CN113724892A - Method and device for analyzing population mobility, electronic equipment and storage medium - Google Patents

Abstract

The application is applicable to the technical field of big data, and provides a population mobility analysis method, a population mobility analysis device, electronic equipment and a storage medium, wherein the method comprises the following steps: receiving population mobility information uploaded by each medical cloud node; each medical cloud node corresponds to one monitoring area; constructing node topological graphs corresponding to all medical cloud nodes according to all the population flow information, and determining flow characteristic vectors of the medical cloud nodes based on the node topological graphs; importing all the flow characteristic vectors into a preset cluster analysis algorithm, and performing cluster analysis on all the medical cloud nodes to obtain at least one cluster center; and identifying abnormal nodes from all the medical cloud nodes according to the clustering center, and generating early warning information about the abnormal nodes. By adopting the method, the accuracy and the reliability of the early warning information are improved.

Description

Method and device for analyzing population mobility, electronic equipment and storage medium

Technical Field

The application belongs to the technical field of big data, and particularly relates to a population mobility analysis method and device, electronic equipment and a storage medium.

Background

Along with the continuous development of society, the population also increases, the population flow among all the areas is more and more frequent, how to accurately and effectively analyze the population flow condition of each area and send out early warning information when an abnormal condition is detected has important significance for social management, such as application to backtracking, analysis and prevention of virus propagation tracks. Especially for the management and control and the propagation prediction of infectious diseases, the system can accurately analyze the population flow and perform corresponding early warning, thereby effectively controlling the propagation of the infectious diseases and isolating and treating patients.

In the existing technology for generating early warning information, when the early warning information of a certain area needs to be determined, population mobility characteristic information of the area is often acquired through a monitoring device corresponding to the area, and is imported into an anomaly identification algorithm configured based on artificial experience based on the population mobility characteristic information to generate corresponding early warning information. However, the population mobility in a certain area is closely related to the population mobility in other areas, and when the early warning information is generated by the method, the population mobility in one area is only analyzed independently, so that the accuracy and the reliability of the early warning information are greatly reduced.

Disclosure of Invention

The embodiment of the application provides a population mobility analysis method and device, electronic equipment and a storage medium, which can solve the problems that the accuracy and reliability of early warning information are greatly reduced because the generation technology of the early warning information only analyzes the population mobility of one region when the early warning information is generated.

In a first aspect, an embodiment of the present application provides a method for analyzing population mobility, including:

receiving population mobility information uploaded by each medical cloud node; each medical cloud node corresponds to one monitoring area; the population mobility information is used for determining population mobility of a monitoring area corresponding to the medical cloud node;

constructing node topological graphs corresponding to all medical cloud nodes according to all the population flow information, and determining flow characteristic vectors of the medical cloud nodes based on the node topological graphs;

importing all the flow characteristic vectors into a preset cluster analysis algorithm, and performing cluster analysis on all the medical cloud nodes to obtain at least one cluster center;

and identifying abnormal nodes from all the medical cloud nodes according to the clustering center, and generating early warning information about the abnormal nodes.

In a possible implementation manner of the first aspect, the constructing a node topology map corresponding to all medical cloud nodes according to all the population mobility information, and determining a mobility feature vector of the medical cloud node based on the node topology map includes:

according to the position relation between the monitoring areas corresponding to the medical cloud nodes, data nodes are established for the medical cloud nodes in a preset topology template; the initial state of the data node is determined based on the population state of the monitoring area corresponding to the medical cloud node;

generating a movement track of a user according to the population mobility information of each medical cloud node;

determining an incidence relation between the data nodes according to the monitoring area passed by the moving track;

and connecting each data node based on the incidence relation to generate the node topological graph.

In a possible implementation manner of the first aspect, the constructing a node topology map corresponding to all medical cloud nodes according to all the population mobility information, and determining a mobility feature vector of the medical cloud node based on the node topology map includes:

generating an association feature vector of any medical cloud node according to population mobility information of the association cloud node having the association relation with the medical cloud node; the associated feature vector specifically includes:

Connect[v]＝[Side(v,w₁),Side(v,w₂),...,Side(v,w_n)]

wherein, Connect [ v ]]Is the associated feature vector; v is the any medical cloud node; w is a_nThe nth associated cloud node; side (v, w)_n) Population flow information between the any medical cloud node and the nth associated cloud node; n isA total number of the associated cloud nodes;

constructing an associated node state vector of any medical cloud node according to the initial states of all the associated cloud nodes corresponding to the medical cloud node; the associated node state vector is specifically:

LinkPoint[v]＝[State(w₁),...,State(w_n)]

wherein LinkPoint [ v ]]Is the correlation node state vector; state (w)_n) Is the initial state of the nth association node;

determining a feature transformation vector of any medical cloud node in a preset multidimensional space according to the associated feature vector, the associated node state vector and the initial state of any medical cloud node; the feature transformation vector specifically comprises:

TurnVector[v]＝Switch_d(State[v],Connect[v],Emd[v],LinkPoint[v])

wherein TurnVector [ v ] is the feature transformation vector; emd [ v ] is an embedding vector determined based on the associated node;

calculating to obtain the feature transformation vector through a preset updating iterative transformation algorithm;

and generating the flow feature vector according to the feature conversion vector and the initial state of any medical cloud node.

In a possible implementation manner of the first aspect, the receiving population flow information uploaded by each medical cloud node includes:

receiving position information fed back by a user terminal to generate a user flow track of each user terminal;

determining the stay time of the user flow track at the node corresponding to each medical cloud node according to the stay time of the position corresponding to each piece of position information;

selecting a flow track section with the node stay time being larger than a preset stay time threshold value from all the user flow tracks as an effective flow track section;

and determining population flow information corresponding to each medical cloud node based on all the effective flow track sections.

In a possible implementation manner of the first aspect, the identifying, according to the clustering center, an abnormal node from all the medical cloud nodes, and generating early warning information about the abnormal node includes:

acquiring an initial identification state of each medical cloud node;

if the initial identification state of any medical cloud node is an abnormal state, calculating the center offset between any medical cloud node and the clustering center;

if the central offset is smaller than a preset offset threshold, identifying any medical cloud node as a normal node;

if the central offset is greater than or equal to the offset threshold, identifying any medical cloud node as an abnormal node;

determining at least one abnormal center according to all the abnormal nodes;

respectively determining abnormal characteristic information of each abnormal node based on the abnormal offset between each abnormal node and the abnormal center; the abnormal feature information includes: exception type and exception level.

In a possible implementation manner of the first aspect, after the obtaining the initial identification state of each medical cloud node, the method further includes:

if the initial identification state of any medical cloud node is a normal state, identifying an abnormal center according to the medical cloud nodes contained in each clustering center;

and identifying the abnormal nodes from all the medical cloud nodes according to the abnormal offset between each medical cloud node and the abnormal center.

In a possible implementation manner of the first aspect, after the identifying, according to the clustering center, an abnormal node from all the medical cloud nodes and generating early warning information about the abnormal node, the method further includes:

receiving a verification result fed back based on the early warning information, and determining a parameter adjustment range based on a verification deviation amount between the early warning information and the verification result;

determining a plurality of candidate adjusting parameters from the parameter adjusting range based on a preset adjusting step length, adjusting the clustering analysis algorithm based on the candidate adjusting parameters, and obtaining a verification accuracy rate corresponding to each candidate adjusting parameter;

and determining target adjustment parameters from a plurality of candidate adjustment parameters based on the verification accuracy, and adjusting the cluster analysis algorithm based on the target adjustment parameters to obtain an adjusted cluster analysis algorithm.

In a second aspect, an embodiment of the present application provides an apparatus for analyzing population mobility, including:

the population mobility information acquisition unit is used for receiving population mobility information uploaded by each medical cloud node; each medical cloud node corresponds to one monitoring area; the population mobility information is used for determining population mobility of a monitoring area corresponding to the medical cloud node;

the flow characteristic vector generating unit is used for constructing node topological graphs corresponding to all medical cloud nodes according to all the population flow information and determining flow characteristic vectors of the medical cloud nodes based on the node topological graphs;

the cluster center determining unit is used for performing cluster analysis on all the medical cloud nodes based on all the flow characteristic vectors to obtain at least one cluster center;

and the early warning information generating unit is used for identifying abnormal nodes from all the medical cloud nodes according to the clustering center and generating early warning information about the abnormal nodes.

In a third aspect, an embodiment of the present application provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor, when executing the computer program, implements the method according to any one of the above first aspects.

In a fourth aspect, the present application provides a computer-readable storage medium, which stores a computer program, and when the computer program is executed by a processor, the computer program implements the method according to any one of the above first aspects.

In a fifth aspect, embodiments of the present application provide a computer program product, which, when run on an electronic device, causes the electronic device to perform the method of any one of the above first aspects.

Compared with the prior art, the embodiment of the application has the advantages that: the population flow information of each monitoring area is acquired through medical cloud nodes distributed in each monitoring area, then the electronic equipment receives the population flow information uploaded by each medical cloud node, a corresponding node topological graph is constructed based on the population flow information, flow characteristic vectors corresponding to each medical cloud node are determined according to the constructed node topological graph, all the medical cloud nodes are subjected to cluster analysis according to the flow characteristic vectors to obtain corresponding cluster centers, corresponding abnormal nodes are identified through the cluster centers to generate early warning information about the abnormal nodes, and automatic early warning of abnormal conditions is achieved. Compared with the existing generation technology of the early warning information, when the early warning information is generated, the corresponding medical cloud systems are not independently analyzed, the corresponding medical cloud systems are built through the plurality of medical cloud nodes to receive population flow information of different monitoring areas and generate corresponding node topological graphs, the incidence relation among the medical cloud nodes needs to be considered when the flow characteristic vectors are generated, so that the accuracy of the flow characteristic vectors can be improved, the clustering analysis is carried out, abnormal nodes deviating from normal flow behaviors are identified and obtained, the corresponding early warning information is output, and the accuracy and the reliability of the early warning information are improved.

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 embodiments or the prior art descriptions will be briefly described 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 without creative efforts.

Fig. 1 is a flowchart of an implementation of a method for analyzing population mobility according to an embodiment of the present disclosure;

fig. 2 is a schematic structural diagram of a medical cloud system provided in an embodiment of the present application;

fig. 3 is a schematic diagram illustrating generation of warning information according to an embodiment of the present application;

fig. 4 is a schematic diagram of an implementation manner of S102 of a method for analyzing population mobility according to an embodiment of the present application;

fig. 5 is a schematic diagram of an implementation manner of S101 of a method for analyzing population mobility according to an embodiment of the present application;

fig. 6 is a schematic diagram of an implementation manner of S104 of a method for analyzing population mobility according to an embodiment of the present application;

FIG. 7 is a schematic diagram illustrating an implementation of a method for analyzing population mobility according to an embodiment of the present disclosure;

fig. 8 is a schematic structural diagram of an apparatus for analyzing population mobility according to an embodiment of the present disclosure;

fig. 9 is a schematic structural diagram of an electronic device provided in an embodiment of the present application.

Detailed Description

In the following description, for purposes of explanation and not limitation, specific details are set forth, such as particular system structures, techniques, etc. in order to provide a thorough understanding of the embodiments of the present application. It will be apparent, however, to one skilled in the art that the present application may be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present application with unnecessary detail.

It will be understood that the terms "comprises" and/or "comprising," when used in this specification and the appended claims, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.

Furthermore, in the description of the present application and the appended claims, the terms "first," "second," "third," and the like are used for distinguishing between descriptions and not necessarily for describing or implying relative importance.

The method for analyzing the population mobility provided by the embodiment of the application can be applied to electronic equipment such as a smart phone, a server, a tablet computer, a notebook computer, an ultra-mobile personal computer (UMPC), a netbook and the like. The embodiment of the present application does not set any limit to the specific type of the electronic device. Particularly, the electronic device can also be a central node of a medical cloud system, and generates early warning information for abnormal nodes by receiving population flow information uploaded by each medical cloud node so as to perform early warning and management and control on abnormal conditions. For example, the above analysis method for population mobility can be applied to epidemic situation monitoring scenes about infectious diseases, by acquiring population mobility conditions of each monitored area and identified users with infectious diseases, abnormal areas with propagation risks are identified, and early warning information about the abnormal areas is generated.

Referring to fig. 1, fig. 1 is a flowchart illustrating an implementation of a method for analyzing population mobility according to an embodiment of the present application, where the method includes the following steps:

in S101, receiving population flow information uploaded by each medical cloud node; each medical cloud node corresponds to one monitoring area; and the population mobility information is used for determining the population mobility condition of the monitoring area corresponding to the medical cloud node.

In this embodiment, the electronic device may establish a communication connection with each medical cloud node and receive population mobility information acquired by each medical cloud node. Wherein, the demographic flow information includes but is not limited to: the number of the population flowing, the flowing direction of the population, the flowing ratio of the population, the flowing frequency of the population and other parameters are related to the flowing condition of the population.

Illustratively, fig. 2 shows a schematic structural diagram of a medical cloud system provided by an embodiment of the present application. As shown in fig. 1, the medical operation system includes medical cloud nodes 01 deployed in a plurality of different monitoring areas and acentral node 02, and each medical cloud node is configured to acquire population mobility conditions of a corresponding monitoring area, where each medical cloud node 01 may establish communication connection with thecentral node 02, and feed population mobility information acquired by the medical cloud node back to thecentral node 02 through the communication connection. In a possible implementation manner, the medical cloud system may further use a medical cloud node in any monitoring area as a central node.

The electronic equipment can be configured with corresponding uploading trigger conditions, the uploading trigger conditions can be time trigger conditions or event trigger conditions, when the preset uploading trigger conditions are met, the electronic equipment can send an information uploading instruction to each medical cloud node, and then the medical cloud nodes can send population mobility information to the electronic equipment; in a possible implementation manner, the electronic device may send each upload trigger condition to each medical cloud node, and the medical cloud node may determine whether the upload trigger condition is satisfied according to a population mobility condition of a monitoring area to which the medical cloud node belongs, and if so, send the population mobility information to the electronic device; otherwise, the flow condition of the human mouth is continuously monitored.

For example, if the medical cloud system is applied to an epidemic situation monitoring scene of an infectious disease, the upload trigger condition may be to identify a patient with the infectious disease, and when a patient with a specific infectious disease is confirmed, it may be determined whether the patient passes through any monitoring area in the medical cloud system, and if so, it is determined that the upload trigger condition is satisfied, and each medical cloud node uploads population flow information of the monitoring area.

In a possible implementation manner, if the method for analyzing the population mobility is applied to an epidemic situation monitoring scene of an infectious disease, the population mobility information may further include: the number of fever, absenteeism, abnormal body temperature, etc.

In a possible implementation manner, the content of the demographic information is related to the area type of the monitoring area, and the demographic information collected by different area types may include different parameters. For example, if the monitoring area is a hospital type, the demographic flow information may include: specifying the number of purchased drugs, the number of confirmed users, the number of inpatient users, etc.; if the monitoring area is of school type, the population mobility information may include: the number of people arriving at school, the number of people leaving a lesson, etc.; if the monitoring type is a mall type, the demographic flow information may include: the number of purchases, the number of visits to a store, etc. Certainly, the population movement information may be related to an early warning purpose besides the monitoring type, the electronic device may broadcast the early warning purpose to each medical cloud node in the system, and the medical cloud node may determine population movement parameters to be collected according to the early warning purpose and the area type of the monitoring area, generate the population movement information based on the collected population movement parameters, and send the population movement information to the electronic device.

In S102, a node topological graph corresponding to all medical cloud nodes is constructed according to all the population mobility information, and the mobility characteristic vector of the medical cloud node is determined based on the node topological graph.

In this embodiment, the electronic device may determine an association relationship between the medical cloud nodes according to the population flow information uploaded by the medical cloud nodes, and generate the node topology map based on the association relationship between the medical cloud nodes. Each medical cloud node in the node topological graph can serve as a topological node, and the topological nodes are connected according to the incidence relation among the medical cloud nodes, so that the node topological graph is generated.

In one possible implementation, since the demographic flow information may change over time, the medical cloud nodes are typically more fixed. In this case, the electronic device may update the historical topology map generated in the history according to the acquired population mobility information to generate the node topology map.

In this embodiment, the node topological graph generated in the electronic device includes population flow information between the medical cloud nodes, so that when the flow feature vector of the medical cloud node is determined through the node topological graph, not only the population flow condition of the monitoring area to which the medical cloud node belongs but also the population flow condition of the medical cloud node having an association relationship can be considered, and thus the flow feature vector is a vector capable of representing the local population flow condition and the population flow condition of the adjacent node, so that the accuracy of subsequent abnormal node identification is improved.

In this embodiment, the electronic device may store a flow feature extraction algorithm. The flow feature extraction algorithm may specifically be a pair of image analysis algorithms. The electronic device may import the generated node topological graph into the flow feature extraction algorithm, and may identify each node included in the node topological graph and associated feature information between each node through the flow feature extraction algorithm, so as to obtain a flow feature vector corresponding to each node according to an initial state of each node and the associated feature information of each associated node.

In S103, all the flow feature vectors are led into a preset cluster analysis algorithm, and all the medical cloud nodes are subjected to cluster analysis to obtain at least one cluster center.

In this embodiment, the flow feature vectors may be used to determine population flow conditions of corresponding medical cloud nodes, and all the flow feature vectors are introduced into a cluster analysis algorithm to perform cluster analysis on all the medical cloud nodes, so that the medical cloud nodes having the same population flow behavior can be identified, and at least one cluster center is obtained. Each cluster center can correspond to one type of population flow characteristic, and if the deviation amount between the medical cloud node and the cluster center is smaller, the population flow characteristic of the medical cloud node is similar to the population flow characteristic of the cluster center; conversely, if the deviation amount of the medical cloud node and the cluster center is larger, the population flow characteristic of the medical cloud node is not similar to the population flow characteristic of the cluster center. Wherein, the cluster analysis algorithm includes but is not limited to: various algorithms such as kmeans, dbscan density clustering and hierarchical clustering.

In a possible implementation manner, the number of the cluster centers may be multiple. If the electronic device identifies a medical cloud node with an abnormal condition, namely an abnormal node, all the medical cloud nodes can be divided into two groups, namely an abnormal node group and a node group to be classified, and the abnormal cluster center is identified by performing cluster analysis on the abnormal node group and performing cluster analysis on the node group to be classified, so that the cluster center is identified.

In S104, abnormal nodes are identified from all the medical cloud nodes according to the clustering center, and early warning information about the abnormal nodes is generated.

In this embodiment, the electronic device may identify, according to each cluster center, a medical cloud node having an abnormal population flow characteristic, identify the medical cloud node having the abnormal population flow characteristic as an abnormal node, and generate warning information about the abnormal node to notify the user of the abnormal condition obtained through the identification.

Taking an epidemic situation monitoring scene of a sensing disease as an example for explanation, when the node topological graph is generated, the electronic device can identify an area where a sensing disease patient exists as an abnormal node, and identify a monitoring area corresponding to the abnormal node as an area where an epidemic situation exists. Exemplarily, fig. 3 shows a schematic diagram of generation of the warning information provided in an embodiment of the present application. Referring to fig. 3, the node topological graph records initial states corresponding to the medical cloud nodes, and the initial states correspond to monitoring areas where sensing patients appear, and are identified as areas where epidemic situations exist, and the medical cloud nodes corresponding to the areas are abnormal nodes. For the area with the sensing patient, the area is identified as a normal area, the medical cloud node corresponding to the area is a normal node, the above operations belong to identification of the initial state of each medical cloud node, the electronic device may set a plurality of initial abnormal thresholds, so as to divide each medical cloud node into a normal node and an abnormal node, where the initial abnormal threshold may be: whether the number of normal people with fever exceeds 5, whether the number of people in class is more than 10, and whether the single-day sales increase of the fever medicines exceeds 20% can be determined by comparing the number with an abnormal threshold value, and the initial state of the corresponding medical cloud node can be determined.

Then, the electronic device can lead the generated node topological graph into a preset flow characteristic conversion algorithm, determining flow characteristic vectors corresponding to the medical cloud nodes, then leading the flow characteristic vectors into a preset cluster analysis algorithm, all medical cloud nodes are subjected to cluster analysis, so that a plurality of cluster centers can be obtained, and as the flow characteristic vector can represent the flow condition of a patient with a sensing disease, thereby identifying and obtaining the nodes with the abnormal population flow characteristics, dividing the nodes with the abnormal population flow characteristics into a cluster center after cluster analysis, and the nodes without abnormal population flow characteristics are divided into another clustering center, so that all medical cloud nodes can be classified by identifying to obtain the clustering center, and thus, the abnormal nodes are identified. After the abnormal nodes are identified based on the cluster center, the electronic equipment can determine the threshold adjustment parameter according to the comparison between the final identification state for identifying each medical cloud node and the initial state, and adjust the initial abnormal threshold based on the threshold adjustment parameter, so that the accuracy for adjusting the identification of the subsequent initial state can be improved, for example, whether the number of the original heating normal people exceeds 5 people or not is adjusted to whether the number of the heating normal people exceeds 6 people or not.

As can be seen from the above, according to the analysis method for population mobility provided in the embodiment of the present application, population mobility information of each monitoring area is acquired through medical cloud nodes distributed in each monitoring area, then, the electronic device receives the population mobility information uploaded by each medical cloud node, constructs a corresponding node topological graph based on the population mobility information, determines a mobility feature vector corresponding to each medical cloud node according to the constructed node topological graph, performs cluster analysis on all medical cloud nodes according to the mobility feature vector to obtain a corresponding cluster center, obtains a corresponding abnormal node through the cluster center identification, generates early warning information about the abnormal node, and achieves automatic early warning for abnormal situations. Compared with the existing generation technology of the early warning information, when the early warning information is generated, the corresponding medical cloud systems are not independently analyzed, the corresponding medical cloud systems are built through the plurality of medical cloud nodes to receive population flow information of different monitoring areas and generate corresponding node topological graphs, the incidence relation among the medical cloud nodes needs to be considered when the flow characteristic vectors are generated, so that the accuracy of the flow characteristic vectors can be improved, the clustering analysis is carried out, abnormal nodes deviating from normal flow behaviors are identified and obtained, the corresponding early warning information is output, and the accuracy and the reliability of the early warning information are improved.

Fig. 4 is a flowchart illustrating a specific implementation of the method S102 for analyzing population mobility according to the second embodiment of the present invention. Referring to fig. 4, with respect to the embodiment described in fig. 1, in the analysis method for population mobility provided in this embodiment, S102 includes: s1021 to S1029 are specifically described as follows:

further, the constructing a node topological graph corresponding to all medical cloud nodes according to all the population mobility information includes:

in S1021, according to the position relationship between the monitoring areas corresponding to the medical cloud nodes, creating a data node for each medical cloud node in a preset topology template; the initial state of the data node is determined based on a population state of the monitoring area corresponding to the medical cloud node.

In this embodiment, the electronic device may obtain position information between monitoring areas corresponding to the medical cloud nodes, so as to determine a position relationship between the monitoring areas, thereby determining a spatial position relationship of the medical cloud nodes, that is, determining the medical cloud nodes having an adjacent relationship, and marking the medical cloud nodes in the topology template based on the adjacent relationship between the medical cloud nodes, where each medical cloud node corresponds to one data node in the topology template.

The electronic device can acquire the population state of each medical cloud node, and determine the corresponding initial state of the monitoring area based on the population state of the monitoring area corresponding to each medical cloud node. For example, if an abnormal user exists in a monitoring area corresponding to the medical cloud node, the initial state of the medical cloud node may be an abnormal node; of course, population characteristic data of the monitoring area corresponding to the medical cloud node, such as the number of people who generate heat, the number of people who have no lessons, the number of people who leave the office, and the like, may also be obtained, and the initial state corresponding to the medical cloud node is determined based on the comparison of the population characteristic data and the corresponding abnormal threshold.

In S1022, a movement trajectory of the user is generated according to the population flow information of each of the medical cloud nodes.

In this embodiment, the electronic device may analyze population mobility information of the medical cloud node, acquire and determine cross-region behaviors of the mobile users in each monitoring region, that is, determine a leaving monitoring region and a entering monitoring region, and integrate all cross-region behaviors belonging to the same mobile user, so that a movement track of the mobile users can be obtained. The medical cloud node can be configured with a data acquisition device, and movement tracks of mobile users can be tracked through the data acquisition device, so that the population mobility information is generated.

In a possible implementation manner, the electronic device may receive location information fed back by the user terminal of each user, for example, an operator server of a mobile network may obtain an accessed mobile base station of a mobile phone of each user, determine a movement track of each user based on the accessed mobile base station, generate population movement information based on the movement track, and send the population movement information to the electronic device.

In S1023, the association relationship between the data nodes is determined according to the monitoring area passed by the moving track.

In this embodiment, if a moving track of a certain user passes through a plurality of monitoring areas, the medical cloud nodes in the plurality of monitoring areas through which the moving track passes may be identified as medical cloud nodes having an association relationship, and mapped into the topology template, so as to identify that an association relationship exists between the data nodes.

In a possible implementation manner, the electronic device may set a preset maximum number of associated nodes, and if the number of monitoring areas through which a certain movement track passes is greater than the maximum number of associated nodes, a sliding frame may be performed on the movement track based on the maximum number of associated nodes, and the medical cloud nodes in the multiple monitoring areas in the same sliding frame are identified as medical cloud nodes having an associated relationship.

In S1024, each data node is connected based on the association relationship, and the node topology map is generated.

In this embodiment, the electronic device may connect data nodes having all association relationships in the topology template, so as to generate the node topology map

In the embodiment of the application, the association relationship between different data nodes is determined by obtaining the movement track of the user in the monitoring area, the spatial position relationship between the monitoring areas is considered, the placement position of each data node is determined in the topology template, then the data nodes with the association relationship are connected, and the node topology graph is generated, so that the accuracy of the node topology graph can be improved.

Further, the determining flow feature vectors of the medical cloud nodes based on the node topology map comprises:

in S1025, generating an associated feature vector of any medical cloud node according to population mobility information of the associated cloud node having the association relation with the medical cloud node; the associated feature vector specifically includes:

Connect[v]＝[Side(v,w₁),Side(v,w₂),...,Side(v,w_n)]

wherein, Connect [ v ]]Is the associated feature vector; v is the any medical cloud node; w is a_nFor the nth associated cloudA node; side (v, w)_n) Population flow information between the any medical cloud node and the nth associated cloud node; n is the total number of the associated cloud nodes.

In this embodiment, the electronic device may determine, through the node topology map, a medical cloud node that has an association relationship with a certain medical cloud node, that is, an association cloud node, and may determine, according to population mobility information between the medical cloud node and the association cloud node, an association feature vector between the medical cloud node and the association cloud node, so as to determine a population mobility condition between two monitoring areas.

For example, if the analysis method of population mobility is applied to an epidemic situation control scenario, the associated feature vector may be represented as: [ the proportion of the outflow personnel of the node, the proportion of the outflow personnel of the node in the adjacent nodes, and the distance between the nodes ]. Wherein: the ratio of the person flowing out of the node specifies the number of persons flowing out of the node/the total number of persons staying in the node within a preset time period (for example, 1 day). Determining whether to egress may be estimated using the dwell time at the node and the path trajectory. The flow trajectory and information of the person can be estimated using the base station location data of the communication carrier and the registered mobile phone number data. Because the proportion data is calculated, the estimation can be carried out by adopting a sampling mode, and the flow condition of all personnel monitored by the computing node is not required. The proportion of the outflow personnel of the node to the adjacent nodes is as follows: scanning all the personnel information appearing near the adjacent node and recording the personnel information as a set (A), then scanning which personnel information in the set (A) comes from the node, and calculating to obtain a set (B) from the node. The proportion of the outgoing staff of the node to the adjacent nodes is set (B)/set (A).

In S1026, constructing an associated node state vector of any medical cloud node according to the initial states of all the associated cloud nodes corresponding to the any medical cloud node; the associated node state vector is specifically:

LinkPoint[v]＝[State(w₁),...,State(w_n)]

wherein LinkPoint [ v ]]Is the correlation node state vector; state (w)_n) As the initial state of the nth associated node。

In this embodiment, each associated cloud node may also correspond to its own initial state, and the initial state may be determined according to the population state of the monitoring area, and based on the initial states of all associated cloud nodes, an associated node state vector corresponding to the medical cloud node may be obtained.

In S1027, determining a feature transformation vector of the medical cloud node in a preset multidimensional space according to the associated feature vector, the associated node state vector, and an initial state of the medical cloud node; the feature transformation vector specifically comprises:

TurnVector[v]＝Switch_d(State[v],Connect[v],Emd[v],LinkPoint[v])

wherein TurnVector [ v ] is the feature transformation vector; emd [ v ] is an embedding vector determined based on the associated node.

In this embodiment, the electronic device may combine the four vectors, introduce the four vectors into a preset feature vector conversion algorithm, and calculate to obtain a feature conversion vector corresponding to the medical cloud node. The feature conversion algorithm is specifically configured to project the multiple vectors obtained by the combination into a corresponding multi-dimensional space, where the multi-dimensional space includes d feature dimensions, so that all vectors can be normalized and unified, and accuracy of subsequent flow feature vectors is improved.

In S1028, the feature transformation vector is calculated by a preset updated iterative transformation algorithm.

In this embodiment, the electronic device is further configured with an update iterative conversion algorithm, and the feature transformation vector may be introduced into the update iterative conversion algorithm, so that the corresponding feature transformation vector may be uniquely solved. The update iterative conversion algorithm may specifically be:

TurnVector[v]^t+1＝Switch_d(TurnVector[v]^t,State[v])

wherein, Turnvector [ v ]]^t+1To update the iterated feature transformation vector, State [ v ]]Is a characteristic vector of a node.

In S1029, the flow feature vector is generated according to the feature transformation vector and the initial state of any medical cloud node.

In this embodiment, the electronic device may import, according to the feature transformation vector and the initial state of the medical cloud node, the feature transformation vector into a full-connection network that generates a flow feature vector, so as to obtain a corresponding flow feature vector through calculation.

In a possible implementation manner, the electronic device may perform training learning on the calculation process of the flow feature vector, where the loss function called in the training learning specifically may be:

wherein loss is the corresponding loss amount in training and learning; state (v) is the initial state of the v medical cloud node; final (v) is a flow feature vector of the v-th medical cloud node, and N is the total number of the medical cloud nodes.

In the embodiment of the application, the association characteristic vectors among the medical cloud nodes are determined and converted into the corresponding flow characteristic vectors, so that the association degree of the flow characteristic vectors among the medical cloud nodes can be improved, and the accuracy of the generation of the subsequent early warning information is improved.

Fig. 5 is a flowchart illustrating a specific implementation of the method S101 for analyzing population mobility according to the third embodiment of the present invention. Referring to fig. 5, with respect to the embodiment described in fig. 1, in the analysis method for population mobility provided in this embodiment, S101 includes: s1011 to S1014 are specifically described as follows:

in S1011, the position information fed back by the user terminal is received to generate a user flow trajectory of each user terminal.

In this embodiment, the user terminals in each monitoring area may send their own location information to the electronic device in a preset feedback period, and the electronic device may classify all the received location information according to the terminal identifiers to which the location information belongs, determine the location information belonging to different user terminals, and then sequentially connect the location information according to the sequence of the feedback time corresponding to the location information, so as to generate the user flow trajectory corresponding to the user terminal.

In S1012, according to the position staying time corresponding to each piece of position information, the node staying time of the user flow trajectory at each medical cloud node is determined.

In this embodiment, when the user terminal sends the location information to the electronic device, the location information may carry the location residence time corresponding to the location information, and the electronic device may identify a flow trajectory segment passing through a monitoring area corresponding to a certain medical cloud node according to the generated user flow trajectory, and may determine the node residence time corresponding to the medical cloud node according to the location residence time of each location information fed back in the monitoring area.

In S1013, a flow trajectory segment in which the node stay time is greater than a preset stay time threshold is selected from all the user flow trajectories as an effective flow trajectory segment.

In this embodiment, due to the length of the staying time, the influence degree of the relevant behavior performed in the monitoring area by the user corresponding to the user terminal may be determined, for example, in an epidemic situation monitoring scenario, the risk of spreading an infectious disease and the risk of infecting an infectious disease of the user may be determined, and if the staying time is short, the influence degree of the relevant behavior performed in the monitoring behavior is low and can be ignored; on the contrary, if the time of the willow branch is longer, the related behaviors performed in the monitoring behaviors have certain influence and are identified as effective flow tracks. Based on this, the electronic device can compare the node stay time of the flow track section corresponding to each medical cloud node with the stay time threshold, so that invalid flow track sections can be filtered, and valid flow track sections can be identified.

In S1014, based on all the effective flow trajectory segments, population flow information corresponding to each of the medical cloud nodes is determined.

In this embodiment, the electronic device may generate the population flow information of the medical cloud node according to all the effective flow track segments corresponding to the medical cloud node.

In the embodiment of the application, by identifying the residence time of the user flow track at the node corresponding to each medical cloud node, the invalid flow track section with short residence time can be filtered to obtain the effective flow track section, and the population flow information is generated extremely based on the effective flow ghost, so that the population flow information can be primarily screened, and the accuracy of subsequent identification is improved.

Fig. 6 is a flowchart illustrating a specific implementation of a method for analyzing population mobility according to a fourth embodiment of the present invention. Referring to fig. 6, with respect to the embodiment described in any one of fig. 1 to 5, in the analysis method for the population mobility provided in this embodiment, S104 includes: s1041 to S1048, which are detailed as follows:

in S1041, an initial identification state of each of the medical cloud nodes is acquired.

In this embodiment, each medical cloud node may determine an initial identification state corresponding to the medical cloud node according to population conditions in the associated monitoring area. Wherein, the population situation may include: the number of people staying on the day, the number of people with fever and the change rate, the number of people in class and the change rate, the drug sales and the change rate, and the number of people with abnormal body temperature and the change rate. And determining an initial identification state corresponding to the medical cloud node by comparing the population situation with the corresponding population characteristic threshold value. The initial recognition state includes an abnormal state and a normal state. If the initial identification state of the medical cloud node is an abnormal state, executing the operation of S1042; otherwise, if the initial identification state is the normal state, the operation of S1047 is performed.

In S1042, if the initial identification state of any medical cloud node is an abnormal state, a center offset between the any medical cloud node and the cluster center is calculated.

In S1043, if the central offset amount is smaller than a preset offset threshold, identifying the any medical cloud node as a normal node.

In S1044, if the center offset is greater than or equal to the offset threshold, identifying the any medical cloud node as an abnormal node.

In this embodiment, in order to avoid generating the early warning information by mistake and improve the accuracy of the early warning information, the electronic device may verify the initial identification state, and based on this, when it is determined that the initial identification state of the medical cloud node is an abnormal state, the electronic device may calculate a center offset between the medical cloud node and the cluster center. If the offset between the medical cloud node and the cluster center is large, the difference between the medical cloud node and the normal node is large, and the probability that the medical cloud node is an abnormal node is higher; on the contrary, if the offset between the medical cloud node and the clustering center is smaller, the probability that the medical cloud node is an abnormal node is smaller. Based on this, if the central offset is smaller than the offset threshold, the initial identification state is inaccurate, and the medical cloud node is identified as a normal node at this time, that is, the identification state is changed into a normal state; on the contrary, if the offset between the medical cloud node and the cluster center is greater than or equal to the offset threshold, it indicates that the initial identification state of the medical cloud node is accurately identified, and the medical cloud node can be identified as an abnormal node.

In a possible implementation manner, if there are multiple identified clustering centers, the electronic device may determine a center type corresponding to each clustering center, such as an abnormal clustering center and a normal clustering center. The electronic equipment can respectively calculate the center offset between the medical cloud node and each normal clustering center, and if the center offset between the medical cloud node and any one of the normal clustering centers is smaller than a preset offset threshold, the medical cloud node is identified as a normal node; otherwise, if the central offset between the medical cloud node and each normal clustering center is greater than or equal to the offset threshold, or the central offset between the medical cloud node and each abnormal clustering center is smaller than the offset threshold, the medical cloud node is identified as an abnormal node.

In S1045, at least one abnormal center is determined according to all the abnormal nodes.

In this embodiment, after determining each abnormal node, the electronic device may perform cluster analysis on all the abnormal nodes, so as to obtain a corresponding abnormal center.

In S1046, determining abnormal feature information of each abnormal node based on an abnormal offset between each abnormal node and the abnormal center, respectively; the abnormal feature information includes: exception type and exception level.

In this embodiment, the electronic device may calculate an abnormal offset between each abnormal node and each abnormal center, so as to determine an abnormal degree corresponding to each abnormal node, where if the abnormal offset between each abnormal node and each abnormal center is closer, the corresponding abnormal level is higher; and different abnormal centers can correspond to different abnormal types, and the abnormal type corresponding to the abnormal node can be identified and obtained according to the abnormal center with the nearest distance from the abnormal node. After the abnormal characteristic information of the abnormal center is determined, the abnormal characteristic information can be added into the early warning information, so that the accuracy of the early warning information on the description of the abnormal condition is improved.

In S1047, if the initial identification states of all the medical cloud nodes are normal states, an abnormal center is identified according to the medical cloud nodes included in each clustering center.

In this embodiment, for the case that the initial states of all the medical cloud nodes are normal states, the cluster centers whose number is smaller than the normal threshold value may be identified according to the number of the medical cloud nodes included in each cluster center, and the cluster centers obtained through the identification may be used as abnormal centers.

In S1048, the abnormal node is identified from all the medical cloud nodes according to an abnormal offset between each of the medical cloud nodes and the abnormal center.

In this embodiment, by determining the abnormal offset between each medical cloud node and the abnormal center, the medical cloud node with the abnormal offset smaller than the preset offset threshold is taken as the abnormal node.

In the embodiment of the application, the initial identification state can be verified by performing abnormal identification on the initial state of the medical cloud node in different identification modes, so that the accuracy of abnormal node identification is improved, and the occurrence of error early warning is avoided.

Fig. 7 is a flowchart illustrating a specific implementation of a method for analyzing population mobility according to a fifth embodiment of the present invention. Referring to fig. 7, with respect to any one of the embodiments in fig. 1 to 5, after the identifying, according to the clustering center, an abnormal node from all the medical cloud nodes and generating the warning information about the abnormal node, the method for analyzing population mobility according to this embodiment further includes: S701-S703 are detailed as follows:

in S701, a verification result fed back based on the warning information is received, and a parameter adjustment range is determined based on a verification deviation amount between the warning information and the verification result.

In S702, a plurality of candidate adjustment parameters are determined from the parameter adjustment range based on a preset adjustment step length, the cluster analysis algorithm is adjusted based on the candidate adjustment parameters, and a verification accuracy corresponding to each candidate adjustment parameter is obtained.

In S703, a target adjustment parameter is determined from the plurality of candidate adjustment parameters based on the verification accuracy, and the cluster analysis algorithm is adjusted based on the target adjustment parameter, so as to obtain an adjusted cluster analysis algorithm.

In this embodiment, the electronic device may reversely optimize the threshold parameter of the early warning according to the abnormal node list obtained by the identification, for example, the threshold parameter includes a parameter of a cluster analysis algorithm, a parameter of a flow feature vector generation algorithm, and a parameter for the initial identification state. The early warning threshold parameter can be adjusted by adopting a parameter searching mode. The parameter search is to perform grid search on the parameters of which the threshold needs to be adjusted to find the optimal parameters (which can ensure that the early warning result can be achieved and give consideration to the early warning accuracy and recall rate indexes) which can meet the final early warning result, and if the threshold to be adjusted is large in parameter and the search space is large, optimization methods such as variable step length and gradient descent search can be adopted to accelerate the parameter optimization and search speed. Meanwhile, besides the parameter search mode, a model reasoning method can be adopted, for example, after a batch of data is collected, a classification model is constructed through an early warning target and decision variable parameters, model training is carried out, and the early warning rule is automatically adjusted by using a decision rule and a threshold learned by a decision tree model.

In the embodiment of the application, the accuracy of parameter adjustment can be improved by adjusting the step length of the parameter, and the final adjustment parameter is determined and obtained in the effective combination, so that the algorithm accuracy is improved.

Fig. 8 is a block diagram illustrating a method and apparatus for analyzing population mobility according to an embodiment of the present invention, where the electronic device includes units for performing the steps in the embodiment corresponding to fig. 1. Please refer to fig. 1 and fig. 1 for the corresponding description of the embodiment. For convenience of explanation, only the portions related to the present embodiment are shown.

Referring to fig. 8, the apparatus for analyzing the population mobility includes:

the population mobilityinformation acquiring unit 81 is configured to receive population mobility information uploaded by each medical cloud node; each medical cloud node corresponds to one monitoring area; the population mobility information is used for determining population mobility of a monitoring area corresponding to the medical cloud node;

the flow characteristicvector generating unit 82 is used for constructing node topological graphs corresponding to all medical cloud nodes according to all the population flow information and determining flow characteristic vectors of the medical cloud nodes based on the node topological graphs;

a clustercenter determining unit 83, configured to perform cluster analysis on all the medical cloud nodes based on all the flow feature vectors to obtain at least one cluster center;

the early warninginformation generating unit 84 is configured to identify an abnormal node from all the medical cloud nodes according to the clustering center, and generate early warning information about the abnormal node.

Optionally, the flow featurevector generating unit 82 includes:

a position relation determining unit, configured to create a data node for each medical cloud node in a preset topology template according to a position relation between the monitoring areas corresponding to the medical cloud nodes; the initial state of the data node is determined based on the population state of the monitoring area corresponding to the medical cloud node;

a movement track generation unit, configured to generate a movement track of a user according to the population flow information of each medical cloud node;

the association relationship establishing unit is used for determining the association relationship among the data nodes according to the monitoring area passed by the moving track;

and the node connecting unit is used for connecting each data node based on the incidence relation and generating the node topological graph.

Optionally, the flow featurevector generating unit 82 includes:

the association feature vector generation unit is used for generating an association feature vector of any medical cloud node according to population flow information of the association cloud node which has the association relation with the medical cloud node; the associated feature vector specifically includes:

Connect[v]＝[Side(v,w₁),Side(v,w₂),...,Side(v,w_n)]

wherein, Connect [ v ]]Is the associated feature vector; v is the any medical cloud node; w is a_nThe nth associated cloud node; side (v, w)_n) Population flow information between the any medical cloud node and the nth associated cloud node; n is the total number of the associated cloud nodes;

the associated node state vector generating unit is used for constructing an associated node state vector of any medical cloud node according to the initial states of all the associated cloud nodes corresponding to the medical cloud node; the associated node state vector is specifically:

LinkPoint[v]＝[State(w₁),...,State(w_n)]

wherein LinkPoint [ v ]]Is the correlation node state vector; state (w)_n) Is the initial state of the nth association node;

a conversion vector generation unit, configured to determine a feature conversion vector of any medical cloud node in a preset multidimensional space according to the associated feature vector, the associated node state vector, and an initial state of the any medical cloud node; the feature transformation vector specifically comprises:

TurnVector[v]＝Switch_d(State[v],Connect[v],Emd[v],LinkPoint[v])

wherein TurnVector [ v ] is the feature transformation vector; emd [ v ] is an embedding vector determined based on the associated node;

the conversion vector calculation unit is used for calculating to obtain the feature conversion vector through a preset updating iterative conversion algorithm;

and the flow characteristic vector calculation unit is used for generating the flow characteristic vector according to the characteristic conversion vector and the initial state of any medical cloud node.

Optionally, the population flowinformation obtaining unit 81 includes:

the user flow track generating unit is used for receiving position information fed back by the user terminal and generating a user flow track of each user terminal;

a node stay time determining unit, configured to determine, according to a position stay time corresponding to each piece of position information, a node stay time of the user flow trajectory in each medical cloud node;

an effective flow track segment identification unit, configured to select, from all the user flow tracks, a flow track segment whose node dwell time is greater than a preset dwell time threshold as an effective flow track segment;

and the effective flow track segment packaging unit is used for determining population flow information corresponding to each medical cloud node based on all the effective flow track segments.

Optionally, the warninginformation generating unit 84 includes:

the initial identification state determining unit is used for acquiring the initial identification state of each medical cloud node;

the center offset calculation unit is used for calculating the center offset between any medical cloud node and the clustering center if the initial identification state of any medical cloud node is an abnormal state;

a normal node identification unit, configured to identify any one of the medical cloud nodes as a normal node if the center offset is smaller than a preset offset threshold;

the first abnormal node identification unit is used for identifying any medical cloud node as an abnormal node if the central offset is greater than or equal to the offset threshold;

the first abnormal center determining unit is used for determining at least one abnormal center according to all the abnormal nodes;

an abnormal feature information determining unit, configured to determine, based on an abnormal offset between each abnormal node and the abnormal center, abnormal feature information of each abnormal node respectively; the abnormal feature information includes: exception type and exception level.

Optionally, the warninginformation generating unit 84 includes:

the second abnormal center identification unit is used for identifying an abnormal center according to the medical cloud nodes contained in each clustering center if the initial identification states of all the medical cloud nodes are normal states;

and the second abnormal center identification unit is used for identifying the abnormal nodes from all the medical cloud nodes according to the abnormal offset between each medical cloud node and the abnormal center.

Optionally, the analysis apparatus further comprises:

the parameter adjustment range determining unit is used for receiving a verification result fed back based on the early warning information and determining a parameter adjustment range based on a verification deviation amount between the early warning information and the verification result;

the verification accuracy determining unit is used for determining a plurality of candidate adjusting parameters from the parameter adjusting range based on a preset adjusting step length, adjusting the clustering analysis algorithm based on the candidate adjusting parameters and obtaining the verification accuracy corresponding to each candidate adjusting parameter;

and the parameter adjusting unit is used for determining a target adjusting parameter from a plurality of candidate adjusting parameters based on the verification accuracy, and adjusting the clustering analysis algorithm based on the target adjusting parameter to obtain an adjusted clustering analysis algorithm.

Therefore, the method and the device for analyzing the population flow, provided by the embodiment of the invention, can also improve the accuracy of the flow characteristic vector, perform cluster analysis, identify and obtain abnormal nodes deviating from the normal flow behavior, and output corresponding early warning information, thereby improving the accuracy and the reliability of the early warning information.

It should be understood that, in the structural block diagram of the method and apparatus for analyzing a population flow shown in fig. 8, each module is used to execute each step in the embodiment corresponding to fig. 1 to 7, and each step in the embodiment corresponding to fig. 1 to 7 has been explained in detail in the above embodiment, specifically please refer to the relevant description in the embodiments corresponding to fig. 1 to 7 and fig. 1 to 7, which is not repeated herein.

Fig. 9 is a block diagram of an electronic device according to another embodiment of the present application. As shown in fig. 9, theelectronic apparatus 900 of this embodiment includes: aprocessor 910, amemory 920, and acomputer program 930, such as a program for a method of analyzing a population flow, stored in thememory 920 and executable on theprocessor 910. Theprocessor 910, when executing thecomputer program 930, implements the steps in the embodiments of the analysis method for human mobility described above, such as S101 to S105 shown in fig. 1. Alternatively, theprocessor 910, when executing thecomputer program 930, implements the functions of the modules in the embodiment corresponding to fig. 8, for example, the functions of theunits 81 to 84 shown in fig. 8, please refer to the related description in the embodiment corresponding to fig. 8.

Illustratively, thecomputer program 930 may be partitioned into one or more modules, which are stored in thememory 920 and executed by theprocessor 910 to accomplish the present application. One or more of the modules may be a series of computer program instruction segments capable of performing certain functions, the instruction segments being used to describe the execution ofcomputer program 930 inelectronic device 900. For example, thecomputer program 930 may be divided into a data analysis request receiving unit, a task record acquiring unit, a task data acquiring unit, and a data analysis report generating unit, each module having the above-described specific functions.

Theelectronic device 900 may include, but is not limited to, aprocessor 910, amemory 920. Those skilled in the art will appreciate that fig. 9 is merely an example of anelectronic device 900 and does not constitute a limitation of theelectronic device 900 and may include more or fewer components than shown, or some components may be combined, or different components, e.g., the electronic device may also include input-output devices, network access devices, buses, etc.

Theprocessor 910 may be a central processing unit, but may also be other general purpose processors, digital signal processors, application specific integrated circuits, off-the-shelf programmable gate arrays or other programmable logic devices, discrete hardware components, and the like. A general purpose processor may be a microprocessor or any conventional processor or the like.

Thestorage 920 may be an internal storage unit of theelectronic device 900, such as a hard disk or a memory of theelectronic device 900. Thememory 920 may also be an external storage device of theelectronic device 900, such as a plug-in hard disk, a smart card, a flash memory card, etc. provided on theelectronic device 900. Further, thememory 920 may also include both internal storage units and external storage devices of theelectronic device 900.

The above embodiments are only used to illustrate the technical solutions of the present application, and not to limit the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; such modifications and substitutions do not substantially depart from the spirit and scope of the embodiments of the present application and are intended to be included within the scope of the present application.

Claims (10)

1. A method for analyzing the flow of a population, comprising:

receiving population mobility information uploaded by each medical cloud node; each medical cloud node corresponds to one monitoring area; the population mobility information is used for determining population mobility of a monitoring area corresponding to the medical cloud node;

constructing node topological graphs corresponding to all medical cloud nodes according to all the population flow information, and determining flow characteristic vectors of the medical cloud nodes based on the node topological graphs;

importing all the flow characteristic vectors into a preset cluster analysis algorithm, and performing cluster analysis on all the medical cloud nodes to obtain at least one cluster center;

and identifying abnormal nodes from all the medical cloud nodes according to the clustering center, and generating early warning information about the abnormal nodes.

2. The analysis method according to claim 1, wherein the constructing a node topological graph corresponding to all medical cloud nodes according to all the population mobility information, and determining a mobility feature vector of the medical cloud node based on the node topological graph comprises:

according to the position relation between the monitoring areas corresponding to the medical cloud nodes, data nodes are established for the medical cloud nodes in a preset topology template; the initial state of the data node is determined based on the population state of the monitoring area corresponding to the medical cloud node;

generating a movement track of a user according to the population mobility information of each medical cloud node;

determining an incidence relation between the data nodes according to the monitoring area passed by the moving track;

and connecting each data node based on the incidence relation to generate the node topological graph.

3. The analysis method according to claim 2, wherein the constructing a node topological graph corresponding to all medical cloud nodes according to all the population mobility information, and determining a mobility feature vector of the medical cloud node based on the node topological graph comprises:

generating an association feature vector of any medical cloud node according to population mobility information of the association cloud node having the association relation with the medical cloud node; the associated feature vector specifically includes:

Connect[v]＝[Side(v,w₁),Side(v,w₂),...,Side(v,w_n)]

wherein, Connect [ v ]]Is the associated feature vector; v is the any medical cloud node; w is a_nThe nth associated cloud node; side (v, w)_n) Population flow information between the any medical cloud node and the nth associated cloud node; n is the total number of the associated cloud nodes;

constructing an associated node state vector of any medical cloud node according to the initial states of all the associated cloud nodes corresponding to the medical cloud node; the associated node state vector is specifically:

LinkPoint[v]＝[State(w₁),...,State(w_n)]

wherein LinkPoint [ v ]]Is the correlation node state vector; state (w)_n) Is the initial state of the nth association node;

determining a feature transformation vector of any medical cloud node in a preset multidimensional space according to the associated feature vector, the associated node state vector and the initial state of any medical cloud node; the feature transformation vector specifically comprises:

TurnVector[v]＝Switch_d(State[v],Connect[v],Emd[v],LinkPoint[v])

wherein TurnVector [ v ] is the feature transformation vector; emd [ v ] is an embedding vector determined based on the associated node;

calculating to obtain the feature transformation vector through a preset updating iterative transformation algorithm;

and generating the flow feature vector according to the feature conversion vector and the initial state of any medical cloud node.

4. The analysis method according to claim 1, wherein the receiving of the population flow information uploaded by each medical cloud node comprises:

receiving position information fed back by a user terminal to generate a user flow track of each user terminal;

determining the stay time of the user flow track at the node corresponding to each medical cloud node according to the stay time of the position corresponding to each piece of position information;

selecting a flow track section with the node stay time being larger than a preset stay time threshold value from all the user flow tracks as an effective flow track section;

and determining population flow information corresponding to each medical cloud node based on all the effective flow track sections.

5. The analysis method according to any one of claims 1 to 4, wherein the identifying abnormal nodes from all the medical cloud nodes according to the clustering center and generating early warning information about the abnormal nodes comprises:

acquiring an initial identification state of each medical cloud node;

if the initial identification state of any medical cloud node is an abnormal state, calculating the center offset between any medical cloud node and the clustering center;

if the central offset is smaller than a preset offset threshold, identifying any medical cloud node as a normal node;

if the central offset is greater than or equal to the offset threshold, identifying any medical cloud node as an abnormal node;

determining at least one abnormal center according to all the abnormal nodes;

respectively determining abnormal characteristic information of each abnormal node based on the abnormal offset between each abnormal node and the abnormal center; the abnormal feature information includes: exception type and exception level.

6. The analysis method according to claim 5, wherein after obtaining the initial identification state of each medical cloud node, further comprising:

if the initial identification states of all the medical cloud nodes are normal states, identifying an abnormal center according to the medical cloud nodes contained in each clustering center;

and identifying the abnormal nodes from all the medical cloud nodes according to the abnormal offset between each medical cloud node and the abnormal center.

7. The analysis method according to any one of claims 1 to 4, wherein after the identifying abnormal nodes from all the medical cloud nodes according to the clustering center and generating early warning information about the abnormal nodes, the method further comprises:

receiving a verification result fed back based on the early warning information, and determining a parameter adjustment range based on a verification deviation amount between the early warning information and the verification result;

determining a plurality of candidate adjusting parameters from the parameter adjusting range based on a preset adjusting step length, adjusting the clustering analysis algorithm based on the candidate adjusting parameters, and obtaining a verification accuracy rate corresponding to each candidate adjusting parameter;

and determining target adjustment parameters from a plurality of candidate adjustment parameters based on the verification accuracy, and adjusting the cluster analysis algorithm based on the target adjustment parameters to obtain an adjusted cluster analysis algorithm.

8. An apparatus for analyzing the flow of a person, comprising:

the population mobility information acquisition unit is used for receiving population mobility information uploaded by each medical cloud node; each medical cloud node corresponds to one monitoring area; the population mobility information is used for determining population mobility of a monitoring area corresponding to the medical cloud node;

the flow characteristic vector generating unit is used for constructing node topological graphs corresponding to all medical cloud nodes according to all the population flow information and determining flow characteristic vectors of the medical cloud nodes based on the node topological graphs;

the cluster center determining unit is used for performing cluster analysis on all the medical cloud nodes based on all the flow characteristic vectors to obtain at least one cluster center;

and the early warning information generating unit is used for identifying abnormal nodes from all the medical cloud nodes according to the clustering center and generating early warning information about the abnormal nodes.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method of any of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, in which a computer program is stored which, when being executed by a processor, carries out the method according to any one of claims 1 to 7.

Images