-described exemplary methods of the present invention, and/or other desired functions.
In other words, the memory 42 comprises one or more physical devices configured to hold machine-readable instructions executable by the processor 41 to implement the methods and processes described herein. In implementing these methods and processes, the state of the memory 42 may be transformed (e.g., to hold different data). The memory 42 may comprise a removable and/or built-in device. The memory 42 may include optical memory (e.g., CD, DVD, HD-DVD, blu-ray disc, etc.), semiconductor memory (e.g., RAM, EPROM, EEPROM, etc.), and/or magnetic memory (e.g., hard-disk drive, floppy-disk drive, tape drive, MRAM, etc.), among others. The memory 42 may include volatile, nonvolatile, dynamic, static, read/write, read-only, random-access, sequential-access, location-addressable, file-addressable, and/or content-addressable devices.
It is understood that the memory 42 comprises one or more physical devices. However, aspects of the instructions described herein may alternatively be propagated by a communication medium (e.g., an electromagnetic signal, an optical signal, etc.) that is not held by a physical device for a limited period of time. Aspects of the processor 41 and the memory 42 may be integrated together into one or more hardware logic components. These hardware logic components may include, for example, Field Programmable Gate Arrays (FPGAs), program and application specific integrated circuits (PASIC/ASIC), program and application specific standard products (PSSP/ASSP), system on a chip (SOC), and Complex Programmable Logic Devices (CPLDs).
In one example, as shown in FIG. 7, the electronic device 40 may also include input and output devices that are interconnected via a bus system and/or other form of connection mechanism (not shown). For example, the input device may be, for example, a camera module or the like for capturing image data or video data. As another example, the input device may include or interface with one or more user input devices such as a keyboard, mouse, touch screen, or game controller. In some embodiments, the input device may include or interface with a selected Natural User Input (NUI) component. Such component parts may be integrated or peripheral and the transduction and/or processing of input actions may be processed on-board or off-board. Example NUI components may include a microphone for speech and/or voice recognition; infrared, color, stereo display and/or depth camera for machine vision and/or gesture recognition; a head tracker, eye tracker, accelerometer and/or gyroscope for motion detection and/or intent recognition; and an electric field sensing component for assessing brain activity and/or body movement; and/or any other suitable sensor.
The output device can output various information including classification results and the like to the outside. The output devices may include, for example, a display, speakers, a printer, and a communication network and remote output devices connected thereto, among others.
Of course, the electronic device 40 may further include the communication apparatus, wherein the communication apparatus may be configured to communicatively couple the electronic device 40 with one or more other computer devices. The communication means may comprise wired and/or wireless communication devices compatible with one or more different communication protocols. As a non-limiting example, the communication subsystem may be configured for communication via a wireless telephone network or a wired or wireless local or wide area network. In some embodiments, the communication device may allow the electronic device 40 to send and/or receive messages to and/or from other devices via a network such as the internet.
It will be appreciated that the configurations and/or approaches described herein are exemplary in nature, and that these specific embodiments or examples are not to be considered in a limiting sense, because numerous variations are possible. The specific routines or methods described herein may represent one or more of any number of processing strategies. As such, various acts illustrated and/or described may be performed in the sequence illustrated and/or described, in other sequences, in parallel, or omitted. Also, the order of the above-described processes may be changed.
Of course, for simplicity, only some of the components of the electronic device 40 relevant to the present invention are shown, and components such as buses, input/output interfaces, and the like are omitted. In addition, electronic device 40 may include any other suitable components, depending on the particular application.
According to another aspect of the present invention, the present invention further provides an electronic device such as a smart phone, a smart robot, etc., wherein the electronic device is configured with the above-mentioned variational-autocoder-based disease prediction system for performing disease prediction on a user. Illustratively, the electronic device comprises a smart phone and the variational-autocoder-based disease prediction system 1, wherein the variational-autocoder-based disease prediction system 1 is configured to the smart phone for performing disease prediction on historical diagnosis results input via the smart phone. It is to be understood that the smartphone may be, but is not limited to being, implemented as a camera-enabled smartphone.
In addition to the above-described methods and apparatus, embodiments of the present invention may also be a computer program product comprising computer program instructions that, when executed by a processor, cause the processor to perform the steps in the methods according to various embodiments of the present invention described in the "exemplary methods" section above of this specification.
The computer program product may write program code for carrying out operations for embodiments of the present invention in any combination of one or more programming languages, including an object oriented programming language such as Java, C + + or the like and conventional procedural programming languages, such as the C language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server.
Furthermore, an embodiment of the present invention may also be a computer-readable storage medium having stored thereon computer program instructions, which, when executed by a processor, cause the processor to perform the steps of the above-described method of the present specification.
The computer readable storage medium may take any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may include, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable disk, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
The basic principles of the present invention have been described above with reference to specific embodiments, but it should be noted that the advantages, effects, etc. mentioned in the present invention are only examples and are not limiting, and the advantages, effects, etc. must not be considered to be possessed by various embodiments of the present invention. Furthermore, the foregoing disclosure of specific details is for the purpose of illustration and description and is not intended to be limiting, since the invention is not limited to the specific details described above.
The block diagrams of devices, apparatuses, systems involved in the present invention are only given as illustrative examples and are not intended to require or imply that the connections, arrangements, configurations, etc. must be made in the manner shown in the block diagrams. These devices, apparatuses, devices, systems may be connected, arranged, configured in any manner, as will be appreciated by those skilled in the art. Words such as "including," "comprising," "having," and the like are open-ended words that mean "including, but not limited to," and are used interchangeably therewith. The words "or" and "as used herein mean, and are used interchangeably with, the word" and/or, "unless the context clearly dictates otherwise. The word "such as" is used herein to mean, and is used interchangeably with, the phrase "such as but not limited to".
It should also be noted that in the apparatus, devices and methods of the present invention, the components or steps may be broken down and/or re-combined. These decompositions and/or recombinations are to be regarded as equivalents of the present invention.
The previous description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these aspects will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other aspects without departing from the scope of the invention. Thus, the present invention is not intended to be limited to the aspects shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
It will be appreciated by persons skilled in the art that the embodiments of the invention described above and shown in the drawings are given by way of example only and are not limiting of the invention. The objects of the invention have been fully and effectively accomplished. The functional and structural principles of the present invention have been shown and described in the examples, and any variations or modifications of the embodiments of the present invention may be made without departing from the principles.
Claims (10)
1. A variational-autocoder-based disease prediction system, comprising:
the data acquisition module is used for acquiring historical diagnosis data related to a user;
the preprocessing module is used for extracting a diagnosis result from historical diagnosis data; and
and the disease prediction module is used for processing the diagnosis result based on the trained variational self-encoder to generate a disease prediction result.
2. The disease prediction system of claim 1, wherein the preprocessing module is further to:
preprocessing the historical diagnostic data to obtain a diagnostic result in the historical diagnostic data; and
and mapping the diagnosis result based on the ICD-10 code to obtain the mapped diagnosis result.
3. The disease prediction system of claim 2, wherein mapping the diagnostic result based on the ICD-10 code to obtain the mapped diagnostic result comprises:
and mapping the diagnosis result based on the ICD-10 code to obtain a multidimensional binary vector, wherein the multidimensional binary vector is the mapped diagnosis result.
4. The disease prediction system of claim 1, wherein the variational self-encoder comprises an input layer, an encoding layer, a sampling layer, a decoding layer, and an output layer;
wherein the input layer is configured to receive the mapped diagnostic result;
the coding layer is used for coding the mapped diagnosis result;
wherein the sampling layer is used for sampling the diagnosis result after encoding;
wherein the decoding layer is used for decoding the sampled diagnosis result to generate a disease prediction result;
wherein the output layer comprises a first output layer for outputting the disease prediction result.
5. A disease prediction system according to claim 4, wherein the first output layer is a Softmax layer.
6. The disease prediction system of claim 4, wherein the output layer further comprises a second output layer for outputting the diagnosis reconstructed based on a Sigmoid activation function.
7. The disease prediction system of claim 1, wherein, in training the variational auto-encoder, a loss function of the variational auto-encoder is weighted by a first loss function and a second loss function, wherein the first loss function is a negative log-likelihood function; the second loss function is a binary cross entropy between the input and the reconstructed output.
8. The disease prediction system of claim 1, wherein the variational auto-encoder is optimized using a neural network Adam optimization algorithm in training the variational auto-encoder.
9. The disease prediction system of claim 1, wherein in training the variational self-encoder, the variational self-encoder is evaluated with prediction accuracy, coverage and average ranking.
10. An electronic device comprising a memory for storing information comprising program instructions and a processor for controlling the execution of the program instructions, wherein the program instructions when loaded and executed by the processor implement a variational auto-encoder based disease prediction system as claimed in any one of claims 1 to 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911404700.6A CN111161883A (en) | 2019-12-31 | 2019-12-31 | Disease prediction system based on variational self-encoder and electronic equipment thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911404700.6A CN111161883A (en) | 2019-12-31 | 2019-12-31 | Disease prediction system based on variational self-encoder and electronic equipment thereof |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111161883A true CN111161883A (en) | 2020-05-15 |
Family
ID=70559547
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911404700.6A Pending CN111161883A (en) | 2019-12-31 | 2019-12-31 | Disease prediction system based on variational self-encoder and electronic equipment thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111161883A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112202696A (en) * | 2020-10-12 | 2021-01-08 | 青岛科技大学 | Underwater sound signal automatic modulation identification method based on fuzzy self-encoder |
| CN113380417A (en) * | 2021-06-17 | 2021-09-10 | 哈尔滨理工大学 | LR-N based cardiovascular disease prediction method |
| CN113671421A (en) * | 2021-08-24 | 2021-11-19 | 华北电力大学（保定） | Transformer state evaluation and fault early warning method |
| CN115604365A (en) * | 2022-11-15 | 2023-01-13 | 南京芯传汇电子科技有限公司（Cn） | Data encoding and decoding method and device, electronic equipment and readable storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106326925A (en) * | 2016-08-23 | 2017-01-11 | 南京邮电大学 | Apple disease image identification method based on deep learning network |
| CN109036553A (en) * | 2018-08-01 | 2018-12-18 | 北京理工大学 | A kind of disease forecasting method based on automatic extraction Medical Technologist's knowledge |
-
2019
- 2019-12-31 CN CN201911404700.6A patent/CN111161883A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106326925A (en) * | 2016-08-23 | 2017-01-11 | 南京邮电大学 | Apple disease image identification method based on deep learning network |
| CN109036553A (en) * | 2018-08-01 | 2018-12-18 | 北京理工大学 | A kind of disease forecasting method based on automatic extraction Medical Technologist's knowledge |
Non-Patent Citations (1)
| Title |
|---|
| 韩浩先 等: "基于聚类变分自编码器的协同过滤算法", 《计算机系统应用》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112202696A (en) * | 2020-10-12 | 2021-01-08 | 青岛科技大学 | Underwater sound signal automatic modulation identification method based on fuzzy self-encoder |
| CN113380417A (en) * | 2021-06-17 | 2021-09-10 | 哈尔滨理工大学 | LR-N based cardiovascular disease prediction method |
| CN113671421A (en) * | 2021-08-24 | 2021-11-19 | 华北电力大学（保定） | Transformer state evaluation and fault early warning method |
| CN115604365A (en) * | 2022-11-15 | 2023-01-13 | 南京芯传汇电子科技有限公司（Cn） | Data encoding and decoding method and device, electronic equipment and readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11937929B2 (en) | Systems and methods for using mobile and wearable video capture and feedback plat-forms for therapy of mental disorders | |
| CN108351862B (en) | Method and apparatus for determining developmental progress using artificial intelligence and user input | |
| CN111161883A (en) | Disease prediction system based on variational self-encoder and electronic equipment thereof | |
| US11270565B2 (en) | Electronic device and control method therefor | |
| JP7191443B2 (en) | Target object attribute prediction method based on machine learning, related equipment and computer program | |
| WO2019137538A1 (en) | Emotion representative image to derive health rating | |
| KR102637133B1 (en) | On-device activity recognition | |
| EP3786882A1 (en) | Movement state recognition model learning device, movement state recognition device, method, and program | |
| CN111640512A (en) | Kidney replacement therapy starting strategy evaluation method and device and electronic equipment | |
| Ali et al. | Multitask deep learning for cost-effective prediction of patient's length of stay and readmission state using multimodal physical activity sensory data | |
| CN111785372A (en) | Collaborative filtering disease prediction system based on association rule and electronic equipment thereof | |
| Kramer et al. | Reconstructing nonlinear dynamical systems from multi-modal time series | |
| Yu et al. | Semi-supervised learning and data augmentation in wearable-based momentary stress detection in the wild | |
| Slim et al. | IoT System Based on parameter optimization of Deep Learning using Genetic Algorithm. | |
| Wang et al. | A wearable-HAR oriented sensory data generation method based on spatio-temporal reinforced conditional GANs | |
| US20220029986A1 (en) | Methods and systems of biometric identification in telemedicine using remote sensing | |
| Zhao et al. | Attention‐based sensor fusion for emotion recognition from human motion by combining convolutional neural network and weighted kernel support vector machine and using inertial measurement unit signals | |
| D’Aniello et al. | Situation identification in smart wearable computing systems based on machine learning and Context Space Theory | |
| WO2023031235A1 (en) | Semi-supervised machine learning method and system suitable for identification of patient subgroups in electronic healthcare records | |
| Raina et al. | Intelligent and Interactive Healthcare System (I 2 HS) Using Machine Learning | |
| Tan et al. | Convolutional Neural Network with Multi-Head Attention for Human Activity Recognition | |
| Christal et al. | Heart diseases diagnosis using chaotic Harris Hawk optimization with E-CNN for IoMT framework | |
| Olsen | Detecting Human Emotions Using Smartphone Accelerometer Data | |
| CN115496105B (en) | Sleep prediction model training method, sleep condition prediction method and related devices | |
| CN114449701B (en) | Light type regulation and control method and device based on medical record information and electronic equipment |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20200515 |