KR102583585B1 - Method and system of predicting infectious diseases using big data and artificial intelligence - Google Patents

Abstract

A method of predicting an infectious disease using an infectious disease prediction system according to an embodiment of the present specification includes generating a plurality of grids of a certain size on a digital map; Based on the generated grid, a first input variable related to the floating population collected through a wireless communication network within the grid, a second input variable related to public transportation collected through a traffic information network, public information collected through a public information network, and calculating a related third input variable; Calculating a plurality of infectious disease variables provided from a disease surveillance information module; Extracting feature values from the first input variable, the second input variable, the third input variable, and the plurality of infectious disease variables, and inputting the extracted feature values into a pre-trained deep learning model; and predicting infectious disease information about infectious diseases on the digital map based on the output of the deep learning model.

Description

Infectious disease prediction method and system using big data and artificial intelligence {METHOD AND SYSTEM OF PREDICTING INFECTIOUS DISEASES USING BIG DATA AND ARTIFICIAL INTELLIGENCE}

This specification relates to a method and system for predicting infectious diseases using big data and artificial intelligence. More specifically, the occurrence and death of infectious diseases by administrative district is based on artificial intelligence and mathematical modeling using various information collected from real-time wireless and public communication networks. , relates to an infectious disease prediction method and system that can support infectious disease quarantine by predicting hospitalization and increase/decrease.

Infectious diseases are one of the most important problems related to human health, and humanity is making great efforts to treat and prevent various infectious diseases (e.g. cholera, tuberculosis, measles, COVID-19, etc.) there is. In order to deal with infectious diseases, a system is needed that can take preliminary measures to prevent the spread of infectious diseases and take measures to end the epidemic when it occurs. For this purpose, a general infectious disease analysis and prevention system identifies the past epidemic periods and regions of infectious diseases and supports visual confirmation.

To describe the above in technical detail, a general infectious disease analysis and prevention system only supports the simple past occurrence of infectious diseases by showing them on a map so that they can be visually identified.

[Registered Patent 10-1976189], (Title of invention: Method for providing analysis service of floating population)

This specification has been calculated to solve the above problems, and includes floating population information collected from a real-time wireless communication network, bus/subway road traffic information, daily flight number information, infectious disease epidemiology information, weather information, date information, economic indicators, It provides a method of predicting infectious diseases using an infectious disease prediction system that can support infectious disease quarantine by predicting the occurrence, death, hospitalization, and increase/decrease of infectious diseases by administrative district through artificial intelligence and mathematical modeling using population and population movement information. There is a purpose.

The technical objectives to be achieved in this specification are not limited to the matters mentioned above, and other technical tasks not mentioned can be solved by those skilled in the art from the embodiments of this specification described below. can be considered.

Hereinafter, an infectious disease prediction method and system using big data and artificial intelligence will be described as embodiments of the present specification.

An infectious disease prediction method according to an embodiment of the present specification includes generating a plurality of grids of a certain size on a digital map; Based on the generated grid, a first input variable related to the floating population collected through a wireless communication network within the grid, a second input variable related to public transportation collected through a traffic information network, public information collected through a public information network, and calculating a related third input variable; Calculating a plurality of infectious disease variables provided from a disease surveillance information module; Extracting feature values from the first input variable, the second input variable, the third input variable, and the plurality of infectious disease variables, and inputting the extracted feature values into a deep learning model; and predicting infectious disease information about infectious diseases on the digital map based on the output of the deep learning model.

In addition, the deep learning model is supervised based on the extracted feature values, and may include constructing an infectious disease prediction model through this and periodically updating it.

In addition, the step of predicting the infectious disease information may include calculating N days by predicting variable values output through the deep learning model, and predicting variable values for a specific date based on the calculated N days. .

In addition, the deep learning model includes mathematical modeling SI, SIS, SIR, SIRS, SEIR, SEIRS, etc., and the step of predicting the infectious disease information involves predicting the SEIR variable value of a specific date through the SEIR, The SEIR variable value on the specific date is calculated by multiplying the SEIR variable value on the date before the specific date by the average increase rate of the value of each SEIR variable on the previous N days prior to the specific date and the date following the N days prior to the specific date. It may include predicting.

In addition, the pre-trained deep learning model may include setting a certain lag time between the infectious disease transmission risk and its increase/decrease rate and infectious disease occurrence, death, and its increase/decrease rate when constructing the infectious disease prediction model.

Another embodiment of the present invention provides an infectious disease prediction system, comprising: a communication module for transmitting and receiving data based on at least one of a wireless communication network, a traffic information network, and a public information network; A disease surveillance information module that stores infectious disease information; and generating a plurality of grids of a certain size on a digital map, and based on the generated grids, a first input variable related to the floating population collected through the wireless communication network within the grid, collected through the traffic information network. Calculate a second input variable related to public transportation, a third input variable related to public information collected through the public information network, calculate a plurality of infectious disease variables based on infectious disease information provided from the disease surveillance information module, and calculate the first input variable. 1 Extract feature values from the input variable, the second input variable, the third input variable, and the plurality of infectious disease variables, input the extracted feature values into a deep learning model, and based on the output of the deep learning model A system for predicting infectious diseases is provided, including a processor that predicts infectious disease information about infectious diseases on the digital map.

In addition, the deep learning model is supervised based on the extracted feature values, and through this, an infectious disease prediction model is built and periodically updated.

In addition, the processor predicts variable values output through the deep learning model to calculate N days, and predicts variable values for a specific date based on the calculated N days.

In addition, the deep learning model includes mathematical modeling SI, SIS, SIR, SIRS, SEIR, SEIRS, etc., and when predicting the SEIR variable value of a specific date through the SEIR, the processor predicts the SEIR variable value of the specific date. Predicting the variable value of SEIR on the specific date by multiplying the average growth rate of the value of each SEIR variable on the previous N days previous and the next N days prior to the specific date by the SEIR variable value on the date prior to the specific date. It is characterized by

In addition, the deep learning model is characterized by setting a certain lag time between the risk of infectious disease transmission and its increase/decrease rate and the occurrence of infectious disease, death, and its increase/decrease rate when constructing the infectious disease prediction model.

The aspects of the present specification described above are only some of the preferred embodiments of the present specification, and various embodiments reflecting the technical features of the present specification can be understood by those skilled in the art. It can be derived and understood based on the explanation.

According to the embodiments of the present specification, the following effects can be achieved.

The infectious disease prediction system according to the embodiment of this specification calculates a social distancing index using floating population information collected in real time from wireless carriers and population movement information on buses, subways, road traffic, and flights, and uses it in a prediction model to determine social distancing. Accuracy can be improved in predicting the increase or decrease in the number of new infectious diseases, deaths, and hospitalizations, such as COVID-19, which plays an important role in the spread of the virus.

The effects that can be obtained from the embodiments of the present specification are not limited to the effects mentioned above, and other effects not mentioned are commonly known in the technical field to which the present specification pertains from the description of the embodiments of the present specification below. It can be clearly derived and understood by those with knowledge. That is, unintended effects resulting from implementing the present specification may also be derived from the embodiments of the present specification by those skilled in the art.

BRIEF DESCRIPTION OF THE DRAWINGS To facilitate understanding of the present specification, the accompanying drawings are included as part of the detailed description and provide various embodiments of the present specification. Additionally, the accompanying drawings, together with the detailed description, are used to describe embodiments of the present specification.
Figure 1 illustrates the configuration of a system for acquiring learning data to which this specification can be applied.
Figure 2 is a diagram for explaining an infectious disease prediction system to which this specification can be applied.
Figure 3 is a diagram for explaining a method of collecting a first input variable according to an embodiment of the present specification.
4 to 7 are diagrams for explaining a method of collecting second input variables according to an embodiment of the present specification.
Figures 8 to 16 are diagrams for explaining a method of calculating predictor values for a specific date using a deep learning model according to an embodiment of the present specification.

Hereinafter, a method for predicting an infectious disease using an infectious disease prediction system will be described as embodiments of the present specification.

The following embodiments combine the elements and features of this specification in a predetermined form. Each component or feature may be considered optional unless explicitly stated otherwise. Each component or feature may be implemented in a form that is not combined with other components or features. Additionally, some components and/or features may be combined to configure embodiments of the present specification. The order of operations described in the embodiments of this specification may be changed. Some features or features of one embodiment may be included in other embodiments or may be replaced with corresponding features or features of other embodiments.

In the description of the drawings, parts, devices, and/or configurations that may obscure the gist of the present specification are not described, and parts, devices, and/or configurations that can be understood at a level by those skilled in the art are not described. Additionally, parts referred to using the same reference numerals in the drawings refer to the same components or steps in the device configuration or method.

Throughout the specification, when a part is said to “comprise or include” a certain element, this means that it does not exclude other elements but may further include other elements, unless specifically stated to the contrary. do. In addition, terms such as "~unit" or "~gi" used in the specification mean a unit that processes at least one function or operation. Also, "a or an", "one", " "The" and similar related terms include both the singular and the plural in the context of describing the invention (especially in the context of the claims below), unless otherwise indicated herein or clearly contradicted by context. It can be used with meaning.

In addition, specific terms and/or symbols used in the embodiments of the present specification are provided to aid understanding of the present specification, and the use of such specific terms may be used in other forms without departing from the technical spirit of the present specification. It can be changed to .

Figure 1 illustrates the configuration of a system for acquiring learning data to which this specification can be applied.

This specification includes infectious disease epidemiological information, weather information, date information, economic indicators, population and population movement information, as well as floating population information collected from real-time wireless communication networks, bus, subway, and road traffic information, and daily flight number information. It helps prevent new infectious diseases such as COVID-19, where social distancing is important, by using social distancing indicators that reflect the population and predicting the probability of infectious disease occurrence, death, and hospitalization by administrative district through a learning model using artificial intelligence. can be given.

Looking at Figure 1, the configuration of a system for acquiring training data for such an AI model is illustrated. Referring to FIG. 1 , the infection system 100 may include a processor 110, a memory 130, and a communication module 150.

The processor 110 may include one or more application processors (AP), one or more communication processors (CP), or at least one artificial intelligence processor (AI processor). The application processor, communication processor, or AI processor may each be included in different integrated circuit (IC) packages or may be included in one IC package.

The application processor runs an operating system or application program, controls multiple hardware or software components connected to the application processor, and can perform various data processing/computations, including multimedia data. As an example, an application processor may be implemented as a system on chip (SoC). The processor 110 may further include a graphics processing unit (GPU).

The communication processor can perform the function of managing data links and converting communication protocols in communication with networked wireless communication networks, traffic information networks, public information networks, disease surveillance information modules, and prediction-based service modules. As an example, a communication processor may be implemented as a SoC. The communications processor may perform at least some of the multimedia control functions.

Additionally, the communication processor may control data transmission and reception of the communication module 150. The communication processor may be implemented to be included as at least part of the application processor. For example, the data may include a first input variable related to floating population, a second input variable related to public transportation, a third input variable related to public information, a plurality of infectious disease variables, and predicted infectious disease data (or information). .

The application processor or communication processor may load and process commands or data received from at least one of the non-volatile memory or other components connected to the volatile memory. Additionally, the application processor or communication processor may store data received from or generated by at least one of the other components in a non-volatile memory.

Memory 130 may include internal memory or external memory. Built-in memory may be volatile memory (e.g., dynamic RAM (DRAM), static RAM (SRAM), synchronous dynamic RAM (SDRAM), etc.) or non-volatile memory (e.g., one time programmable ROM (OTPROM), It may include at least one of programmable ROM (PROM), erasable and programmable ROM (EPROM), electrically erasable and programmable ROM (EEPROM), mask ROM, flash ROM, NAND flash memory, NOR flash memory, etc.). According to one example, the internal memory may take the form of a solid state drive (SSD). External memory may be a flash drive, such as compact flash (CF), secure digital (SD), micro secure digital (Micro-SD), mini secure digital (Mini-SD), extreme digital (xD), or It may include a memory stick, etc.

The communication module 150 may include a wireless communication module or an RF module. The wireless communication module may include, for example, Wi-Fi, BT, GPS or NFC. For example, a wireless communication module may provide a wireless communication function using radio frequencies. Additionally or alternatively, the wireless communication module includes a wireless communication network 210, a traffic information network 230, a public information network 250, a disease surveillance information module 270, and a prediction-based service module 290 (see FIG. 2) over a network (e.g. : May include a network interface or modem for connection to (Internet, LAN, WAN, telecommunication network, cellular network, satellite network, POTS or 5G network, etc.).

The RF module may be responsible for transmitting and receiving data, for example, RF signals or called electronic signals. As an example, the RF module may include a transceiver, a power amp module (PAM), a frequency filter, or a low noise amplifier (LNA). Additionally, the RF module may include components for transmitting and receiving electromagnetic waves in free space in wireless communication, for example, conductors or wires.

The processor 110 may include an AI processor 111, a data learning unit 111a, a data pre-processing unit 111b, a data selection unit 111c, a model evaluation unit 111d, and a prediction module 115. .

The AI processor 111 can learn a neural network using a program stored in the memory 330. In particular, the AI processor 111 decodes the prediction of infectious diseases using the first input variable, second input variable, third input variable, and infectious disease variable input from the wireless communication network, traffic information network, public information network, and disease surveillance information module. You can learn a neural network for Here, the neural network may be designed to simulate human brain structure (eg, neuron structure of a human neural network) on a computer. A neural network may include an input layer, an output layer, and at least one hidden layer. Each layer includes at least one neuron with a weight, and the neural network may include synapses connecting neurons. In a neural network, each neuron can output an input signal input through a synapse as a function value of an activation function for weight and/or bias.

Multiple network modes can exchange data according to each connection relationship to simulate the synaptic activity of neurons sending and receiving signals through synapses. Here, the neural network may include a deep learning model developed from a neural network model. In a deep learning model, multiple network nodes are located in different layers and can exchange data according to convolutional connection relationships. Examples of neural network models include deep neural networks (DNN), convolutional neural networks (CNN), recurrent neural networks, restricted Boltzmann machines, and deep belief networks. ), deep Q-Network, and various deep learning techniques, and can be applied in fields such as vision recognition, speech recognition, natural language processing, and voice/signal processing.

Meanwhile, the processor 110 that performs the functions described above may be a general-purpose processor (eg, CPU), or may be an AI-specific processor (eg, GPU) for artificial intelligence learning.

The memory 130 includes first input variables, second input variables, third input variables, and infectious disease variables input from the wireless communication network 210, traffic information network 230, public information network 250, and disease surveillance information module 270. It is possible to store various programs, infectious disease data, and infectious disease prediction data necessary for the operation of the infection prediction system 100 to decipher the prediction of infectious diseases related to infectious diseases. The memory 130 is accessed by the AI processor 111, and reading/writing/modifying/deleting/updating data by the AI processor 111 can be performed. Additionally, the memory 130 may store a neural network model (eg, deep learning model) generated through a learning algorithm for data classification/recognition according to an embodiment of the present specification. Furthermore, the memory 330 may store not only the learning model 221 but also input data, learning data, and learning history.

Meanwhile, the AI processor 111 may include a data learning unit 111a that learns a neural network for data classification/recognition. The data learning unit 111a can learn standards regarding which learning data to use to determine data classification/recognition and how to classify and recognize data using the learning data. The data learning unit 111a can learn a deep learning model by acquiring training data to be used for learning and applying the acquired training data to the deep learning model.

The data learning unit 111a can be manufactured in the form of at least one hardware chip and mounted on the infection prediction system 100 to decipher the prediction of infectious diseases related to infectious diseases. As an example, the data learning unit 111a may be manufactured in the form of a dedicated hardware chip for artificial intelligence, and may be manufactured as part of a general-purpose processor (CPU) or graphics processor (GPU) to decipher predictions of infectious diseases related to infectious diseases. It can be mounted on the infection prediction system 100. Additionally, the data learning unit 111a may be implemented as a software module. When implemented as a software module (or a program module including instructions), the software module may be stored in a non-transitory computer readable recording medium that can be read by a computer. In this case, at least one software module may be provided to an operating system (OS) or may be provided by an application.

The data learning unit 111a can use the acquired training data to train the neural network model to have a judgment standard on how to classify/recognize certain data. At this time, the learning method by the data learning unit 111a can be classified into supervised learning, unsupervised learning, and reinforcement learning. Here, supervised learning refers to a method of training an artificial neural network with a given label for the learning data, and the label is the correct answer (or result value) that the artificial neural network must infer when the learning data is input to the artificial neural network. It can mean. Unsupervised learning can refer to a method of training an artificial neural network in a state where no labels for training data are given. Reinforcement learning can refer to a method of training an agent defined within a specific environment to select an action or action sequence that maximizes the cumulative reward in each state. Additionally, the data learning unit 111a may learn a neural network model using a learning algorithm including error backpropagation or gradient descent. When a neural network model is learned, the learned neural network model may be referred to as a learning model 131. The learning model 131 is stored in the memory 130 and can be used to infer results for new input data other than training data.

Meanwhile, the AI processor 111 includes a data pre-processing unit 111b and/or a data selection unit to improve the analysis results using the learning model 131 or to save resources or time required for generating the learning model 131. (111c) may further be included.

The data preprocessing unit 111b may preprocess the acquired data so that the acquired data can be used for learning/inference for situation determination.

The data selection unit 111c may select data required for learning from the learning data preprocessed by the data learning unit 111a or the data preprocessing unit 111b. Selected learning data may be provided to the learning model 131. Additionally, the data selection unit 111c may select data required for inference among input data obtained through an input device or input data preprocessed in a preprocessor.

Additionally, the AI processor 111 may further include a model evaluation unit 111d to improve the analysis results of the neural network model. The model evaluation unit 111d inputs evaluation data into the neural network model, and when the analysis result output from the evaluation data does not meet a predetermined standard, it can cause the model learning unit to learn again. In this case, the evaluation data may be preset data for evaluating the learning model 131. As an example, the model evaluation unit 111d determines that among the analysis results of the learned neural network model for the evaluation data, when the number or ratio of evaluation data for which the analysis result is inaccurate exceeds a preset threshold, the model evaluation unit 111d determines that the analysis result does not meet a predetermined standard. It can be evaluated as

By introducing mathematical modeling techniques through artificial intelligence, the prediction module 115 can predict the spread of new infectious diseases whose transmission routes are complex and complex and whose exact transmission routes are not yet known. The prediction module 115 can utilize various algorithms used for epidemic prediction and future prediction. The prediction module 115 can predict infectious diseases for the entire country or for each administrative district.

Figure 2 is a diagram for explaining an infectious disease prediction system to which this specification can be applied.

Looking at Figure 2, the infectious disease prediction system 100 according to an embodiment of the present specification includes a processor 110, a wireless communication network 210, a traffic information network 230, a public information network 250, and a disease surveillance information module 270. may include.

Description of the portion of the processor 110 to be described in FIG. 2 that overlaps with the processor 110 described in FIG. 1 will be omitted.

The processor 110 generates a plurality of grids of a certain size on a digital map, and based on the generated grids, the first input variable related to the floating population collected through the wireless communication network 210 within the grid, the traffic information network ( A second input variable related to public transportation collected through 230) and a third input variable related to public information collected through the public information network 250 can be calculated. The processor 110 may be referred to as an information processing module.

For example, the processor 110 may calculate or extract first to third input variables using the data integration module 11 and the variable extraction model 12.

The processor 110 can calculate a plurality of infectious disease variables provided from the disease surveillance information module 270.

The processor 110 extracts feature values from the first input variable, the second input variable, the third input variable, and a plurality of infectious disease variables, inputs the extracted feature values into a pre-trained deep learning model, and Based on the output, infectious disease information about infectious diseases can be predicted on a digital map.

For example, by introducing mathematical modeling techniques through artificial intelligence, the processor 110 can effectively predict the spread of new infectious diseases whose transmission routes are complex and complex and whose exact transmission routes are not yet known.

Mathematical modeling may include SI, SIS, SIR, SIRS, SEIR, SEIRS, etc. Hereinafter, the technology can be utilized by unifying it into the SEIR model or SEIR model. It is not limited to this. For example, in the case of mathematical modeling through artificial intelligence, algorithms such as Random Forest, ARIMA, LSTM (long short-term memory), XBboost, Light GBM, etc., and the recently developed LSTM (Interpretable sequence learning for covid-19) forecasting models can be used. It is not limited to this, and the processor 110 may utilize various algorithms used for epidemic prediction and future prediction. For example, the processor includes a prediction module 115, and can use the prediction module 115 to predict infectious diseases for the entire country or to predict infectious diseases for each administrative district.

The processor 110 can build a deep learning model using mathematical modeling through artificial intelligence. For example, a pre-trained deep learning model can be called a prediction model, prediction model, infectious disease prediction model, or infectious disease prediction model.

When constructing a prediction model, the processor 110 may set a certain lag time between the infectious disease transmission risk and its increase/decrease rate and the infectious disease occurrence, death, and its increase/decrease rate. At this time, the processor 110 can set the infectious disease transmission risk and its increase/decrease rate as independent variables, and the infectious disease occurrence, death, and its increase/decrease rate as outcome variables. The processor 110 can set the lag time by comprehensively considering the average incubation period of each infectious disease and the time it takes to confirm the diagnosis.

The processor 110 performs real-time supervised learning using artificial intelligence, and through this, the infectious disease prediction model can be periodically updated. For example, the outcome variables (or predictor variables) obtained through this are the daily number of cases by infectious disease by administrative district, sex/age group, rate of increase/decrease in the number of cases, number of outpatients, rate of increase/decrease in the number of outpatients, number of inpatients, rate of increase/decrease in the number of inpatients. , the number of patients admitted to the intensive care unit, the rate of change in the number of patients admitted to the intensive care unit, the number of deaths, the rate of change in the number of deaths, etc., the number of daily cases by infectious disease by sex/age group in the entire country, the rate of change in the number of cases, the number of outpatients, the rate of change in the number of outpatients, hospitalization It may be the number of patients, the rate of increase or decrease in the number of hospitalized patients, the number of patients admitted to the intensive care unit, the rate of increase or decrease in the number of patients hospitalized in the intensive care unit, the number of deaths, or the rate of increase or decrease in the number of deaths.

The processor 110 receives information related to the floating population by grid and time zone from the wireless communication network 210 through the communication module 150, selection of weekly floating population hotspots, daily summation of floating population in the hotspot grid, and daily summing of floating population by administrative district. A first input variable may be provided. A detailed explanation of this will be provided later.

The processor 110 receives a second input variable, which is information related to the total number of passengers getting on and off the bus per day, the total number of passengers getting on and off the subway per day, the daily road traffic volume, and the number of daily flights, from the traffic information network 230 through the communication module 150. You can. A detailed explanation of this will be provided later.

The processor 110 can receive a third input variable, which is information related to weather information, economic indicator information, population and population movement information, and date information/other statistical information, from the public information network 250 through the communication module 150.

The processor 110 extracts some of the third input variables using real-time weather information, and can be used for modeling the risk of infectious disease transmission based on some of the extracted input variables. For example, weather information may include daily average relative humidity, daily average temperature, daily average wind speed, daily maximum temperature, daily minimum temperature, daily total solar radiation, daily precipitation, daily total sunlight hours, etc.

The processor 110 can extract some of the third input variables using real-time economic indicators and use them for modeling the risk of infectious disease transmission based on some of the extracted input variables. For example, economic indicators include KRW/US dollar (basic trading rate), KRW/Japanese yen (100 yen), KRW/Euro, KRW/British pound, KRW/Canadian dollar, KRW/Hong Kong dollar, KRW/Australian dollar, KRW/ Dollar (closing price), KOSPI index, trading volume (stock market, provisional value), trading amount (stock market, provisional value), foreign net purchase (stock market, provisional value), stock market-trading volume (including 10,000 shares, after-hours trading), stock market- Trading amount (KRW 100 million, including after-hours trading), KOSDAQ index, trading volume (Manchuria: KOSDAQ market, provisional value), Trading amount (KRW 100 million: KOSDAQ market, provisional value), foreign net purchase (KOSDAQ market, provisional value), market capitalization (stock market, provisional value) It may include the provisional value), Bank of Korea base interest rate, fund-adjusted loan interest rate, and fund-adjusted deposit interest rate.

The processor 110 can extract some of the third input variables using real-time date information and use them for modeling the risk of infectious disease transmission based on some of the extracted input variables. For example, date information may include month, day of the week, whether it is a holiday, whether it is a summer vacation period, the date on which a specific event occurs, and other date types.

In addition, the processor 110 can receive a plurality of infectious disease variables from the disease surveillance information module 270 through the communication module 150 and use them for modeling the risk of infectious disease transmission. Multiple infectious disease variables may include infectious disease epidemiology information, infectious disease information in foreign countries, infectious disease prevention information, and infectious disease characteristic information.

For example, infectious disease epidemiology information includes the cumulative number of confirmed cases by administrative district, the number of changes compared to the previous day (number of new confirmed cases) by administrative district, the number of quarantined people by administrative district, the number of release from quarantine by administrative district, the number of deaths by administrative district, Number of new overseas inflows by administrative district, death rate compared to national confirmation rate, nationwide cumulative confirmation rate, nationwide cumulative number of tests in progress, nationwide number of completed tests, number of patients under treatment nationwide, number of people released from quarantine nationwide, number of deaths nationwide, cumulative number of confirmed cases nationwide Number, number of nationwide tests conducted, number of negative test results nationwide, cumulative number of confirmed cases by administrative district (sum of cumulative number of confirmed cases by autonomous district), cumulative number of overseas inflows by administrative district (sum of cumulative number of overseas inflows by autonomous district), by administrative district It can include the number of infected subjects, the current number of people tested by administrative district, the current number of confirmed cases by administrative district, the current number of quarantined people by administrative district, and the current number of recoveries + number of deaths by administrative district.

Infectious disease information in overseas countries may include the daily number of confirmed cases and deaths in each overseas country, and the search frequency of keywords related to COVID-19 in each overseas country. Here, the information on the history of infectious diseases is information on infectious diseases that have occurred in the country and whether the infectious diseases have become endemic, as reported by each country. Informatics, where each country reports the infectious diseases that have occurred in the country and whether the infectious diseases have become endemic, etc. It can be obtained from an analysis site. (Example: GIDEON, ProMED-mail, WHO COVID-19 Dashboard, Korea Centers for Disease Control and Prevention website, etc.)

Infectious disease prevention information may include whether a vaccine has been developed, the vaccination rate by province or province, and the vaccination rate by country.

Additionally, multiple infectious disease variables can utilize other statistical information. Multiple infectious disease variables can utilize comprehensive big data information that includes the following variables, such as demographic information collected in non-real time, infectious disease characteristic information, and other statistical information by administrative district. For example, demographic information may include gender by administrative district, population by age group, population density by administrative district, etc., and infectious disease characteristic information may include the name of the infectious disease and the type of transmission route (water-borne, insect-borne, air-borne, contact). transmission, droplet transmission, etc.), criteria for classification of legally infectious diseases, incubation period, and whether or not it is a new infectious disease. Other statistical information by administrative district can be obtained from the National Statistical Office.

As described above, the infection prediction system 100 according to an embodiment of the present specification may transmit a service using the prediction result for the occurrence of an infectious disease under the control of the processor 110. For example, the infection prediction system 100 provides basic information for customized quarantine by region through regional infectious disease prediction, a service that allows medical resources to be deployed in advance, and preemptive support for damage caused by infectious diseases. , it can be connected to a service that allows you to respond.

Figure 3 is a diagram for explaining a method of collecting a first input variable according to an embodiment of the present specification.

Referring to FIG. 3, the infectious disease prediction system 100 according to an embodiment of the present specification may collect a first input variable related to the floating population through a wireless communication network within the grid. The infectious disease prediction system 100 can collect information collected through subscriber wireless terminals in the wireless communication network 210.

The infectious disease prediction system 100 may generate a grid of a certain size on a digital map under the control of the processor 110. The grid may be adjusted to a predetermined size based on various infectious disease data or infectious disease prediction data under the control of the processor 110. Multiple adjusted grids can be set.

The infectious disease prediction system 100 can calculate, calculate, or collect the floating population of wireless communication subscribers by time zone using the wireless communication network 210 in each grid under the control of the processor 110 (S211). For example, certain time slots may be 1 hour intervals, etc. For example, the infectious disease prediction system 100 can calculate the number of floating populations during daytime hours or the number of floating populations during nighttime hours for each administrative district and grid under the control of the processor 110.

The infectious disease prediction system 100 may select a grid with the largest floating population among a plurality of grids under the control of the processor 110 (S212).

Thereafter, the infectious disease prediction system 100 may calculate the daily combined floating population of the selected grid under the control of the processor 110 (S213). Among multiple grids, the grid with the largest floating population can be called a floating population hotspot. The infectious disease system 100 can calculate the daily floating population for each administrative district based on floating population hotspots (S214). The calculated floating population number can be referred to as the first input variable. The first input variable may be a variable related to the floating population collected through the wireless communication network 210 within the grid.

4 to 7 are diagrams for explaining a method of collecting second input variables according to an embodiment of the present specification.

Referring to FIGS. 4 to 7 , the infectious disease prediction system 100 according to an embodiment of the present specification may collect second input variables related to public transportation through the traffic information network 2300.

The second input variable may be a variable that generates an index reflecting the daily floating population through various transportation information such as buses, subways, road traffic, and flights.

Referring to FIG. 4, the infectious disease prediction system 100 can collect the floating population by administrative district through information on the number of passengers getting on and off buses.

The infectious disease prediction system 100 uses the traffic information network 230 under the control of the processor 110 to calculate the total number of passengers getting on and off per day by adding the total number of daily boarding passengers (S221) and the total number of daily disembarking passengers (S222) for each bus stop. It can be calculated (S223).

The infection prediction system 110, under the control of the processor 110, adds up the total number of daily boarding and alighting passengers for each administrative district to which the corresponding bus stop belongs (S223), and calculates the total number of daily boarding and alighting passengers for each administrative district based on this (S224) ), the floating population can be collected.

Referring to FIG. 5, the infectious disease prediction system 100 can collect the floating population by administrative district through information on the number of passengers getting on and off the subway.

The infectious disease prediction system 100 uses the traffic information network 230 under the control of the processor 110 to calculate the total number of passengers getting on and off per day by adding the total number of daily passengers getting on (S225) and the total number of daily getting off (S226) for each subway station. You can do it (S227).

The infectious disease prediction system 100, under the control of the processor 110, adds up the total number of passengers getting on and off per day for each administrative district to which the subway station belongs, and calculates the total number of passengers getting on and off per day for each administrative district based on this (S228). Population numbers can be collected.

Referring to FIG. 6, the infectious disease prediction system 100 can collect daily traffic volume for each road or business office and the floating population for each administrative district to which the road or business office belongs.

The infectious disease prediction system 100 uses the traffic information network 230 under the control of the processor 110 to add up the daily traffic volume (S229) for each road or business office by administrative district to which the road or business office belongs, and based on this, each administrative district. By calculating the daily traffic volume (S230), the number of floating populations can be collected.

Referring to FIG. 7, the infectious disease prediction system 100 can collect floating population by utilizing real-time aircraft operation information using the traffic information network 230.

The infectious disease prediction system 100, under the control of the processor 110, classifies flights arriving and departing from Korea on international flights by country and can extract the number of daily flights to Korea and the number of arrivals for each country (S231 ).

The infectious disease prediction system 100 can access the Korea Disease Control and Prevention Agency under the control of the processor 110 and extract the number of infectious disease cases in each overseas country in real time (S232).

The infectious disease prediction system 100 can use wireless communication data to extract the number of roaming customers entering Korea from each overseas country (S233).

The infectious disease prediction system 100 matches the extracted number of entrants, the number of infectious disease outbreaks by country, and the number of roaming customer entrants (S234), and calculates the daily risk of spreading infectious diseases imported from overseas based on the matched data (S235). ).

Figures 8 to 16 are diagrams for explaining a method of calculating predictor values for a specific date using a deep learning model according to an embodiment of the present specification.

Referring to FIG. 8, the processor 110 of the present specification calculates the N days that best predict the variable value using a deep learning model, and can predict the variable value of a specific date based on this. In this specification, the explanation will be based on SEIR, a mathematical modeling among deep learning models. It is not limited to this and various deep learning models can be applied. SEIR, which will be described later, can be referred to as the SEIR model, and the population group is classified into four groups: S: Susceptible, E: Exposed, I: Infective, and R: Recovered. It can be defined as a model that mathematically predicts infectious diseases.

The variable value may be a predicted value obtained by analyzing feature values extracted from the above-described first to third input variables and infectious disease variables.

When predicting the SEIR variable value of a specific date through mathematical modeling, SEIR, the processor multiplies the SEIR variable value of the date before the specific date by the average increase rate of each SEIR variable value of the previous date and the next date N days before the specific date to obtain the average increase rate of the SEIR variable value of the date before the specific date. SEIR variable values can be predicted.

For example, to predict the SEIR predictor variable value on the 21st, the processor uses the SEIR variable value on the 20th, the SEIR variable value on the 19th, ..., the SEIR variable value on the 2nd, the SEIR variable value on the 1st, 19 The increase rate of SEIR variable values between the SEIR variable value of day 1 and the SEIR variable value of day 20, ..., the increase rate of SEIR variable values between the SEIR variable value of day 1 and the SEIR variable value of day 2, and the increase rate of N days. You can calculate the average, perform statistical analysis based on this, and calculate the N days that best predict the SEIR variable value for a specific date through statistical analysis.

At this time, the calculated N days may vary depending on the type of infectious disease, transmission period, transmission area, and transmission characteristics. The processor can use the calculated N days as a fixed value when doing mathematical modeling (SEIR) later. For example, the processor can use the calculated N days as a fixed value when doing predictive modeling for the infectious disease in the future.

For example, if the processor determines through statistical analysis that the SEIR variable value on a specific date is best predicted when N is 30 days in the case of COVID-19, in the case of COVID-19 prediction modeling, 30 days is used as the COVID-19 prediction model. Can be applied to all.

Referring to FIGS. 9 to 11, when predicting the SEIR variable value of a specific date in the future using SEIR, which is mathematical modeling, the processor sequentially predicts the SEIR variable value for each future date, one day at a time, based on the current time, and predicts the SEIR variable value of the specific date. Variable values can be predicted.

If N is set to 30 days, the method for calculating the SEIR predictor value for a specific date in the future is as follows.

For example, when the processor predicts the SEIR variable value for 4 days (+4 days) in the future based on the current date (day 0) using SEIR, which is mathematical modeling, 1 day later ( You can sequentially predict the SEIR variable values after (+1 day), 2 days later (+2 days), and 3 days later (+3 days), and then predict the SEIR variable values 4 days later (+4 days) based on this. there is.

In other words, the processor can use mathematical modeling, SEIR, to calculate the SEIR variable value 2 days later by multiplying the SEIR variable value predicted 1 day from the current date by the average increase rate of the SEIR variable value for the previous 30 days. In this way, the processor can calculate the SEIR variable value 3 days later by multiplying the SEIR variable value predicted 2 days from the current date by the average increase rate of the SEIR variable value of the previous 30 days. The processor of the present specification can predict the SEIR variable value on a specific date by sequentially predicting the SEIR variable value one day for future dates based on the current time.

For example, when building a SEIR model, which is predictive modeling for predicting result values after a specific date (e.g. M day), the processor uses the values of other explanatory variables (a1) other than SEIR before a specific date, The SEIR variable value (a2) and the result value predicted on a date before a specific date (a3) are used as explanatory variables (a), and the difference between the predicted result value after a specific date and the actual result value is used as the result variable (b). Therefore, supervised learning can be performed through artificial intelligence to build an infectious disease prediction model (110) after a specific date.

Referring to Figures 12 and 13, when the processor predicts an infectious disease on the 21st, 7 days after the 14th, other explanatory variables other than SEIR 7 days (14th) before the 21st, the date to be predicted, value (c1), the SEIR variable value predicted on the 21st (c2), and the result predicted before the 21st (for example, if the infectious disease was predicted for the 15th, 16th, and 19th, the The result value (c3)) is used as the explanatory variable (c), and the difference between the predicted result value on the 21st and the actual result value is used as the result variable (d), and supervised learning is performed through artificial intelligence to obtain the result on the 21st day (14 days). An infectious disease prediction model (110) can be built for (7 days after).

Referring to Figures 14 and 15, when predicting an infectious disease for M days in the future based on the current time, the processor predicts the value of other explanatory variables (e1) other than SEIR before a specific date and the specific date (M day) to be predicted. The predicted result on day M is obtained by applying the previously developed infectious disease prediction model after M days using the SEIR variable value (e2) and the predicted result value (e3) on the day before a specific date (M day) as the explanatory variable (e). The value (f) can be calculated.

In other words, if you are trying to make a prediction for M days in the future based on the current point in time, the infection prediction system can apply “M day future predictive modeling,” which was developed to predict the result after M days under the control of the processor. You can. For example, under the control of the processor, the infection prediction system applies the model developed as a 1-day prediction model when trying to predict after 1 day, and when trying to predict after 2 days, it applies the model developed as a 2-day prediction model. can be applied.

In addition, as shown in Figure 16, the infection prediction system uses a number of artificial intelligence prediction models (e.g. Random forest, Extremely randomized trees, GBM, XGBoost, GLM, You can calculate multiple prediction results using (Deep learning, Stacked ensemble, etc.) and the maximum, minimum, average, median, etc. of the results, and explain them by date using a graph or table.

As described above, the infection prediction system of the present specification can provide a service utilizing the infectious disease outbreak prediction results by accurately calculating the prediction result value for the specific date to be predicted under the control of the processor. For example, the infection prediction system of this specification utilizes the results of infectious disease outbreak predictions to provide basic information for customized quarantine by region through regional infectious disease predictions, a service that allows medical resources to be deployed in advance, and a service that allows for the pre-positioning of medical resources, You can connect with services that provide proactive support and response to damage.

The embodiments of the present invention described above may be embodied in other specific forms without departing from the essential characteristics of the present invention. Accordingly, the above detailed description should not be construed as restrictive in all respects and should be considered illustrative. The scope of the present invention should be determined by reasonable interpretation of the appended claims, and all changes within the equivalent scope of the present invention are included in the scope of the present invention. In addition, claims that do not have an explicit reference relationship in the patent claims can be combined to form an embodiment or included as a new claim through amendment after filing.

Claims (10)

In a method of predicting an infectious disease using an infectious disease system,
Creating a plurality of grids of a certain size on a digital map;
Based on the generated grid, a first input variable related to the number of floating population collected through a wireless communication network within the grid, a second input variable showing the number of floating population related to public transportation collected through a traffic information network, and a public information network. Calculating a third input variable related to public information collected through;
Calculating a plurality of infectious disease variables provided from a disease surveillance information module;
Extracting feature values from the first input variable, the second input variable, the third input variable, and the plurality of infectious disease variables, and inputting the extracted feature values into a deep learning model; and
Predicting infectious disease information about infectious diseases on the digital map based on the output of the deep learning model;
Including,
The first input variable calculates the floating population of wireless communication subscribers by time zone using the wireless communication network in the grid, selects a hotspot with the largest floating population among the grid, and calculates the daily floating population based on the hotspot. Calculate,
The second input variable is obtained by adding the total number of daily passengers boarding and the total number of daily disembarking passengers by administrative district for public transportation such as subways and buses, or, in the case of flights, the number of flights arriving in and departing from Korea by country. A method of categorizing each country based on the number of daily flights and arrivals to Korea. According to paragraph 1,
The deep learning model is,
A method of supervised learning based on the extracted feature values, through which an infectious disease prediction model is built and periodically updated. According to paragraph 2,
The step of predicting the infectious disease information is,
N days are calculated by predicting variable values output through the deep learning model,
A method of predicting the value of a variable on a specific date based on the calculated N days. According to paragraph 3,
The deep learning model is,
Includes mathematical modeling SI, SIS, SIR, SIRS, SEIR, SEIRS, etc.
The step of predicting the infectious disease information is,
When predicting the SEIR variable value of a specific date through the SEIR, the average increase rate of each SEIR variable value on the previous N days prior to the specific date and the date following the N days prior to the specific date is calculated as the average increase rate of the value of each SEIR variable on the date prior to the specific date. A method of predicting the SEIR variable value of the specific date by multiplying the SEIR variable value of . According to paragraph 2,
The deep learning model is,
When constructing the infectious disease prediction model, a method of setting a certain lag time between the risk of infectious disease transmission and its increase/decrease rate and the occurrence of infectious disease, death, and its increase/decrease rate. In a system for predicting infectious diseases,
A communication module that transmits and receives data based on at least one of a wireless communication network, a traffic information network, and a public information network;
A disease surveillance information module that stores infectious disease information; and
A plurality of grids of a certain size are created on a digital map, and based on the generated grids, a first input variable related to the number of floating population collected through the wireless communication network within the grid, and a first input variable collected through the traffic information network Calculating a second input variable showing the number of floating population related to public transportation and a third input variable related to public information collected through the public information network,
Calculate a plurality of infectious disease variables based on the infectious disease information provided from the disease surveillance information module,
Extracting feature values from the first input variable, the second input variable, the third input variable, and the plurality of infectious disease variables, and inputting the extracted feature values into a deep learning model,
A processor that predicts infectious disease information about infectious diseases on the digital map based on the output of the deep learning model
Including,
The first input variable calculates the floating population of wireless communication subscribers by time zone using the wireless communication network in the grid, selects a hotspot with the largest floating population among the grid, and calculates the daily floating population based on the hotspot. Calculated and saved,
The second input variable is obtained by adding the total number of daily passengers boarding and the total number of daily disembarking passengers by administrative district for public transportation such as subways and buses, or, in the case of flights, the number of flights arriving in and departing from Korea by country. A system that classifies and predicts infectious diseases based on the daily number of flights and arrivals to Korea for each country. According to clause 6,
The deep learning model is,
A system for predicting infectious diseases that is supervised based on the extracted feature values, builds an infectious disease prediction model through this, and updates periodically. In clause 7,
The processor,
N days are calculated by predicting variable values output through the deep learning model,
A system for predicting infectious diseases that predicts variable values for a specific date based on the calculated N days. According to clause 8,
The deep learning model is,
Includes mathematical modeling SI, SIS, SIR, SIRS, SEIR, SEIRS, etc.
The processor,
When predicting the SEIR variable value of a specific date through the SEIR, the average increase rate of each SEIR variable value on the previous N days prior to the specific date and the date following the N days prior to the specific date is calculated as the average increase rate of the value of each SEIR variable on the date prior to the specific date. A system for predicting infectious diseases that predicts the SEIR variable value of the specific date by multiplying the SEIR variable value of . In clause 7,
The deep learning model is,
A system for predicting infectious diseases that sets a certain lag time between the risk of infectious disease transmission and its increase/decrease rate and the occurrence of infectious disease, death, and its increase/decrease rate when constructing the infectious disease prediction model.