CN113936809A - Infectious disease prediction and training method, device, equipment and medium - Google Patents

Infectious disease prediction and training method, device, equipment and medium Download PDF

Info

Publication number
CN113936809A
CN113936809A CN202111201853.8A CN202111201853A CN113936809A CN 113936809 A CN113936809 A CN 113936809A CN 202111201853 A CN202111201853 A CN 202111201853A CN 113936809 A CN113936809 A CN 113936809A
Authority
CN
China
Prior art keywords
user
node
feature vector
region
graph
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
Application number
CN202111201853.8A
Other languages
Chinese (zh)
Inventor
黄予
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tencent Technology Shenzhen Co Ltd
Original Assignee
Tencent Technology Shenzhen Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Tencent Technology Shenzhen Co Ltd filed Critical Tencent Technology Shenzhen Co Ltd
Priority to CN202111201853.8A priority Critical patent/CN113936809A/en
Publication of CN113936809A publication Critical patent/CN113936809A/en
Pending legal-status Critical Current

Links

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
    • G16H50/80—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for detecting, monitoring or modelling epidemics or pandemics, e.g. flu
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/047—Probabilistic or stochastic networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Biomedical Technology (AREA)
  • General Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • Software Systems (AREA)
  • Evolutionary Computation (AREA)
  • Biophysics (AREA)
  • Molecular Biology (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Computational Linguistics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Public Health (AREA)
  • Medical Informatics (AREA)
  • Probability & Statistics with Applications (AREA)
  • Databases & Information Systems (AREA)
  • Pathology (AREA)
  • Epidemiology (AREA)
  • Primary Health Care (AREA)
  • Measuring And Recording Apparatus For Diagnosis (AREA)

Abstract

The application discloses an infectious disease prediction and training method, an infectious disease prediction and training device, infectious disease prediction and training equipment and medium, and belongs to the field of artificial intelligence. The prediction method comprises the following steps: acquiring first physiological information and first position information of a first user; adding a first user node in the basic graph network of the infectious disease based on the first physiological information and the first position information to obtain an updated graph network; the basic graph network comprises at least one user node and/or at least one area node, wherein the user node identifies a user suspected to be infected with the infectious disease, and the area node identifies an area where the user suspected to be infected with the infectious disease is located; inputting the updated graph network into a graph convolution neural network, and predicting to obtain a first user feature vector of a first user; and predicting the probability that the first user has the infectious disease through the probability computation network based on the first user feature vector. The scheme improves the accuracy of infectious disease prediction.

Description

Infectious disease prediction and training method, device, equipment and medium
Technical Field
The present application relates to the field of artificial intelligence, and in particular, to a method, an apparatus, a device, and a medium for infectious disease prediction and training.
Background
In the medical field, infectious disease prediction plays a crucial role in the prevention and control of infectious diseases.
In the related art, the probability that a suspected patient has the target infectious disease is predicted by judging whether the symptom of the suspected patient is the symptom of the target infectious disease; for example, suspected patients exhibit symptoms of cough (symptoms of the target infectious disease), and the related art labels suspected patients as high risk persons.
However, the related art cannot distinguish other similar diseases by symptom information, for example, the related art may mark a patient actually suffering from a cold (cough is also a symptom of the cold) as a high-risk person of the target infectious disease, and the accuracy of infectious disease prediction by the related art is low.
Disclosure of Invention
The application provides an infectious disease prediction and training method, an infectious disease prediction and training device and an infectious disease prediction and training medium, which can improve the accuracy of infectious disease prediction. The technical scheme is as follows:
according to an aspect of the present application, there is provided an infectious disease prediction method, the method including:
acquiring first physiological information and first position information of a first user, wherein the first physiological information is physiological information of the first user related to infectious diseases, and the first position information indicates the position of the first user;
adding a first user node on the basis of the first physiological information and the first position information on the basis of the basic graph network of the infectious disease to obtain an updated graph network; the basic graph network comprises at least one user node and/or at least one area node, wherein the user node identifies a user suspected to be infected with the infectious disease, and the area node identifies an area where the user suspected to be infected with the infectious disease is located;
inputting the updated graph network into a graph convolution neural network, and predicting to obtain a first user feature vector of a first user; and predicting the probability that the first user has the infectious disease through the probability computation network based on the first user feature vector.
According to another aspect of the present application, there is provided a method for training a convolutional neural network, the method using a semi-supervised mode, the method comprising:
acquiring sample physiological information and sample position information of n sample users; the n sample users include n1A first sample user and n2A second sample user, the first sample user carrying a label for determining whether the first sample user suffers fromWith an infectious disease, the second sample user does not carry a label;
inputting a sample graph network corresponding to sample physiological information and sample position information of n sample users into a graph convolution neural network, and calculating n sample characteristic vectors of the n sample users through the graph convolution neural network;
for n1One of the first sample users obtains the predicted probability that the first sample user has the infectious disease through a probability calculation network based on the sample feature vectors of the first sample users;
based on the prediction probability and n1And (3) training the convolutional neural network of the graph according to the loss value of the proportion of the diseased sample users in the first sample users.
According to another aspect of the present application, there is provided a method for training a convolutional neural network, the method using a supervised mode, the method comprising:
obtaining n1Sample physiological information and sample location information for a first sample user; the first sample user carries a tag for determining whether the first sample user has an infectious disease;
will be reacted with n1Inputting the sample graph network corresponding to the sample physiological information and the sample position information of the first sample user into the graph convolution neural network, and obtaining n through the computation of the graph convolution neural network1N of a first sample user1A sample feature vector;
for n1One of the first sample users obtains the predicted probability that the first sample user has the infectious disease through a probability calculation network based on the sample feature vectors of the first sample users;
based on the prediction probability and n1And (3) training the convolutional neural network of the graph according to the loss value of the proportion of the diseased sample users in the first sample users.
According to another aspect of the present application, there is provided an infectious disease prediction apparatus including:
the system comprises an acquisition module, a processing module and a display module, wherein the acquisition module is used for acquiring first physiological information and first position information of a first user, the first physiological information is physiological information of the first user related to infectious diseases, and the first position information indicates the position of the first user;
the updating module is used for adding a first user node in the basic graph network of the infectious disease based on the first physiological information and the first position information to obtain an updated graph network; the basic graph network comprises at least one user node and/or at least one area node, wherein the user node identifies a user suspected to be infected with the infectious disease, and the area node identifies an area where the user suspected to be infected with the infectious disease is located;
the prediction module is used for inputting the updated graph network into the graph convolution neural network to predict and obtain a first user feature vector of the first user; and predicting the probability that the first user has the infectious disease through the probability computation network based on the first user feature vector.
According to another aspect of the present application, there is provided a training apparatus for a convolutional neural network, the apparatus including:
the acquisition module is used for acquiring sample physiological information and sample position information of n sample users; the n sample users include n1A first sample user and n2A second sample user, the first sample user carrying a label for determining whether the first sample user has an infectious disease, the second sample user not carrying a label;
the calculation module is used for inputting a sample graph network corresponding to the sample physiological information and the sample position information of the n sample users into a graph convolution neural network, and obtaining n sample characteristic vectors of the n sample users through the graph convolution neural network;
a calculation module for n1One of the first sample users obtains the predicted probability that the first sample user has the infectious disease through a probability calculation network based on the sample feature vectors of the first sample users;
a training module for training the prediction probability based on n1And (3) training the convolutional neural network of the graph according to the loss value of the proportion of the diseased sample users in the first sample users.
According to another aspect of the present application, there is provided a training apparatus for a convolutional neural network, the apparatus including:
an acquisition module for acquiring n1Sample physiological information and sample location information for a first sample user; the first sample user carries a tag for determining whether the first sample user has an infectious disease;
a calculation module for comparing n with1Inputting the sample graph network corresponding to the sample physiological information and the sample position information of the first sample user into the graph convolution neural network, and obtaining n through the computation of the graph convolution neural network1N of a first sample user1A sample feature vector;
a calculation module for n1One of the first sample users obtains the predicted probability that the first sample user has the infectious disease through a probability calculation network based on the sample feature vectors of the first sample users;
a training module for training the prediction probability based on n1And (3) training the convolutional neural network of the graph according to the loss value of the proportion of the diseased sample users in the first sample users.
According to an aspect of the present application, there is provided a computer device including: a processor and a memory, the memory storing a computer program that is loaded and executed by the processor to implement the infectious disease prediction method, and/or the training method of the atlas neural network, as described above.
According to another aspect of the present application, there is provided a computer-readable storage medium storing a computer program, which is loaded and executed by a processor to implement the infectious disease prediction method as described above, and/or the training method of a atlas neural network.
According to another aspect of the application, a computer program product or computer program is provided, comprising computer instructions stored in a computer readable storage medium. The computer instructions are read by a processor of a computer device from a computer-readable storage medium, and the computer instructions are executed by the processor to cause the computer device to perform the infectious disease prediction method and/or the training method of the graph convolution neural network.
The beneficial effects brought by the technical scheme provided by the embodiment of the application at least comprise:
the method comprises the steps of updating a basic graph network through first physiological information and first position information of a first user to obtain a first user characteristic vector, and converting the first user characteristic vector into the probability that the first user has the infectious disease through a probability calculation network.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
FIG. 1 is a schematic diagram of a computer system provided in an exemplary embodiment of the present application;
FIG. 2 is a flow chart of an infectious disease prediction method provided by an exemplary embodiment of the present application;
FIG. 3 is an interface diagram of a product to which the infectious disease prediction method of FIG. 2 is applied according to an exemplary embodiment of the present application;
FIG. 4 is a flow chart of an infectious disease prediction method provided by another exemplary embodiment of the present application;
FIG. 5 is a schematic diagram of a first graph network provided by an exemplary embodiment of the present application;
FIG. 6 is a flow chart of an infectious disease prediction method according to another exemplary embodiment of the present application;
FIG. 7 is a schematic diagram of a second network provided by an exemplary embodiment of the present application;
FIG. 8 is a flow chart of an infectious disease prediction method according to another exemplary embodiment of the present application;
FIG. 9 is a schematic diagram of a third network provided by an exemplary embodiment of the present application;
FIG. 10 is a flow chart of an infectious disease prediction method provided by another exemplary embodiment of the present application;
FIG. 11 is a schematic diagram of a fourth network provided by an exemplary embodiment of the present application;
FIG. 12 is a schematic diagram of a fifth network and a sixth network provided by an exemplary embodiment of the present application;
FIG. 13 is a flow chart of a method of training a convolutional neural network provided in an exemplary embodiment of the present application;
FIG. 14 is a flow chart of a method of training a convolutional neural network provided by another exemplary embodiment of the present application;
fig. 15 is a block diagram illustrating an infectious disease prediction apparatus according to an exemplary embodiment of the present application;
FIG. 16 is a block diagram of a training apparatus for a convolutional neural network provided in an exemplary embodiment of the present application;
FIG. 17 is a block diagram of a training apparatus for a convolutional neural network according to another exemplary embodiment of the present application;
FIG. 18 is a block diagram of a computer device provided in an exemplary embodiment of the present application.
Detailed Description
To make the objects, technical solutions and advantages of the present application more clear, embodiments of the present application will be described in further detail below with reference to the accompanying drawings.
First, terms referred to in the embodiments of the present application are briefly described:
graph network: in the embodiment of the present application, a graph network refers to data stored in the form of a graph, and is also referred to as graph data, a graph model, a graph representation, and graph structure data. The graph network comprises at least one node and at least one edge, each node is provided with a corresponding characteristic, and the edge is used for representing the connection relation between different nodes.
Graph neural network: the term refers to a general name of a model applied to a Graph Network by a neural Network, and the Graph neural Network includes a Graph Convolutional neural Network (GCN), a Graph attention Network, and the like. The graph neural network is used for predicting the class of the graph according to the structural features of the graph. In particular, the graph neural network may include one or more feature extraction layers. The feature extraction layer is, for example, Graph Convolution Layers (GCL). The feature extraction layer is used for extracting structural features of the graph. If the two graphs are isomorphic, then the structural features of the graphs of the two graphs will be similar after passing through the feature extraction layer. If the two graphs are heterogeneous, the structural features of the graphs of the two graphs will be different after passing through the feature extraction layer. Thus, the graph neural network is able to map graph structures with homogeneous properties into the same representation domain and output the same classes.
Graph convolution neural network: is a type of graph neural network that employs graph convolution. The atlas neural network includes at least one atlas layer. The graph convolution layer functions similarly to a feature extractor, where the object of feature extraction is a graph and the extracted features are structural features contained in the graph. Specifically, the graph convolution layer includes a plurality of convolution operators, the convolution operators are also called convolution kernels, the convolution kernels may be essentially a weight matrix, weight values in the weight matrix are obtained through a model training stage, and each weight matrix formed by the trained weight values may be used to extract features from an input graph, so that the graph convolution neural network performs correct prediction in an application stage.
The function of the graph convolution layer for realizing feature extraction is realized through graph convolution processing. The graph convolution process is an operation of performing nonlinear transformation on input data. For the first map convolution layer of the map convolution neural network, the input data of the map convolution processing is a map; for the second to last convolutional layers, the input data for the graph convolutional layer processing is the output result of the previous graph convolutional layer.
The scheme of the embodiment of the application comprises a training phase and a using phase of a convolutional neural network, and fig. 1 shows a training device 101 and a computer system using the device 102 of the convolutional neural network provided by an exemplary embodiment of the application. As shown in fig. 1, a convolutional neural network is obtained by training a training device 101 of the convolutional neural network, and the convolutional neural network is transmitted to a user device 102, where the user device 102 of the convolutional neural network can use the convolutional neural network.
The training device 101 and the using device 102 of the graph convolutional neural network may be computer devices with machine learning capability, for example, the computer devices may be terminals or servers.
Optionally, the training device 101 and the using device 102 of the graph convolution neural network may be the same computer device, or the training device 101 and the using device 102 of the graph convolution neural network may be different computer devices. Also, when the training device 101 of the convolutional neural network and the usage device 102 of the convolutional neural network are different devices, the training device 101 of the convolutional neural network and the usage device 102 of the convolutional neural network may be the same type of device, for example, the training device 101 of the convolutional neural network and the usage device 102 of the convolutional neural network may both be servers; alternatively, the training device 101 of the convolutional neural network and the using device 102 of the convolutional neural network may be different types of devices. The server may be an independent physical server, a server cluster or a distributed system formed by a plurality of physical servers, or a cloud server providing basic cloud computing services such as a cloud service, a cloud database, cloud computing, a cloud function, cloud storage, a Network service, cloud communication, a middleware service, a domain name service, a security service, a CDN (Content Delivery Network), a big data and artificial intelligence platform. The terminal may be, but is not limited to, a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart watch, and the like. The terminal and the server may be directly or indirectly connected through wired or wireless communication, and the application is not limited herein.
To improve the accuracy of infectious disease prediction, fig. 2 is a flowchart of an infectious disease prediction method according to an exemplary embodiment of the present application. This embodiment is exemplified by the method being performed by the terminal (using the device 102) shown in fig. 1, and the method includes:
step 220, acquiring first physiological information and first position information of a first user;
the first user: refers to a user making infectious disease predictions. Optionally, the first user is a user who is not determined to have the infectious disease, that is, the first user does not carry a label of the infectious disease (the label is used for determining whether the user has the infectious disease); optionally, the first user is a user who has determined whether or not to have an infectious disease, at which point the first user again conducts an infectious disease prediction for verifying the determined outcome.
First physiological information: the first user is physiological information related to the infectious disease, and illustratively, the physiological information related to the infectious disease includes at least one of sex, age, symptom, laboratory test index information, and imaging examination information.
The first position information: indicating the location of the first user. In this application, the minimum unit of the location indicated by the location information is not limited, for example, the first location information indicates that the location where the first user is located is shenzhen city, or the first location information indicates that the location where the first user is located is a nan garden street in shenzhen city, or the first location information indicates that the location where the first user is located is a house with a nan garden street number of 5 in shenzhen city.
Step 240, adding a first user node in the basic graph network of the infectious disease based on the first physiological information and the first position information to obtain an updated graph network;
basic graph network: refers to a reusable graph network related to infectious disease, and the basic graph network comprises at least one user node and/or at least one region node, wherein the user node identifies a user suspected to have infectious disease, and the region node identifies a region where the user suspected to have infectious disease is located.
Optionally, the region refers to an administrative region obtained based on administrative planning; optionally, the area refers to an area obtained based on a specific division, for example, in the period of infectious diseases, an area around the a village for 3km is marked as a high risk, and the high risk area is an area where a user suspected to have infectious diseases is located, i.e., an area node is identified.
Specifically, the first user node is added to the basic graph network of the infectious disease, which is described in the following four possible embodiments.
Step 260, inputting the updated graph network into a graph convolution neural network, and predicting to obtain a first user feature vector of the first user;
and the graph convolution neural network is used for extracting the features of the updated graph network and predicting to obtain a first user feature vector of the first user. The first user feature vector is a feature vector obtained based on first physiological information and first location information of the first user. Specifically, the updated graph network is input into the graph convolution neural network, and the first user feature vector of the first user is obtained through prediction, which is described in detail in the following four possible implementation manners.
And step 280, predicting the probability of the first user having the infectious disease through the probability calculation network based on the first user feature vector.
In one embodiment, the first user feature vector is normalized by a normalization function to obtain a value in the range of [0, 1], which is the probability that the first user has an infectious disease. Optionally, the normalization function is a sigmoid function.
In one embodiment, fig. 3 illustrates a product interface diagram to which the infectious disease prediction method illustrated in fig. 2 is applied, wherein the risk score 301 is displayed as the prevalence probability of the first patient, the second patient, the third patient, and the fourth patient.
In summary, in the method, the basic graph network is updated through the first physiological information and the first position information of the first user, the first user feature vector is obtained, and the probability calculation network is used for converting the first user feature vector into the probability that the first user has the infectious disease.
A first possible implementation: in an alternative embodiment shown in fig. 2, fig. 4 is a flowchart illustrating an infectious disease prediction method provided in an exemplary embodiment of the present application, wherein step 240 may be replaced by: step 242-1 and step 242-2, step 260 may be replaced with step 262-1, step 262-2, step 262-3 and step 262-4.
Step 242-1, generating a first user node in the basic graph network based on the first physiological information;
the basic graph network comprises a first area node, and the first area node identifies a first area where a user suspected of having an infectious disease is located. Based on the first physiological information, the terminal generates a first user node in the basic graph network, and the first user node carries the first physiological information of the first user.
Step 242-2, performing an edge connecting operation on the first user node and the first region node in the basic graph network to form a first graph network under the condition that the position of the first user indicated by the first position information falls into the first region;
in an embodiment, the location where the first user of the first location information is located is a nan garden street of shenzhen city, and the first locale is shenzhen city, that is, the location where the first user indicated by the first location information is located falls into the first locale, and the first user node and the first locale node are connected in the basic graph network to form a first graph network (i.e., an updated graph network).
Schematically, fig. 5 shows a first graph network in which a first region node 501 and a first user node 502 are connected.
Step 262-1, obtaining an initial region feature vector based on historical diseased information carried by a first region node in the first graph network; obtaining an initial user feature vector based on first physiological information carried by a first user node in a first graph network;
optionally, the historical illness information includes a historical confirmed number of persons diagnosed with the infectious disease over a past period of time.
In one embodiment, the first region node is represented by d and the first user node is represented by u, based on the first region node in the first graph networkHistorical diseased information of the belt, and the terminal obtains the characteristic vector of the initial region
Figure BDA00033052031300000911
Obtaining an initial user feature vector based on first physiological information carried by a first user node in a first graph network
Figure BDA00033052031300000912
In one embodiment, the initial user feature vector includes at least one of a type-not feature (symptom) represented by 0 or 1, a numerical type feature (age) represented by a floating point number, and a category type feature (gender) represented by a one-hot vector.
Step 262-2, calculating to obtain a layer 1 region feature vector based on the initial region feature vector and the initial user feature vector; calculating to obtain a layer 1 user feature vector based on the initial region feature vector and the initial user feature vector;
in one embodiment, the update formula for the feature vector of the first region node is:
Figure BDA0003305203130000091
where, phi denotes the activation function,
Figure BDA0003305203130000092
representing the flow of information from the first regional node itself at layer i,
Figure BDA0003305203130000093
representing the flow of information from the first user node u at layer i;
Figure BDA0003305203130000094
Figure BDA0003305203130000095
wherein,
Figure BDA0003305203130000096
representing the feature vector of the i-1 st layer region,
Figure BDA0003305203130000097
representing the i-1 level user feature vector,
Figure BDA0003305203130000098
a first parameter matrix representing an ith level of the first region node,
Figure BDA0003305203130000099
a second parameter matrix representing an ith level of the first locale node.
In one embodiment, the update formula of the feature vector of the first user node is:
Figure BDA00033052031300000910
where, phi denotes the activation function,
Figure BDA0003305203130000101
representing the information flow from the first user node itself at layer i,
Figure BDA0003305203130000102
representing the flow of information from the first regional node at layer i.
Figure BDA0003305203130000103
Figure BDA0003305203130000104
Wherein,
Figure BDA0003305203130000105
representing the feature vector of the i-1 st layer region,
Figure BDA0003305203130000106
representing the i-1 level user feature vector,
Figure BDA0003305203130000107
a third parameter matrix representing an ith layer of the first user node,
Figure BDA0003305203130000108
a fourth parameter matrix representing an ith layer of the first user node.
When i is 1, the layer 1 region feature vector
Figure BDA0003305203130000109
Is based on the initial region feature vector
Figure BDA00033052031300001010
And initial user feature vector
Figure BDA00033052031300001011
Resulting, layer 1 user feature vectors
Figure BDA00033052031300001012
Is based on the initial region feature vector
Figure BDA00033052031300001013
And initial user feature vector
Figure BDA00033052031300001014
And (4) obtaining the product.
262-3, calculating to obtain an ith layer region feature vector based on the i-1 layer region feature vector and the i-1 layer user feature vector; calculating to obtain an ith layer user feature vector based on the ith-1 layer region feature vector and the ith-1 layer user feature vector;
with reference to the above formulas (1) to (6), i-th layer region feature vector
Figure BDA00033052031300001015
Is based on the i-1 layer region feature vector
Figure BDA00033052031300001016
And the i-1 layer user feature vector
Figure BDA00033052031300001017
The obtained i-th layer user feature vector
Figure BDA00033052031300001018
Is based on the initial region feature vector
Figure BDA00033052031300001019
And the i-th layer user feature vector
Figure BDA00033052031300001020
And (4) obtaining the product.
And 262-4, outputting the characteristic vector of the k-th layer user as the first characteristic vector of the user.
In one embodiment, the graph convolution neural network includes k graph convolution layers, the terminal outputs the k-th layer user feature vector output by the k-th graph convolution layer as the first user feature vector, k is an integer greater than 1, and i is an integer greater than 0 and less than k.
In summary, in the method, the updated first graph network is obtained by updating the basic graph network, the first graph network includes the first area node and the first user node, data integration of the physiological information and the position information of the user is achieved through the graph network, and further the prevalence probability of the first user is obtained through the graph convolution neural network and the probability calculation network.
A second possible implementation: in an alternative embodiment shown in fig. 2, fig. 6 is a flowchart illustrating an infectious disease prediction method provided in an exemplary embodiment of the present application, wherein step 240 may be replaced by: step 244-1 and step 244-2, step 260 may be replaced with step 264-1, step 264-2, step 264-3 and step 264-4.
Step 244-1, generating a first user node in the base graph network based on the first physiological information;
the basic graph network comprises a first area node and a second area node connected with the first area node, the first area node identifies a first area where a user suspected to be infected is located, and the second area node identifies a second area where the user suspected to be infected is located.
And the population flow number between the second region node and the first region node reaches a number threshold, and/or the distances between the regions respectively indicated by the second region node and the first region node are lower than a first distance threshold.
Illustratively, the first region is Shenzhen city, the second region is Guangzhou city, the number of population flows between the Shenzhen city and the Guangzhou city reaches a quantity threshold value, the distance between the Shenzhen city and the Guangzhou city is lower than a first distance threshold value, and the first region node and the second region node are connected in the basic graph network.
In one embodiment, the terminal generates a first user node in the basic graph network based on the first physiological information, and the first user node carries the first physiological information.
Step 244-2, under the condition that the position of the first user indicated by the first position information falls into the first region, performing edge connection operation on the first user node and the first region node in the basic graph network to form a second graph network;
in an embodiment, the location where the first user of the first location information is located is a nan garden street of shenzhen city, and the first locale is shenzhen city, that is, the location where the first user indicated by the first location information is located falls into the first locale, and the first user node and the first locale node are subjected to edge connection operation in the basic graph network to form a second graph network (i.e., an updated graph network).
Schematically, fig. 7 shows a second graph network in which a first regional node 701 is connected to a second regional node 702, and the first regional node 701 is connected to a first user node 703.
Step 264-1, obtaining an initial region feature vector of a first region node based on historical diseased information carried by the first region node in the second graph network; obtaining an initial region feature vector of a second region node based on historical diseased information carried by the second region node in a second graph network; obtaining an initial user feature vector based on first physiological information carried by a first user node in a second graph network;
in one embodiment, the first area node is represented by d, the second area node is represented by d', the first user node is represented by u, and the terminal obtains an initial area feature vector of the first area node based on historical diseased information carried by the first area node in the second graph network
Figure BDA0003305203130000111
Based on the history ill information carried by the second region node in the second graph network, the terminal obtains the initial region feature vector of the second region node
Figure BDA0003305203130000112
Obtaining an initial user feature vector based on first physiological information carried by a first user node in a second graph network
Figure BDA0003305203130000113
Step 264-2, calculating to obtain a layer 1 area feature vector of the first area node based on the initial area feature vector of the first area node, the initial area feature vector of the second area node and the initial user feature vector by combining the first weight; calculating to obtain a layer 1 user feature vector based on the initial region feature vector and the initial user feature vector of the first region node;
wherein the first weight is associated with a number of population flows between the first region node and the second region node.
In one embodiment, the update formula for the feature vector of the first region node is:
Figure BDA0003305203130000121
where, phi denotes the activation function,
Figure BDA0003305203130000122
representing the information flow from the first regional node d itself at layer i,
Figure BDA0003305203130000123
representing the flow of information from the first user node u at layer i,
Figure BDA0003305203130000124
representing the flow of information, α, from a second regional node d' of the i-th layerd←d′Representing a first weight, Nd(d) Representing the set of all the regional nodes connected to the first regional node d.
Figure BDA0003305203130000125
Figure BDA0003305203130000126
Figure BDA0003305203130000127
Figure BDA0003305203130000128
Wherein d' represents a region node (including a second region node) to which the first region node d is connected, wd←d′Indicating the edge weight of the second region node d' to the first region node d (based on the amount of population movement between the first region and the second region).
Figure BDA0003305203130000129
A fifth parameter matrix representing the ith level of the first region node,
Figure BDA00033052031300001210
a sixth parameter matrix representing the ith level of the first region node,
Figure BDA00033052031300001211
a seventh parameter matrix representing an ith level of the first zone node.
In one embodiment, the update formula of the feature vector of the first user node is:
Figure BDA00033052031300001212
where, phi denotes the activation function,
Figure BDA00033052031300001213
representing the information flow from the first user node itself at layer i,
Figure BDA00033052031300001214
representing the flow of information from the first regional node at layer i.
Figure BDA00033052031300001215
Figure BDA00033052031300001216
Wherein,
Figure BDA00033052031300001217
a level i-1 locale feature vector representing a first locale node,
Figure BDA00033052031300001218
representing the i-1 level user feature vector,
Figure BDA00033052031300001219
an eighth parameter representing an i-th layer of the first user nodeThe matrix is a matrix of a plurality of matrices,
Figure BDA00033052031300001220
a ninth parameter matrix representing an ith layer of the first user node.
When i is 1, the layer 1 region feature vector of the first region node
Figure BDA00033052031300001221
Is based on the initial region feature vector of the first region node
Figure BDA0003305203130000131
Initial region feature vector of second region node
Figure BDA0003305203130000132
And initial user feature vector
Figure BDA0003305203130000133
Resulting, layer 1 user feature vectors
Figure BDA0003305203130000134
Is based on the initial region feature vector
Figure BDA0003305203130000135
And initial user feature vector
Figure BDA0003305203130000136
And (4) obtaining the product.
Step 264-3, calculating to obtain the ith layer area feature vector of the first area node based on the i-1 layer area feature vector of the first area node, the i-1 layer area feature vector of the second area node and the i-1 layer user feature vector by combining the first weight; calculating to obtain an ith layer user feature vector based on the ith-1 layer region feature vector and the ith-1 layer user feature vector of the first region node;
with reference to the above equations (7) to (14), the ith layer region feature vector of the first region node
Figure BDA0003305203130000137
Is based on the i-1 layer region feature vector of the first region node
Figure BDA0003305203130000138
Layer i-1 region feature vector of second region node
Figure BDA0003305203130000139
And the i-1 layer user feature vector
Figure BDA00033052031300001310
Obtained in combination with the first weight. The i-th layer user feature vector is an i-1-th layer region feature vector based on the first region node
Figure BDA00033052031300001311
And the i-1 layer user feature vector
Figure BDA00033052031300001312
And (4) obtaining the product.
And step 264-4, outputting the characteristic vector of the k-th layer user as the first user characteristic vector.
In one embodiment, the graph convolution neural network includes k graph convolution layers, the terminal outputs the k-th layer user feature vector output by the k-th graph convolution layer as the first user feature vector, k is an integer greater than 1, and i is an integer greater than 0 and less than k.
In summary, in the method, the updated second graph network is obtained by updating the basic graph network, and the second graph network includes the first area node, the second area node and the first user node, so that data integration of the physiological information and the position information of the user through the graph network is realized, and further, the prevalence probability of the first user can be obtained through the graph convolution neural network and the probability calculation network.
A third possible implementation: in an alternative embodiment shown in fig. 2, fig. 8 is a flowchart illustrating an infectious disease prediction method provided in an exemplary embodiment of the present application, wherein step 240 may be replaced by: step 246-1 and step 246-2, step 260 may be replaced with step 266-1, step 266-2, step 266-3, and step 266-4.
Step 246-1, generating a first user node in the base graph network based on the first physiological information;
the basic graph network comprises a first area node and a second user node connected with the first area node, the first area node identifies a first area where a user suspected to be infected with the infectious disease is located, and the second user node identifies a second user suspected to be infected with the infectious disease.
The second user node: refers to the existing user node before the network of the basic graph is updated.
In one embodiment, the terminal generates a first user node in the basic graph network based on the first physiological information, and the first user node carries the first physiological information.
Step 246-2, under the condition that the position of the first user indicated by the first position information falls into the first region, performing edge connecting operation on the first user node and the first region node in the basic graph network to form a third graph network;
in an embodiment, the location where the first user of the first location information is located is a nan garden street of shenzhen city, and the first locale is shenzhen city, that is, the location where the first user indicated by the first location information is located falls into the first locale, and the first user node and the first locale node are subjected to edge connection operation in the basic graph network to form a third graph network (i.e., an updated graph network).
Schematically, fig. 9 shows a third graph network in which a first region node 901 is connected to a first user node 902, and a first region node 901 is connected to a second user node 903. And the position of the second user indicated by the second user node falls into the first area.
266-1, obtaining an initial region feature vector based on the historical diseased information carried by the first region node in the third graph network; obtaining an initial user feature vector of a first user node based on first physiological information carried by the first user node in a third graph network; obtaining an initial user feature vector of a second user node based on second physiological information carried by the second user node in a third graph network;
in one embodiment, the first region node is represented by d, and one user node in the user node set formed by the first user node and at least one second user node is represented by uiAnd representing, the first user node adopts u, and based on the historical diseased information carried by the first area node in the third graph network, the terminal obtains the initial area characteristic vector of the first area node
Figure BDA0003305203130000141
Obtaining an initial user feature vector based on first physiological information carried by a first user node in a third graph network
Figure BDA0003305203130000142
266-2, calculating to obtain a layer 1 region feature vector based on the initial region feature vector, the initial user feature vector of the first user node and the initial user feature vector of the second user node; calculating to obtain a layer 1 user feature vector of the first user node based on the initial region feature vector and the initial user feature vector of the first user node;
in one embodiment, the update formula for the feature vector of the first region node is:
Figure BDA0003305203130000143
where, phi denotes the activation function,
Figure BDA0003305203130000144
representing the information flow from the first regional node d itself at layer i,
Figure BDA0003305203130000145
represents one user node u from the i-th layer in a user node set (including a first user node and a second user node) connected to a first region node diOf the information stream, Nu(d) Representing and first regionAnd d is connected with all the user nodes.
Figure BDA0003305203130000151
Figure BDA0003305203130000152
Wherein,
Figure BDA0003305203130000153
representing the feature vector of the i-1 st layer region,
Figure BDA0003305203130000154
a user feature vector representing one user node in the set of user nodes of layer i-1,
Figure BDA0003305203130000155
a tenth parameter matrix representing the ith level of the first region node,
Figure BDA0003305203130000156
an eleventh parameter matrix representing an ith level of the first zone node.
In one embodiment, the update formula of the feature vector of the first user node is:
Figure BDA0003305203130000157
where, phi denotes the activation function,
Figure BDA0003305203130000158
representing the information flow from the first user node itself at layer i,
Figure BDA0003305203130000159
representing the flow of information from the first regional node at layer i.
Figure BDA00033052031300001510
Figure BDA00033052031300001511
Wherein,
Figure BDA00033052031300001512
representing the feature vector of the i-1 st layer region,
Figure BDA00033052031300001513
representing the i-1 level user feature vector,
Figure BDA00033052031300001514
a twelfth parameter matrix representing an ith layer of the first user node,
Figure BDA00033052031300001515
a thirteenth parameter matrix representing an ith layer of the first user node.
When i is 1, the layer 1 region feature vector of the first region node
Figure BDA00033052031300001516
Is based on the initial region feature vector of the first region node
Figure BDA00033052031300001517
The initial user characteristic vector of the first user node and the initial user characteristic vector of the second user node are obtained, and the layer 1 user characteristic vector of the first user node
Figure BDA00033052031300001518
Is based on the initial region feature vector
Figure BDA00033052031300001519
And initial user feature vector
Figure BDA00033052031300001520
And (4) obtaining the product.
266-3, calculating to obtain the ith layer area feature vector based on the ith-1 layer area feature vector, the ith-1 layer user feature vector of the first user node and the ith-1 layer user feature vector of the second user node; calculating to obtain an ith layer user feature vector of the first user node based on the ith-1 layer region feature vector and the ith-1 layer user feature vector of the first user node;
with reference to the above equations (15) to (20), the ith layer region feature vector
Figure BDA00033052031300001521
Is based on the i-1 layer region feature vector of the first region node
Figure BDA00033052031300001522
And obtaining a user feature vector of a user node set formed by the first user node and the second user node of the i-1 layer. The i-th layer user feature vector is an i-1-th layer region feature vector based on the first region node
Figure BDA00033052031300001523
And the i-1 layer user feature vector
Figure BDA00033052031300001524
And (4) obtaining the product.
And step 266-4, outputting the k-th layer user feature vector of the first user node as the first user feature vector.
In one embodiment, the graph convolution neural network includes k graph convolution layers, the terminal outputs a k-th layer user feature vector of a first user node output by the k-th graph convolution layer as a first user feature vector, k is an integer greater than 1, and i is an integer greater than 0 and less than k.
In summary, in the method, the updated third graph network is obtained by updating the basic graph network, and the third graph network includes the first user node and the second user node, so that data integration of the physiological information and the position information of the user through the graph network is realized, and further, the prevalence probability of the first user can be obtained through the graph convolution neural network and the probability calculation network.
In a fourth possible implementation manner, based on the alternative embodiment shown in fig. 2, fig. 10 shows a flowchart of an infectious disease prediction method provided in an exemplary embodiment of the present application, wherein step 240 may be replaced by: step 248-1 and step 248-2, step 260 may be replaced with step 268-1, step 268-2, step 268-3, and step 268-4.
248-1, generating a first user node in the basic graph network based on the first physiological information;
the basic graph network comprises a third user node, the third user node identifies a third user suspected to be infected with the infectious disease, and the third user node carries third physiological information and third position information of the third user;
248-2, under the condition that the similarity between the first physiological information and the third physiological information is greater than a similarity threshold value and/or the distance between the positions respectively indicated by the first position information and the third position information is less than a second distance threshold value, performing edge connection operation on the first user node and the third user node in the basic graph network to form a fourth graph network;
in one embodiment, the first physiological information indicates a symptom of the first user, the third physiological information indicates a symptom of the third user, and in a case that a similarity between the symptom of the first user and the symptom of the third user is greater than a similarity threshold and a distance between a location where the first user is located and a location where the third user is located is less than a second distance threshold, the first user node and the third user node are connected in the basic graph network to form a fourth graph network, which is schematically illustrated in fig. 11, and the first user node 1101 is connected to the third user node 1102.
268-1, obtaining an initial user feature vector of the first user node based on the first physiological information carried by the first user node in the fourth graph network; obtaining an initial user feature vector of a third user node based on third physiological information carried by the third user node in the fourth graph network;
in one embodiment, a first user node is represented by m, a second user node is represented by n, and an initial user feature vector of the first user node m is obtained based on first physiological information carried by the first user node
Figure BDA0003305203130000161
Obtaining an initial user feature vector of a third user node n based on third physiological information carried by the third user node
Figure BDA0003305203130000162
Step 268-2, based on the initial user feature vector of the first user node and the initial user feature vector of the third user node, and in combination with the second weight, calculating to obtain a layer 1 user feature vector of the first user node and a layer 1 user feature vector of the third user node;
wherein the second weight is associated with at least one of a similarity between the first physiological information and the third physiological information, and a distance between positions indicated by the first positional information and the third positional information, respectively;
in one embodiment, the update formula of the feature vector of the first user node is:
Figure BDA0003305203130000171
where σ denotes the activation function, αmnIs a second weight, W14A fourteenth weight, W, representing the first user node at level i15A fifteenth weight representing an ith layer third user node.
Figure BDA0003305203130000172
Figure BDA0003305203130000173
Wherein, aTRepresenting a particular parameter matrix, as W16Training together in a training process, wherein N (m) represents a user node set connected with a first user node,
Figure BDA0003305203130000174
denotes the normalized distance, δdA second distance threshold value is indicated which is,
Figure BDA0003305203130000175
indicating the distance, s, at which the first user and the third user are locatedmnIndicates the degree of similarity between the first physiological information of the first user and the third physiological information of the third user, [ 2 ]]Representing vector stitching, W16A sixteenth parameter matrix representing the first user node of the ith layer.
Figure BDA0003305203130000176
Figure BDA0003305203130000177
wfWeight, S, representing a physiological information f (a physiological information of the first physiological information)mFirst physiological information representing a first user, SnThird physiological information representing a third user, NfIndicating the number of patients diagnosed for the presence of the physiological information f, and N indicating the number of patients currently diagnosed with infectious disease.
Similarly, the eigenvector of the second user node n can be calculated by referring to equation (21) above
Figure BDA0003305203130000178
268-3, calculating to obtain the ith layer user characteristic vector of the first user node and the ith layer user characteristic vector of the third user node based on the ith-1 layer user characteristic vector of the first user node and the ith-1 layer user characteristic vector of the third user node by combining the second weight;
and (4) combining the reference expressions (21), (22) and (23), calculating the ith layer user characteristic vector of the first user node and the ith layer user characteristic vector of the third user node.
And step 268-4, outputting the k-th layer user characteristic vector of the first user node as the first user characteristic vector.
In one embodiment, the graph convolution neural network includes k graph convolution layers, the terminal outputs a k-th layer user feature vector of a first user node output by the k-th graph convolution layer as a first user feature vector, k is an integer greater than 1, and i is an integer greater than 0 and less than k.
In summary, in the method, the updated fourth graph network is obtained by updating the basic graph network, and the fourth graph network includes the first user node and the third user node, so that data integration of the physiological information and the position information of the user through the graph network is realized, and further, the prevalence probability of the first user can be obtained through the graph convolution neural network and the probability calculation network.
It should be noted that the first possible implementation manner to the fourth possible implementation manner may be combined, and fig. 12 schematically illustrates a schematic diagram of a fifth graph network and a sixth graph network provided in an exemplary embodiment of the present application.
The fifth graph network (a) is obtained based on the combination of the first graph network, the second graph network, the third graph network and the fourth graph network. First region node 1201 is connected to first user node 1203, first region node 1201 is connected to second region node 1202, first region node 1201 is further connected to at least one second user node, and first user node 1203 is further connected to at least one third user node.
The sixth graph network (b) is obtained based on a combination of the first graph network, the second graph network, and the third graph network, the first region node 1201 is connected to the first user node 1203, the first region node 1201 is connected to the second region node 1202, and the first region node 1201 is further connected to at least one second user node.
Another point worth mentioning is that all the parameter matrices described above are updated by model training.
In one embodiment, based on the alternative embodiment shown in FIG. 2, step 280 may be replaced with:
s1: splicing the first user characteristic vector and the initial user characteristic vector of the first user node to obtain a first intermediate vector;
s2: inputting the first intermediate vector into a full-connection network to obtain a second intermediate vector;
s3: the second intermediate vector is input into an activation function to predict a probability that the first user has the infectious disease.
Illustratively, the formula for calculating the probability of the first user having an infectious disease is as follows:
Figure BDA0003305203130000181
where FC denotes a fully connected network, σ denotes a sigmoid function,
Figure BDA0003305203130000182
a user feature vector representing a k-th layer output of the first user node,
Figure BDA0003305203130000183
an initial feature vector representing a first user node, [ 2 ]]Representing vector stitching.
In one embodiment, based on the alternative embodiment shown in fig. 2, step 280 further includes:
the first user node is deleted in the updated graph network. The first user node is a user node generated in the basic graph network based on first physiological information of a first user.
To train the above graph convolutional neural network, fig. 13 is a flowchart illustrating a training method of a graph convolutional neural network according to an exemplary embodiment of the present application, which is applied to a terminal (training device 101) of the graph convolutional neural network shown in fig. 1, and the training method adopts a semi-supervised mode, and the method includes:
step 1301, acquiring sample physiological information and sample position information of n sample users;
wherein the n sample users include n1A first sample user and n2And a second sample user, the first sample user carrying a label for determining whether the first sample user has an infectious disease, the second sample user not carrying a label.
In one embodiment, the terminal obtains sample physiological information and sample location information for n sample users.
Step 1302, inputting a sample graph network corresponding to sample physiological information and sample position information of n sample users into a graph convolution neural network, and obtaining n sample feature vectors of the n sample users through computation of the graph convolution neural network;
in one embodiment, the terminal inputs a sample map network corresponding to sample physiological information and sample position information of n sample users into a map convolution neural network, and n sample feature vectors of the n sample users are obtained through calculation of the map convolution neural network.
Optionally, the terminal generates n sample nodes based on the sample physiological information of n sample users, the positions indicated by the n sample position information fall into a first region indicated by a first region node, and the terminal connects the n sample nodes with the first region node to obtain a first sample graph network;
optionally, the terminal generates n sample nodes based on the sample physiological information of the n sample users, and the terminal connects the n sample nodes with the first area node or the second area node based on the position indicated by the n sample position information falling into the first area indicated by the first area node or the second area indicated by the second area node to obtain a second sample graph network;
optionally, the terminal generates n sample nodes based on the sample physiological information of the n sample users, the positions indicated by the n sample position information fall into the first region indicated by the first region node, the terminal connects the n sample nodes with the first region node, and forms a third sample graph network together with the second user node, and the position of the second user indicated by the second user node falls into the first region indicated by the first region node;
optionally, the terminal generates n sample nodes based on the sample physiological information of the n sample users, and based on that a distance between one sample user of the n sample users and a third user indicated by a third user node is lower than a second distance threshold, and/or for that a similarity between the physiological information of one sample user of the n sample users and the third physiological information of the third user indicated by the third user node reaches a similarity threshold, the terminal connects the n sample nodes and the third user node to obtain a fourth sample graph network.
With combined reference to fig. 5, 7, 9, 11, and 12, the first, second, third, and fourth sample graph networks are similar to the first, second, third, and fourth graph networks, respectively;
it should be noted that a new sample graph network may be further constructed based on the first sample graph network, the second sample graph network, the third sample graph network, and the fourth sample graph network, and the new sample graph network may be similar to the fifth graph network and the sixth graph network.
Step 1303, for n1One of the first sample users obtains the predicted probability that the first sample user has the infectious disease through a probability calculation network based on the sample feature vectors of the first sample users;
in one embodiment, for n1And one of the first sample users, the terminal obtains the predicted probability that the first sample user has the infectious disease through the probability calculation network based on the sample feature vector of the first sample user.
In one embodiment, the terminal splices the sample characteristic vector of the first sample user and the initial characteristic vector of the first sample user to obtain a first sample intermediate vector; then, the terminal inputs the first sample intermediate vector into a full-connection network to obtain a second sample intermediate vector; and finally, the terminal inputs the second sample intermediate vector into the activation function to obtain the probability that the first sample user has the infectious disease.
Step 1304, based on the prediction probability and n1And (3) training the convolutional neural network of the graph according to the loss value of the proportion of the diseased sample users in the first sample users.
In one embodiment, the terminal bases the prediction probability on n1And (3) training the convolutional neural network of the graph according to the loss value of the proportion of the diseased sample users in the first sample users.
In summary, the training method sets n of n sample users1A first sample user has a label, n2And (4) training the graph convolutional neural network when the second sample user does not have the label, so that the sample user without the label still participates in the integrated calculation of the data, but the prediction probability of the sample user without the label does not participate in the training of the graph convolutional neural network.
To train the above graph convolutional neural network, fig. 14 is a flowchart illustrating a training method of the graph convolutional neural network provided in an exemplary embodiment of the present application, which is applied to a terminal (training device 101) of the graph convolutional neural network shown in fig. 1, and the training method adopts a supervised mode, and the method includes:
step 1401, obtain n1Sample physiological information and sample location information for a first sample user;
wherein the first sample user carries a tag for determining whether the first sample user has an infectious disease; in one embodiment, the terminal obtains n1Sample physiological information and sample location information for a first sample user
Step 1402, compare n with1Inputting the sample graph network corresponding to the sample physiological information and the sample position information of the first sample user into the graph convolution neural network, and obtaining n through the computation of the graph convolution neural network1N of a first sample user1A sample feature vector;
in one embodiment, the terminal will n1Inputting the sample physiological information and the sample position information of the first sample user into a graph convolution neural network, and calculating by the graph convolution neural network to obtain n1N of a first sample user1A sample feature vector.
Optionally, the terminal is based on n1Generating n sample nodes according to the sample physiological information of the first sample user, enabling the positions indicated by the n sample position information to fall into a first area indicated by a first area node, and enabling the n sample nodes and the first area node to be connected by the terminal to obtain a fifth sample graph network;
optionally, the terminal is based on n1Generating n sample nodes according to the sample physiological information of the first sample user, enabling the positions indicated by the n sample position information to fall into a first region indicated by a first region node or a second region indicated by a second region node, and enabling the terminal to connect the n sample nodes with the first region node or the second region node to obtain a sixth sample graph network;
optionally, the terminal is based on n1The method comprises the steps that n sample nodes are generated according to sample physiological information of a first sample user, the positions indicated by n sample position information fall into a first area indicated by first area nodes, the n sample nodes are connected with the first area nodes by a terminal and form a seventh sample graph network together with second user nodes, and the position of the second user indicated by the second user nodes falls into the first area indicated by the first area nodes;
optionally, the terminal is based on n1And generating n sample nodes according to the sample physiological information of the n first sample users, wherein the n sample nodes are connected with the third user node by the terminal to obtain an eighth sample graph network based on that the distance between one sample user of the n1 first sample users and the third user indicated by the third user node is lower than a second distance threshold, and/or the similarity between the physiological information of one sample user of the n1 first sample users and the third physiological information of the third user indicated by the third user node reaches a similarity threshold.
With combined reference to fig. 5, 7, 9, 11 and 12, fifth, sixth, seventh and eighth sample graph networks are similar to the first, second, third and fourth graph networks, respectively;
it should be noted that a new sample graph network may be further constructed based on the fifth sample graph network, the sixth sample graph network, the seventh sample graph network, and the eighth sample graph network, and the new sample graph network may be similar to the fifth graph network and the sixth graph network.
Step 1403, for n1One of the first sample users obtains the predicted probability that the first sample user has the infectious disease through a probability calculation network based on the sample feature vectors of the first sample users;
in one embodiment, the terminal is for n1And one of the first sample users obtains the predicted probability that the first sample user has the infectious disease through the probability calculation network based on the sample feature vectors of the first sample user.
In one embodiment, the terminal splices the sample characteristic vector of the first sample user and the initial characteristic vector of the first sample user to obtain a first sample intermediate vector; then, the terminal inputs the first sample intermediate vector into a full-connection network to obtain a second sample intermediate vector; and finally, the terminal inputs the second sample intermediate vector into the activation function to obtain the probability that the first sample user has the infectious disease.
Step 1404, based on the prediction probability and n1And (3) training the convolutional neural network of the graph according to the loss value of the proportion of the diseased sample users in the first sample users.
In one embodiment, the terminal bases the prediction probability on n1And (3) training the convolutional neural network of the graph according to the loss value of the proportion of the diseased sample users in the first sample users.
In summary, the training method utilizes n carrying labels1And the first sample user trains the graph convolution neural network to realize the training of the graph convolution neural network.
In one embodiment, fig. 15 shows a block diagram of an infectious disease prediction apparatus provided in an exemplary embodiment of the present application, the apparatus including:
the acquiring module 1501 is configured to acquire first physiological information and first position information of a first user, where the first physiological information is physiological information of the first user related to an infectious disease, and the first position information indicates a position where the first user is located;
an updating module 1502, configured to add a first user node in a basic graph network of an infectious disease based on the first physiological information and the first location information, to obtain an updated graph network; the basic graph network comprises at least one user node and/or at least one area node, wherein the user node identifies a user suspected to be infected with the infectious disease, and the area node identifies an area where the user suspected to be infected with the infectious disease is located;
the prediction module 1503 is configured to input the updated graph network into a graph convolution neural network, and predict to obtain a first user feature vector of the first user; and predicting the probability that the first user has the infectious disease through the probability computation network based on the first user feature vector.
In an alternative embodiment, the base graph network includes a first region node and the updated graph network includes a first graph network, the first region node identifying a first region in which a user suspected of having an infectious disease is located.
The updating module 1502 is further configured to generate a first user node in the base graph network based on the first physiological information.
The updating module 1502 is further configured to perform an edge connecting operation on the first user node and the first area node in the basic graph network to form the first graph network when the location of the first user indicated by the first location information falls into the first area.
In an alternative embodiment, the atlas neural network has k atlas layers.
In an optional embodiment, the prediction module 1503 is further configured to obtain an initial region feature vector based on historical diseased information carried by a first region node in the first graph network; and obtaining an initial user feature vector based on first physiological information carried by a first user node in the first graph network.
In an optional embodiment, the prediction module 1503 is further configured to calculate a layer 1 region feature vector based on the initial region feature vector and the initial user feature vector; and calculating to obtain a layer 1 user feature vector based on the initial region feature vector and the initial user feature vector.
In an optional embodiment, the prediction module 1503 is further configured to calculate an ith layer region feature vector based on the i-1 layer region feature vector and the i-1 layer user feature vector; and calculating to obtain the ith layer user feature vector based on the ith-1 layer region feature vector and the ith-1 layer user feature vector.
In an alternative embodiment, the prediction module 1503 is further configured to output the k-th layer user feature vector as the first user feature vector.
In an alternative embodiment, the base graph network includes a first region node and a second region node coupled to the first region node, and the updated graph network includes a second graph network, the first region node identifying a first region in which a user suspected of having an infection is located, the second region node identifying a second region in which the user suspected of having an infection is located.
In an optional embodiment, the updating module 1502 is further configured to generate a first user node in the base graph network based on the first physiological information.
In an optional embodiment, the updating module 1502 is further configured to perform an edge connecting operation on the first user node and the first region node in the base graph network to form a second graph network, when the location of the first user indicated by the first location information falls into the first region.
The population flow number between the second region node and the first region node reaches a number threshold, and/or the distances between the regions respectively indicated by the second region node and the first region node are lower than a first distance threshold.
In an alternative embodiment, the atlas neural network includes k atlas layers.
In an optional embodiment, the prediction module 1503 is further configured to obtain an initial region feature vector of the first region node based on historical diseased information carried by the first region node in the second graph network; obtaining an initial region feature vector of a second region node based on historical diseased information carried by the second region node in a second graph network; and obtaining an initial user feature vector based on first physiological information carried by a first user node in a second graph network.
In an optional embodiment, the prediction module 1503 is further configured to calculate a layer 1 region feature vector of the first region node based on the initial region feature vector of the first region node, the initial region feature vector of the second region node, and the initial user feature vector, and by combining a first weight, where the first weight is associated with a population mobility number between the first region node and the second region node; calculating to obtain a layer 1 user feature vector based on the initial region feature vector and the initial user feature vector of the first region node;
in an optional embodiment, the prediction module 1503 is further configured to calculate, based on the i-1 th layer region feature vector of the first region node, the i-1 th layer region feature vector of the second region node, and the i-1 th layer user feature vector, and in combination with the first weight, obtain an i-th layer region feature vector of the first region node; and calculating to obtain an i-th layer user feature vector based on the i-1-th layer region feature vector and the i-1-th layer user feature vector of the first region node.
In an alternative embodiment, the prediction module 1503 is further configured to output the k-th layer user feature vector as the first user feature vector.
In an alternative embodiment, the base graph network includes a first region node identifying a first region in which a user suspected of having an infectious disease is located and a second user node connected to the first region node identifying a second user suspected of having an infectious disease, and the updated graph network includes a third graph network.
In an optional embodiment, the updating module 1502 is further configured to generate a first user node in the base graph network based on the first physiological information.
In an optional embodiment, the updating module 1502 is further configured to perform an edge connecting operation on the first user node and the first region node in the base graph network to form a third graph network, when the location of the first user indicated by the first location information falls into the first region.
And the position of the second user indicated by the second user node falls into the first area.
In an alternative embodiment, the atlas neural network has k atlas layers.
In an optional embodiment, the prediction module 1503 is further configured to obtain an initial region feature vector based on historical diseased information carried by a first region node in the third graph network; obtaining an initial user feature vector of a first user node based on first physiological information carried by the first user node in a third graph network; and obtaining an initial user feature vector of the second user node based on second physiological information carried by the second user node in the third graph network.
In an optional embodiment, the prediction module 1503 is further configured to calculate a layer 1 region feature vector based on the initial region feature vector, the initial user feature vector of the first user node, and the initial user feature vector of the second user node; and calculating to obtain a layer 1 user feature vector of the first user node based on the initial region feature vector and the initial user feature vector of the first user node.
In an optional embodiment, the prediction module 1503 is further configured to calculate an i-th layer region feature vector based on the i-1-th layer region feature vector, the i-1-th layer user feature vector of the first user node, and the i-1-th layer user feature vector of the second user node; and calculating to obtain the ith layer user feature vector of the first user node based on the ith-1 layer region feature vector and the ith-1 layer user feature vector of the first user node.
In an alternative embodiment, the prediction module 1503 is further configured to output the k-th layer user feature vector of the first user node as the first user feature vector.
In an alternative embodiment, the base graph network includes a third user node, and the updated graph network includes a fourth graph network, the third user node identifying a third user suspected of having an infection, the third user node carrying third physiological information and third location information of the third user.
In an optional embodiment, the updating module 1502 is further configured to generate a first user node in the base graph network based on the first physiological information.
In an optional embodiment, the updating module 1502 is further configured to perform an edge join operation on the first user node and the third user node in the base graph network to form a fourth graph network, when the similarity between the first physiological information and the third physiological information is greater than a similarity threshold, and/or a distance between the positions respectively indicated by the first position information and the third position information is smaller than a second distance threshold.
In an alternative embodiment, the atlas neural network includes k-layer atlas layers.
In an optional embodiment, the prediction module 1503 is further configured to obtain an initial user feature vector of the first user node based on the first physiological information carried by the first user node in the fourth graph network; and obtaining an initial user feature vector of the third user node based on third physiological information carried by the third user node in the fourth graph network.
In an optional embodiment, the prediction module 1503 is further configured to calculate a layer 1 user feature vector of the first user node and a layer 1 user feature vector of the third user node based on the initial user feature vector of the first user node and the initial user feature vector of the third user node, and by combining a second weight, where the second weight is associated with at least one of a similarity between the first physiological information and the third physiological information, and a distance between positions respectively indicated by the first position information and the third position information.
In an optional embodiment, the prediction module 1503 is further configured to calculate, based on the i-1 layer user feature vector of the first user node and the i-1 layer user feature vector of the third user node, the i-layer user feature vector of the first user node and the i-layer user feature vector of the third user node by combining the second weight.
In an alternative embodiment, the prediction module 1503 is further configured to output the k-th layer user feature vector of the first user node as the first user feature vector.
In an optional embodiment, the apparatus further comprises a deleting module 1504 for deleting the first user node in the updated graph network.
In an alternative embodiment, the probabilistic computing network includes an activation function.
In an optional embodiment, the prediction module 1503 is further configured to splice the first user feature vector and the initial user feature vector of the first user node to obtain a first intermediate vector.
In an alternative embodiment, the prediction module 1503 is further configured to input the first intermediate vector into a fully connected network to obtain a second intermediate vector.
In an alternative embodiment, the prediction module 1503 is further configured to input the second intermediate vector into an activation function to predict a probability that the first user has an infection.
In summary, the device updates the basic graph network through the first physiological information and the first position information of the first user, obtains the first user feature vector, and converts the first user feature vector into the probability that the first user has the infectious disease through the probability calculation network.
Fig. 16 is a block diagram illustrating a structure of an apparatus for training a convolutional neural network, which trains the convolutional neural network in a semi-supervised manner, according to an exemplary embodiment of the present application, and includes:
an obtaining module 1601, configured to obtain sample physiological information and sample location information of n sample users; the n sample users include n1A first sample user and n2A second sample user, the first sample user carrying a label for determining whether the first sample user has an infectious disease, the second sample user not carrying a label;
a calculating module 1602, configured to input a sample graph network corresponding to sample physiological information and sample position information of n sample users into a graph convolution neural network, and obtain n sample feature vectors of the n sample users through computation of the graph convolution neural network;
a calculation module 1602 for n1One of the first sample users obtains the predicted probability that the first sample user has the infectious disease through a probability calculation network based on the sample feature vectors of the first sample users;
a training module 1603 for based on the prediction probability and n1And (3) training the convolutional neural network of the graph according to the loss value of the proportion of the diseased sample users in the first sample users.
In an alternative embodiment, the probabilistic computing network includes an activation function.
In an optional embodiment, the calculating module 1602 is further configured to splice the sample feature vector of the first sample user and the initial feature vector of the first sample user to obtain a first sample intermediate vector.
In an alternative embodiment, the calculating module 1602 is further configured to input the first sample intermediate vector into a full-connection network to obtain a second sample intermediate vector.
In an alternative embodiment, the calculating module 1602 is further configured to input the second sample intermediate vector into the activation function to obtain a probability that the first sample user has an infection.
In summary, the training apparatus sets n of n sample users1A first sample user has a label, n2And (4) training the graph convolutional neural network when the second sample user does not have the label, so that the sample user without the label still participates in the integrated calculation of the data, but the prediction probability of the sample user without the label does not participate in the training of the graph convolutional neural network.
Fig. 17 is a block diagram illustrating a structure of an apparatus for training a convolutional neural network, which trains the convolutional neural network in a supervised manner, according to an exemplary embodiment of the present application, and includes:
an acquisition module 1701 for acquiring n1Sample physiological information and sample location information for a first sample user; the first sample user carries a tag for identifying the first sampleWhether the user has the infectious disease;
a calculation module 1702 for comparing n with1Inputting the sample graph network corresponding to the sample physiological information and the sample position information of the first sample user into the graph convolution neural network, and obtaining n through the computation of the graph convolution neural network1N of a first sample user1A sample feature vector;
a calculation module 1702 for n1One of the first sample users obtains the predicted probability that the first sample user has the infectious disease through a probability calculation network based on the sample feature vectors of the first sample users;
a training module 1703 for predicting probability and n1And (3) training the convolutional neural network of the graph according to the loss value of the proportion of the diseased sample users in the first sample users.
In an alternative embodiment, the probabilistic computing network includes an activation function.
In an alternative embodiment, the calculating module 1702 is further configured to concatenate the sample feature vector of the first sample user and the initial feature vector of the first sample user to obtain a first sample intermediate vector.
In an alternative embodiment, the calculating module 1702 is further configured to input the first sample intermediate vector into a full-connection network to obtain a second sample intermediate vector.
In an alternative embodiment, the calculating module 1702 is further configured to input the second sample intermediate vector into the activation function to obtain a probability that the first sample user has the infectious disease.
In summary, the training apparatus utilizes n carrying labels1And the first sample user trains the graph convolution neural network to realize the training of the graph convolution neural network.
Fig. 18 shows a block diagram of a computer device 1800, provided in an example embodiment of the present application. The computer device may be a terminal or a server, and in this embodiment, it may be described as simply that the terminal separately trains the graph convolutional neural network and/or the terminal separately uses the graph convolutional neural network, or the server separately trains the graph convolutional neural network and/or the server separately uses the graph convolutional neural network, or the terminal and the server jointly train the image retrieval model and/or the terminal and the server jointly use the graph convolutional neural network.
Generally, computer device 1800 includes: a processor 1801 and a memory 1802.
The processor 1801 may include one or more processing cores, such as a 4-core processor, an 8-core processor, and so on. The processor 1801 may be implemented in at least one hardware form of a DSP (Digital Signal Processing), an FPGA (Field-Programmable Gate Array), and a PLA (Programmable Logic Array). The processor 1801 may also include a main processor and a coprocessor, where the main processor is a processor for Processing data in an awake state, and is also called a Central Processing Unit (CPU); a coprocessor is a low power processor for processing data in a standby state. In some embodiments, the processor 1801 may be integrated with a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content required to be displayed by the display screen. In some embodiments, the processor 1801 may further include an AI processor to process computing operations related to machine learning.
Memory 1802 may include one or more computer-readable storage media, which may be non-transitory. Memory 1802 may also include high speed random access memory, as well as non-volatile memory, such as one or more magnetic disk storage devices, flash memory storage devices. In some embodiments, a non-transitory computer readable storage medium in memory 1802 is used to store at least one instruction for execution by processor 1801 to implement a method of predicting an infectious disease or a method of training a convolutional neural network provided by method embodiments herein.
In some embodiments, computer device 1800 may also optionally include: a peripheral interface 1803 and at least one peripheral. The processor 1801, memory 1802, and peripheral interface 1803 may be connected by a bus or signal line. Each peripheral device may be connected to the peripheral device interface 1803 by a bus, signal line, or circuit board. Specifically, the peripheral device includes: at least one of radio frequency circuitry 1804, display 1805, camera assembly 1806, audio circuitry 1807, positioning assembly 1808, and power supply 1809.
The peripheral interface 1803 may be used to connect at least one peripheral associated with I/O (Input/Output) to the processor 1801 and the memory 1802. In some embodiments, the processor 1801, memory 1802, and peripheral interface 1803 are integrated on the same chip or circuit board; in some other embodiments, any one or two of the processor 1801, the memory 1802, and the peripheral device interface 1803 may be implemented on separate chips or circuit boards, which is not limited in this embodiment.
The Radio Frequency circuit 1804 is used for receiving and transmitting RF (Radio Frequency) signals, also called electromagnetic signals. The radio frequency circuitry 1804 communicates with communication networks and other communication devices via electromagnetic signals. The rf circuit 1804 converts electrical signals into electromagnetic signals for transmission, or converts received electromagnetic signals into electrical signals. Optionally, the radio frequency circuitry 1804 includes: an antenna system, an RF transceiver, one or more amplifiers, a tuner, an oscillator, a digital signal processor, a codec chipset, a subscriber identity module card, and so forth. The radio frequency circuitry 1804 may communicate with other terminals via at least one wireless communication protocol. The wireless communication protocols include, but are not limited to: the world wide web, metropolitan area networks, intranets, generations of mobile communication networks (2G, 3G, 4G, and 5G), Wireless local area networks, and/or WiFi (Wireless Fidelity) networks. In some embodiments, the rf circuit 1804 may also include NFC (Near Field Communication) related circuits, which are not limited in this application.
The display screen 1805 is used to display a UI (User Interface). The UI may include graphics, text, icons, video, and any combination thereof. When the display screen 1805 is a touch display screen, the display screen 1805 also has the ability to capture touch signals on or over the surface of the display screen 1805. The touch signal may be input to the processor 1801 as a control signal for processing. At this point, the display 1805 may also be used to provide virtual buttons and/or a virtual keyboard, also referred to as soft buttons and/or a soft keyboard. In some embodiments, the display 1805 may be one, disposed on a front panel of the computer device 1800; in other embodiments, the number of the display screens 1805 may be at least two, respectively disposed on different surfaces of the computer device 1800 or in a foldable design; in other embodiments, the display 1805 may be a flexible display disposed on a curved surface or on a folded surface of the computer device 1800. Even more, the display 1805 may be arranged in a non-rectangular irregular figure, i.e. a shaped screen. The Display 1805 may be made of LCD (Liquid Crystal Display), OLED (Organic Light-Emitting Diode), or the like.
The camera assembly 1806 is used to capture images or video. Optionally, the camera assembly 1806 includes a front camera and a rear camera. Generally, a front camera is disposed at a front panel of the terminal, and a rear camera is disposed at a rear surface of the terminal. In some embodiments, the number of the rear cameras is at least two, and each rear camera is any one of a main camera, a depth-of-field camera, a wide-angle camera and a telephoto camera, so that the main camera and the depth-of-field camera are fused to realize a background blurring function, and the main camera and the wide-angle camera are fused to realize panoramic shooting and VR (Virtual Reality) shooting functions or other fusion shooting functions. In some embodiments, camera assembly 1806 may also include a flash. The flash lamp can be a monochrome temperature flash lamp or a bicolor temperature flash lamp. The double-color-temperature flash lamp is a combination of a warm-light flash lamp and a cold-light flash lamp, and can be used for light compensation at different color temperatures.
The audio circuitry 1807 may include a microphone and a speaker. The microphone is used for collecting sound waves of a user and the environment, converting the sound waves into electric signals, and inputting the electric signals to the processor 1801 for processing or inputting the electric signals to the radio frequency circuit 1804 to achieve voice communication. The microphones may be multiple and placed at different locations on the computer device 1800 for stereo sound capture or noise reduction purposes. The microphone may also be an array microphone or an omni-directional pick-up microphone. The speaker is used to convert electrical signals from the processor 1801 or the radio frequency circuitry 1804 to sound waves. The loudspeaker can be a traditional film loudspeaker or a piezoelectric ceramic loudspeaker. When the speaker is a piezoelectric ceramic speaker, the speaker can be used for purposes such as converting an electric signal into a sound wave audible to a human being, or converting an electric signal into a sound wave inaudible to a human being to measure a distance. In some embodiments, audio circuitry 1807 may also include a headphone jack.
The Location component 1808 is used to locate a current geographic Location of the computer device 1800 for navigation or LBS (Location Based Service). The Positioning component 1808 may be a Positioning component based on a Global Positioning System (GPS) in the united states, a beidou System in china, or a galileo System in russia.
The power supply 1809 is used to power various components within the computer device 1800. The power supply 1809 may be ac, dc, disposable or rechargeable. When the power supply 1809 includes a rechargeable battery, the rechargeable battery may be a wired rechargeable battery or a wireless rechargeable battery. The wired rechargeable battery is a battery charged through a wired line, and the wireless rechargeable battery is a battery charged through a wireless coil. The rechargeable battery may also be used to support fast charge technology.
In some embodiments, computer device 1800 also includes one or more sensors 1810. The one or more sensors 1810 include, but are not limited to: acceleration sensor 1811, gyro sensor 1812, pressure sensor 1813, fingerprint sensor 1814, optical sensor 1815, and proximity sensor 1816.
The acceleration sensor 1811 may detect the magnitude of acceleration in three coordinate axes of a coordinate system established with the computer apparatus 1800. For example, the acceleration sensor 1811 may be used to detect components of gravitational acceleration in three coordinate axes. The processor 1801 may control the display 1805 to display the user interface in a landscape view or a portrait view according to the gravitational acceleration signal collected by the acceleration sensor 1811. The acceleration sensor 1811 may also be used for acquisition of motion data of a game or a user.
The gyro sensor 1812 may detect a body direction and a rotation angle of the computer device 1800, and the gyro sensor 1812 may cooperate with the acceleration sensor 1811 to collect a 3D motion of the user on the computer device 1800. The processor 1801 may implement the following functions according to the data collected by the gyro sensor 1812: motion sensing (such as changing the UI according to a user's tilting operation), image stabilization at the time of photographing, game control, and inertial navigation.
Pressure sensors 1813 may be located on the side bezel of computer device 1800 and/or underneath display 1805. When the pressure sensor 1813 is disposed on a side frame of the computer device 1800, a user's holding signal to the computer device 1800 can be detected, and the processor 1801 performs left-right hand recognition or shortcut operation according to the holding signal collected by the pressure sensor 1813. When the pressure sensor 1813 is disposed at the lower layer of the display screen 1805, the processor 1801 controls the operability control on the UI interface according to the pressure operation of the user on the display screen 1805. The operability control comprises at least one of a button control, a scroll bar control, an icon control and a menu control.
The fingerprint sensor 1814 is used to collect the fingerprint of the user, and the processor 1801 identifies the user according to the fingerprint collected by the fingerprint sensor 1814, or the fingerprint sensor 1814 identifies the user according to the collected fingerprint. Upon recognizing that the user's identity is a trusted identity, the processor 1801 authorizes the user to perform relevant sensitive operations, including unlocking a screen, viewing encrypted information, downloading software, paying, and changing settings, etc. The fingerprint sensor 1814 may be disposed on the front, back, or side of the computer device 1800. When a physical key or vendor Logo is provided on the computer device 1800, the fingerprint sensor 1814 may be integrated with the physical key or vendor Logo.
The optical sensor 1815 is used to collect the ambient light intensity. In one embodiment, the processor 1801 may control the display brightness of the display screen 1805 based on the ambient light intensity collected by the optical sensor 1815. Specifically, when the ambient light intensity is high, the display brightness of the display screen 1805 is increased; when the ambient light intensity is low, the display brightness of the display 1805 is reduced. In another embodiment, the processor 1801 may also dynamically adjust the shooting parameters of the camera assembly 1806 according to the intensity of the ambient light collected by the optical sensor 1815.
A proximity sensor 1816, also known as a distance sensor, is typically provided on the front panel of the computer device 1800. The proximity sensor 1816 is used to gather the distance between the user and the front of the computer device 1800. In one embodiment, the processor 1801 controls the display 1805 to switch from the bright screen state to the dark screen state when the proximity sensor 1816 detects that the distance between the user and the front of the computer device 1800 is gradually decreased; when the proximity sensor 1816 detects that the distance between the user and the front of the computer device 1800 is gradually increasing, the display 1805 is controlled by the processor 1801 to switch from the breath-screen state to the bright-screen state.
Those skilled in the art will appreciate that the configuration illustrated in FIG. 18 is not intended to be limiting with respect to the computer device 1800 and may include more or fewer components than those illustrated, or some components may be combined, or a different arrangement of components may be employed.
The present application further provides a computer-readable storage medium, having at least one instruction, at least one program, a set of codes, or a set of instructions stored therein, which is loaded and executed by a processor to implement the infectious disease prediction method and the training method of the graph convolution neural network provided by the above method embodiments.
A computer program product or computer program is provided that includes computer instructions stored in a computer readable storage medium. The processor of the computer device reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions to cause the computer device to execute the infectious disease prediction method and the training method of the graph convolution neural network provided by the method embodiment.
The above-mentioned serial numbers of the embodiments of the present application are merely for description and do not represent the merits of the embodiments.
It will be understood by those skilled in the art that all or part of the steps for implementing the above embodiments may be implemented by hardware, or may be implemented by a program instructing relevant hardware, where the program may be stored in a computer-readable storage medium, and the above-mentioned storage medium may be a read-only memory, a magnetic disk or an optical disk, etc.
The above description is only exemplary of the present application and should not be taken as limiting, as any modification, equivalent replacement, or improvement made within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims (15)

1. An infectious disease prediction method, comprising:
acquiring first physiological information and first position information of a first user, wherein the first physiological information is physiological information of the first user related to the infectious disease, and the first position information indicates the position of the first user;
adding a first user node in the basic graph network of the infectious disease based on the first physiological information and the first position information to obtain an updated graph network; the base graph network comprises at least one user node and/or at least one region node, wherein the user node identifies a user suspected to be suffering from the infectious disease, and the region node identifies a region where the user suspected to be suffering from the infectious disease is located;
inputting the updated graph network into a graph convolution neural network, and predicting to obtain a first user feature vector of the first user node; and predicting, by a probability computation network, a probability that the first user has the infectious disease based on the first user feature vector.
2. The method of claim 1, wherein the base graph network comprises a first region node and the updated graph network comprises a first graph network, the first region node identifying a first region in which a user suspected of having the infectious disease is located;
adding a first user node in the basic graph network of the infectious disease based on the first physiological information and the first position information to obtain an updated graph network, wherein the method comprises the following steps:
generating a first user node in the base graph network based on the first physiological information;
and under the condition that the position of the first user indicated by the first position information falls into the first region, performing edge connection operation on the first user node and the first region node in the basic graph network to form a first graph network.
3. The method of claim 2, wherein the graph convolutional neural network comprises k graph convolutional layers;
inputting the updated graph network into a graph convolution neural network, and predicting to obtain a first user feature vector of the first user, wherein the method comprises the following steps:
obtaining the initial region feature vector based on historical diseased information carried by the first region node in the first graph network; obtaining an initial user feature vector based on first physiological information carried by the first user node in the first graph network;
calculating to obtain a layer 1 region feature vector based on the initial region feature vector and the initial user feature vector; calculating to obtain a layer 1 user feature vector based on the initial region feature vector and the initial user feature vector;
calculating to obtain an ith layer region feature vector based on the ith-1 layer region feature vector and the ith-1 layer user feature vector; calculating to obtain an ith layer user feature vector based on the ith-1 layer region feature vector and the ith-1 layer user feature vector;
and outputting the k-th layer user feature vector as the first user feature vector, wherein k is an integer larger than 1, and i is an integer larger than 0 and smaller than k.
4. The method of claim 1, wherein the base graph network comprises a first region node and a second region node connected to the first region node, wherein the updated graph network comprises a second graph network, wherein the first region node identifies a first region in which a user suspected of having the infection is located, and wherein the second region node identifies a second region in which the user suspected of having the infection is located;
adding a first user node in the basic graph network of the infectious disease based on the first physiological information and the first position information to obtain an updated graph network, wherein the method comprises the following steps:
generating a first user node in the base graph network based on the first physiological information;
under the condition that the position of the first user indicated by the first position information falls into the first region, performing edge connection operation on the first user node and the first region node in the basic graph network to form a second graph network;
the population flow number between the second region node and the first region node reaches a number threshold, and/or the distances between the regions respectively indicated by the second region node and the first region node are lower than a first distance threshold.
5. The method of claim 4, wherein the graph convolutional neural network comprises k graph convolutional layers;
inputting the updated graph network into a graph convolution neural network, and predicting to obtain a first user feature vector of the first user, wherein the method comprises the following steps:
obtaining an initial region feature vector of the first region node based on historical diseased information carried by the first region node in the second graph network; obtaining an initial region feature vector of the second region node based on historical diseased information carried by the second region node in the second graph network; obtaining an initial user feature vector based on first physiological information carried by the first user node in the second graph network;
calculating to obtain a layer 1 region feature vector of the first region node based on an initial region feature vector of the first region node, an initial region feature vector of the second region node and the initial user feature vector in combination with a first weight, wherein the first weight is associated with the population flow quantity between the first region node and the second region node; calculating to obtain a layer 1 user feature vector based on the initial region feature vector of the first region node and the initial user feature vector;
calculating to obtain an ith layer region feature vector of the first region node based on the ith-1 layer region feature vector of the first region node, the ith-1 layer region feature vector of the second region node and the ith-1 layer user feature vector in combination with the first weight; calculating to obtain an i-th layer user feature vector based on the i-1-th layer region feature vector of the first region node and the i-1-th layer user feature vector;
and outputting the k-th layer user feature vector as the first user feature vector, wherein k is an integer larger than 1, and i is an integer larger than 0 and smaller than k.
6. The method of claim 1, wherein the base graph network comprises a first region node and a second user node connected to the first region node, wherein the updated graph network comprises a third graph network, wherein the first region node identifies a first region in which a user suspected of having the infectious disease is located, and wherein the second user node identifies a second user suspected of having the infectious disease;
adding a first user node in a first graph network of the infectious disease based on the first physiological information and the first position information to obtain a processed graph network, wherein the processing comprises:
generating a first user node in the base graph network based on the first physiological information;
under the condition that the position of the first user indicated by the first position information falls into the first region, performing edge connection operation on the first user node and the first region node in the basic graph network to form a third graph network;
and the position of the second user indicated by the second user node falls into the first area.
7. The method of claim 6, wherein the graph convolutional neural network comprises k graph convolutional layers;
inputting the updated graph network into a graph convolution neural network, and predicting to obtain a first user feature vector of the first user, wherein the method comprises the following steps:
obtaining an initial region feature vector based on historical diseased information carried by the first region node in the third graph network; obtaining an initial user feature vector of the first user node based on the first physiological information carried by the first user node in the third graph network; obtaining an initial user feature vector of the second user node based on second physiological information carried by the second user node in the third graph network;
calculating to obtain a layer 1 region feature vector based on the initial region feature vector, the initial user feature vector of the first user node and the initial user feature vector of the second user node; calculating to obtain a layer 1 user feature vector of the first user node based on the initial region feature vector and the initial user feature vector of the first user node;
calculating to obtain an ith layer region feature vector based on the ith-1 layer region feature vector, the ith-1 layer user feature vector of the first user node and the ith-1 layer user feature vector of the second user node; calculating to obtain an i-th layer user feature vector of the first user node based on the i-1-th layer region feature vector and the i-1-th layer user feature vector of the first user node;
and outputting the k-th layer user feature vector of the first user node as the first user feature vector, wherein k is an integer larger than 1, and i is an integer larger than 0 and smaller than k.
8. The method of claim 1, wherein the base graph network comprises a third user node, wherein the updated graph network comprises a fourth graph network, wherein the third user node identifies a third user suspected of having the infectious disease, and wherein the third user node carries third physiological information and third location information of the third user;
adding a first user node in a first graph network of the infectious disease based on the first physiological information and the first position information to obtain a processed graph network, wherein the processing comprises:
generating a first user node in the base graph network based on the first physiological information;
and under the condition that the similarity between the first physiological information and the third physiological information is greater than a similarity threshold value and/or the distance between the positions respectively indicated by the first position information and the third position information is smaller than a second distance threshold value, performing edge connection operation on the first user node and the third user node in the basic graph network to form a fourth graph network.
9. The method of claim 8, wherein the graph convolutional neural network comprises k graph convolutional layers;
the inputting the processed graph network into a graph convolution neural network to obtain a first user feature vector of the first user includes:
obtaining an initial user feature vector of the first user node based on the first physiological information carried by the first user node in the fourth graph network; obtaining an initial user feature vector of the third user node based on the third physiological information carried by the third user node in the fourth graph network;
calculating to obtain a layer 1 user feature vector of the first user node and a layer 1 user feature vector of the third user node based on the initial user feature vector of the first user node and the initial user feature vector of the third user node in combination with a second weight, wherein the second weight is associated with at least one of a similarity between the first physiological information and the third physiological information and a distance between positions respectively indicated by the first position information and the third position information;
calculating to obtain an ith layer user characteristic vector of the first user node and an ith layer user characteristic vector of the third user node based on the ith-1 layer user characteristic vector of the first user node and the ith-1 layer user characteristic vector of the third user node by combining the second weight;
and outputting the k-th layer user feature vector of the first user node as the first user feature vector, wherein k is an integer larger than 1, and i is an integer larger than 0 and smaller than k.
10. The method of claim 3 or 5 or 7 or 9, wherein the probability computation network comprises an activation function;
the predicting, by a probability computation network, a probability that the first user has the infectious disease based on the first user feature vector, comprising:
splicing the first user characteristic vector and the initial user characteristic vector of the first user node to obtain a first intermediate vector;
inputting the first intermediate vector into a full-connection network to obtain a second intermediate vector;
inputting the second intermediate vector into an activation function that predicts a probability that the first user has the infectious disease.
11. A method for training a convolutional neural network as claimed in claim 1, wherein the method is used for training the convolutional neural network, and the training method adopts a semi-supervised mode, and the method comprises:
acquiring sample physiological information and sample position information of n sample users; the n sample users comprise n1A first sample user and n2A second sample user, the first sample user carrying a label for determining the first sample userWhether or not said infectious disease is present, said second sample user not carrying said tag;
inputting a sample graph network corresponding to the sample physiological information and the sample position information of the n sample users into the graph convolution neural network, and calculating n sample characteristic vectors of the n sample users through the graph convolution neural network;
for said n1One of the first sample users, obtaining the predicted probability that the first sample user has the infectious disease through the probability calculation network based on the sample feature vector of the first sample user;
based on the prediction probability and the n1And training the graph convolutional neural network according to the loss value of the proportion of the diseased sample users in the first sample users.
12. An infectious disease prediction apparatus, characterized in that the apparatus comprises:
the system comprises an acquisition module, a processing module and a display module, wherein the acquisition module is used for acquiring first physiological information and first position information of a first user, the first physiological information is physiological information of the first user related to the infectious disease, and the first position information indicates the position of the first user;
the updating module is used for adding a first user node in the basic graph network of the infectious disease based on the first physiological information and the first position information to obtain an updated graph network; the base graph network comprises at least one user node and/or at least one region node, wherein the user node identifies a user suspected to be suffering from the infectious disease, and the region node identifies a region where the user suspected to be suffering from the infectious disease is located;
the prediction module is used for inputting the updated graph network into a graph convolution neural network to predict and obtain a first user feature vector of a first user; and predicting, by a probability computation network, a probability that the first user has the infectious disease based on the first user feature vector.
13. An apparatus for training a convolutional neural network, the apparatus comprising:
the acquisition module is used for acquiring sample physiological information and sample position information of n sample users; the n sample users comprise n1A first sample user and n2A second sample user, the first sample user carrying a label for determining whether the first sample user has the infectious disease, the second sample user not carrying the label;
the calculation module is used for inputting the sample physiological information and the sample position information of the n sample users into the graph convolution neural network, and obtaining n sample characteristic vectors of the n sample users through the graph convolution neural network;
a calculation module for the n1One of the first sample users, obtaining the predicted probability that the first sample user has the infectious disease through the probability calculation network based on the sample feature vector of the first sample user;
a training module to train the prediction probability based on the n1And training the graph convolutional neural network according to the loss value of the proportion of the diseased sample users in the first sample users.
14. A computer device, characterized in that the computer device comprises: a processor and a memory, the memory storing a computer program that is loaded and executed by the processor to implement the infectious disease prediction method of any one of claims 1 to 10, and/or the training method of the graph-convolution neural network of claim 11.
15. A computer-readable storage medium storing a computer program which is loaded and executed by a processor to implement the infectious disease prediction method according to any one of claims 1 to 10 and/or the training method of the convolutional neural network according to claim 11.
CN202111201853.8A 2021-10-15 2021-10-15 Infectious disease prediction and training method, device, equipment and medium Pending CN113936809A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111201853.8A CN113936809A (en) 2021-10-15 2021-10-15 Infectious disease prediction and training method, device, equipment and medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111201853.8A CN113936809A (en) 2021-10-15 2021-10-15 Infectious disease prediction and training method, device, equipment and medium

Publications (1)

Publication Number Publication Date
CN113936809A true CN113936809A (en) 2022-01-14

Family

ID=79279870

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111201853.8A Pending CN113936809A (en) 2021-10-15 2021-10-15 Infectious disease prediction and training method, device, equipment and medium

Country Status (1)

Country Link
CN (1) CN113936809A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115862888A (en) * 2023-02-17 2023-03-28 之江实验室 Infectious disease infection prediction method, system, device and storage medium

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115862888A (en) * 2023-02-17 2023-03-28 之江实验室 Infectious disease infection prediction method, system, device and storage medium

Similar Documents

Publication Publication Date Title
CN112884770B (en) Image segmentation processing method and device and computer equipment
CN111104980B (en) Method, device, equipment and storage medium for determining classification result
CN108288032B (en) Action characteristic acquisition method, device and storage medium
CN111897996A (en) Topic label recommendation method, device, equipment and storage medium
CN111680697B (en) Method, device, electronic equipment and medium for realizing field adaptation
CN108363982B (en) Method and device for determining number of objects
CN111027490B (en) Face attribute identification method and device and storage medium
CN110933468A (en) Playing method, playing device, electronic equipment and medium
CN111127509A (en) Target tracking method, device and computer readable storage medium
WO2022193973A1 (en) Image processing method and apparatus, electronic device, computer readable storage medium, and computer program product
CN110570460A (en) Target tracking method and device, computer equipment and computer readable storage medium
CN111738365B (en) Image classification model training method and device, computer equipment and storage medium
CN112002391A (en) Method and device for determining checking sequence, electronic equipment and storage medium
CN113918767A (en) Video clip positioning method, device, equipment and storage medium
CN113257412B (en) Information processing method, information processing device, computer equipment and storage medium
CN110166275B (en) Information processing method, device and storage medium
CN113936809A (en) Infectious disease prediction and training method, device, equipment and medium
CN111353513B (en) Target crowd screening method, device, terminal and storage medium
CN113516723A (en) Face picture encryption method and device, computer equipment and storage medium
CN111898535A (en) Target identification method, device and storage medium
CN111563201A (en) Content pushing method, device, server and storage medium
CN112925922A (en) Method, device, electronic equipment and medium for obtaining address
CN110990549A (en) Method and device for obtaining answers, electronic equipment and storage medium
CN110414673B (en) Multimedia recognition method, device, equipment and storage medium
CN114328948A (en) Training method of text standardization model, text standardization method and device

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