KR20230144301A - Encironmental health monitoring system and the method - Google Patents

Abstract

The present invention relates to an environmental health monitoring system and method, which is linked with a first IoT device including a plurality of environmental sensor modules and a second IoT device including a plurality of health sensor modules to collect environmental hazards and health influencing factors data. A platform that collects and monitors, a management unit that collects environmental harmful factors and health-impacting factors from the first IoT device and the second IoT device, respectively, and analyzes a plurality of data including the environmental harmful factors and the health-impacting factors. A data collection unit that generates meta information for each of the plurality of data and classifies the plurality of data based on the meta information, and creates a data set by linking the plurality of data based on at least one of the meta information. A data set generation unit that analyzes the correlation between the environmental harmful factors, the correlation between the health influencing factors, and the environmental harmful factors and the health influencing factors based on the data set, and a correlation analysis unit that generates correlation analysis data. and a platform including a monitoring unit that provides spatiotemporal information visualizing the plurality of data and the correlation analysis data to a user terminal.

Description

Environmental health monitoring system and method {ENCIRONMENTAL HEALTH MONITORING SYSTEM AND THE METHOD}

The present invention relates to an environmental health monitoring system and method, which collects raw data on the environment and health, preprocesses the data, analyzes the spatiotemporal correlation between environmental hazards and health-impacting factors, and monitors and predicts areas vulnerable to environmental health. It relates to environmental health monitoring systems and methods.

The development of the Internet of Things (IoT) and platform services are being applied to a wide range of fields and are the core of global growth. In addition, interest in and anxiety about health has been increasing since the recent pandemic, and there is a continuing demand for a causal relationship between the environment and health.

Environment-related ministries and organizations, such as the Ministry of Environment and the National Institute of Environmental Research, hold and disclose environmental and health data. However, it is difficult to check the reliability and integrity of data due to the sporadic collection of environmental harmful factors, the connection between environmental harmful factors and health data is insufficient, and the operation of the system centered on data providers makes it difficult to provide valid data to consumers. There are limits.

In addition, consumers secure users' health data through wearable devices, but correlations and health prediction data related to health data, that is, environmental harmful factors that cause changes in health data, are not provided.

Accordingly, as a big data solution that combines environmental harmful factors and health influencing factors, it collects reliable environmental harmful factor data and health influencing factor data from IoT sensors and provides correlation between environmental harmful factors and health influencing factors and health prediction data for each user. This is being demanded.

The problem that the present invention aims to solve is to increase the reliability of sensing data measured from IoT devices, and to analyze and/or predict the correlation between environmental harmful factors and health influencing factors according to environmental harmful factor data, health influencing factor data, and regional characteristics. The goal is to provide an environmental health monitoring system and method that provides time and space.

In addition, the goal is to provide an environmental health monitoring system and method that provides customized health information for each user, data on health abnormalities, and correlations between environmental harmful factors and abnormal health symptoms.

The technical problems of the present invention are not limited to the technical problems mentioned above, and other technical problems not mentioned can be clearly understood by those skilled in the art from the description below.

The platform according to one embodiment is a platform that collects and monitors environmental hazards and health impact factor data by linking with a first IoT device including a plurality of environmental sensor modules and a second IoT device including a plurality of health sensor modules. As a management unit that collects environmental harmful factors and health-impacting factors from the first IoT device and the second IoT device, respectively, analyzes a plurality of data including the environmental harmful factors and health-impacting factors, and analyzes each of the plurality of data A data collection unit that generates meta information and classifies the plurality of data based on the meta information, a data set generation unit that links the plurality of data based on at least one of the meta information to generate a data set, Based on the data set, a correlation analysis unit that analyzes the correlation between the environmental harmful factors, the correlation between the health influencing factors, and the correlation between the environmental harmful factors and the health influencing factors, and generates correlation analysis data, and the plurality of data, and It includes a monitoring unit that provides spatial and temporal information visualizing the correlation analysis data to the user terminal.

When generating the correlation analysis data, the correlation analysis unit may perform machine learning for temporal and spatial correlation analysis.

When performing machine learning to analyze the correlation between the environmental harmful factors, the correlation analysis unit sets at least one of temporal information, geographic information, meteorological information, and environmental harmful factors as an input variable, and sets at least one of the environmental harmful factors. It can be set as an output variable.

When the correlation analysis unit performs machine learning to analyze the correlation between the health influencing factors and the correlation between the health influencing factors and the environmental harmful factors, at least one of the environmental harmful factors, humanities and social information, geographical information, and health impact At least one of the factors can be set as an input variable, and at least one of the health impact factors can be set as an output variable.

The correlation analysis unit may generate correlation prediction data between the environmental harmful factors and the health influencing factors.

The management unit collects data from an external linkage system that provides any one of environmental health-related data, living environment hazard data, and environmental pollution area information, and the plurality of data includes data collected from the external linkage system, The monitoring unit may analyze data collected from the external linkage system to derive characteristics of each of a plurality of regions and classify areas vulnerable to environmental pollution.

The correlation analysis unit may assign weights when analyzing the correlation between the environmental harmful factors and the health influencing factors for the area vulnerable to environmental pollution.

The monitoring unit may provide user-specific health information, plurality of abnormal health sign information, and relationship data between the environmental harmful factors and the plurality of health abnormality sign information.

The monitoring unit converts each of the plurality of health abnormality sign information generated in a period set by the user into a graph to provide visualized data, and provides the data corresponding to the period set by the user for each of the plurality of health abnormality signs. It can provide average values and evaluation information of environmental harmful factors.

The environmental health monitoring method according to an embodiment is one in which the platform collects environmental harmful factors and health influencing factors from a first IoT device including a plurality of environmental sensor modules and a second IoT device including a plurality of health sensor modules. Step, the platform analyzes a plurality of data including the environmental harmful factors and the health influencing factors to generate meta information for each of the plurality of data, the platform classifies the plurality of data based on the meta information A step of generating a data set by the platform linking the plurality of data based on at least one of the meta information, the platform generating a data set based on the data set, the correlation between the environmental harmful factors, the health influencing factors Analyzing the correlation between the environmental hazards and the health-impacting factors, generating correlation analysis data, and providing the platform with spatiotemporal information visualizing the plurality of data and the correlation analysis data to the user terminal. Includes.

The step of the platform generating the correlation analysis data may include the step of the platform performing machine learning for temporal and spatial correlation analysis.

When the platform performs machine learning to analyze the correlation between the environmental harmful factors, at least one of temporal information, geographic information, weather information, and the environmental harmful factors is set as an input variable, and at least one of the environmental harmful factors is set as an input variable. can be set as an output variable.

When the platform performs machine learning to analyze the correlation between the health influencing factors and the correlation between the health influencing factors and the environmental harmful factors, at least one of the environmental harmful factors, humanities and social information, geographical information, and the health impact At least one of the factors may be set as an input variable, and at least one of the health impact factors may be set as an output variable.

The platform may further include generating correlation prediction data between the environmental harmful factors and the health influencing factors.

The platform further includes a step of collecting data from an external linkage system that provides any one of environmental health-related data, living environment hazard data, and environmental pollution area information, and the plurality of data is collected from the external linkage system. Including data, the platform can analyze data collected from the external linkage system to derive characteristics of each of a plurality of regions and classify areas vulnerable to environmental pollution.

The step of generating the correlation analysis data by the platform further includes the step of the platform analyzing the correlation between the environmental harmful factors and the health-impacting factors for the area vulnerable to environmental pollution, wherein the platform Weights can be assigned when analyzing the correlation between the environmental hazards and health-impacting factors for vulnerable areas.

The platform may further include providing user-specific health information, information on a plurality of abnormal health signs, and relationship data between the environmental harmful factors and the plurality of information on abnormal health signs.

The step of the platform providing relationship data between the environmental harmful factors and the plurality of health abnormality symptom information includes the platform converting each of the plurality of health abnormality symptom information generated during the period set by the user into a graph and visualizing it. provided data, and for each of the plurality of health abnormalities, the average value and evaluation information of the environmental harmful factors corresponding to the period set by the user may be provided.

Specific details of other embodiments are included in the detailed description and drawings.

According to the environmental health monitoring system and method according to embodiments of the present invention, the correlation between environmental harmful factors and health influencing factors is analyzed to provide effective linkage data between environmental harmful factors and health influencing factors, correlation analysis data, and/ Alternatively, it provides correlation prediction data and can be advantageous for improving the accuracy of correlation analysis through mutual iterative machine learning between environmental hazards and health-impacting factors.

In addition, regions of interest vulnerable to environmental pollution are set according to the characteristics of each region, and weights are assigned when analyzing the correlation between environmental hazards and health-impacting factors in the region of interest, thereby predicting the situation in areas vulnerable to environmental health. It can be advantageous in managing the impact of exposure to environmental hazards in areas vulnerable to environmental health.

Furthermore, the environmental health monitoring system and method according to embodiments provide user-specific health information, health abnormality sign information, and relationship data between environmental harmful factors and health abnormality symptom information, so that each user can determine the factors of his or her health abnormality. It can be advantageous to quickly recognize and improve the activity environment accordingly to eliminate health abnormalities related to environmental harmful factors.

Effects according to embodiments of the present invention are not limited to the contents exemplified above, and further various effects are included in the present specification.

1 is a diagram schematically showing an environmental health monitoring system according to an embodiment of the present invention.
Figure 2 is a block diagram schematically showing the configuration of a platform according to an embodiment of the present invention.
Figure 3 is a block diagram schematically showing the configuration of a first IoT device according to an embodiment of the present invention.
Figure 4 is a block diagram schematically showing the configuration of a second IoT device according to an embodiment of the present invention.
Figure 5 is a flowchart illustrating a basic environmental health monitoring method according to an embodiment of the present invention.
6 to 14 are diagrams showing screens output by a monitoring unit through a user interface according to another embodiment of the present invention.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the attached drawings to clarify the technical idea of the present invention. In describing the present invention, if it is determined that a detailed description of related known functions or components may unnecessarily obscure the gist of the present invention, the detailed description will be omitted. Components having substantially the same functional configuration among the drawings are given the same reference numbers and symbols as much as possible, even if they are shown in different drawings. For convenience of explanation, if necessary, the device and method should be described together.

Although first, second, etc. are used to describe various components, these components are of course not limited by these terms. These terms are merely used to distinguish one component from another. Therefore, it goes without saying that the first component mentioned below may also be a second component within the technical spirit of the present invention.

1 is a diagram schematically showing an environmental health monitoring system according to an embodiment of the present invention.

Referring to FIG. 1, the environmental health monitoring system according to one embodiment includes a platform 100, a first IoT device 200, a second IoT device 300, an external linkage system 400, and a network 500. can do.

The platform 100 collects environmental hazard data (hereinafter referred to as environmental hazards) and health impact factor data (hereinafter referred to as health) from the first IoT device 200, the second IoT device 300, and the external linkage system 400. (referred to as influencing factors) can be collected. The platform 100 collects a plurality of collected data and performs a data pre-processing process including data cleaning, data conversion, and distributed processing for raw data to convert environmental hazards and health-impacting factors into structured core information. can be provided. In addition, the platform 100 analyzes a plurality of data in time and space to generate time series data, real-time monitoring data, and correlation analysis unit 150 for each environmental hazard and health impact factor as a spatial information system (GIS). Correlation analysis data and correlation prediction data can be provided visually.

Hereinafter, for convenience of explanation, the description will focus on the platform 100 collecting data from one first IoT device 200, one second IoT device 300, and one external linkage system 400. , but is not limited to this. For example, the platform 100 may collect environmental harmful factors and health-impacting factors from a plurality of first IoT devices 200, a plurality of second IoT devices 300, and a plurality of external linkage systems 400. . A detailed description of the platform 100 will be described later with reference to FIG. 2.

The first IoT device 200 and the second IoT device 300 may be Internet of Things (IoT) devices that include a plurality of environmental sensor modules and a plurality of health sensor modules, respectively. The first IoT device 200 and the second IoT device 300 may transmit data collected through a plurality of sensor modules to the platform 100. For example, the first IoT device 200 and the second IoT device 300 may each transmit data to the platform 100 through a gateway (not shown), but are not limited thereto.

The first IoT device 200 may include a plurality of environmental sensor modules. For example, the first IoT device 200 includes a temperature sensor module, a humidity sensor module, a fine dust/ultrafine dust sensor module, an acoustic sensor module, a Total Volatile Organic Compounds (TVOC) sensor module, and a carbon dioxide (CO ₂ ) sensor module. , carbon monoxide (CO) sensor module, hydrogen sulfide (H ₂ S) sensor module, formaldehyde (HCHO) sensor module, nitrogen oxides (NOx) sensor module, ammonia (NH ₃ ) sensor module, odor sensor module, and/or radon. May include a sensor module.

The second IoT device 300 may include a plurality of health sensor modules. For example, the second IoT device 300 may include a lung sound sensor module, a blood oxygen saturation sensor module, a respiration measurement sensor module, an electromyography sensor module, a heart rate measurement sensor module, an electrocardiogram sensor module, a pulse sensor module, a temperature sensor module, and/ Alternatively, it may include an electroencephalography (EEG) measurement sensor module.

For example, the second IoT device 300 is a smart device worn or attached by a user who agrees to provide information about sensing data, and is an accessory wearable device such as a smart watch, smart band, smart glasses, HoloLens, bio device, etc. It may include clothing-type wearable devices composed of fabric circuit boards such as shirts, body-worn or implantable devices such as electronic skin, smart contact lenses, and bio-stamps.

The first IoT device 200 and/or the second IoT device 300 may be equipped with a function to process missing values of sensor data in order to increase the reliability of various collected sensor data. For example, the first IoT device 200 and/or the second IoT device 300 has the function of processing missing values caused by errors in various sensor devices, such as the measurement range of each sensor module, sudden changes in data, etc. can be provided. Additionally, in some embodiments, the first IoT device 200 and/or the second IoT device 300 may be linked with the platform 100 and have a function for processing missing values due to communication instability.

The external linkage system 400 may be a system that provides at least one of data such as various environmental health-related data, data on hazardous factors in the living environment, information on environmental pollution areas, and regional weather information.

The network 500 is a communication network in which the first IoT device 200, the second IoT device 300, the external linkage system 400, and the platform 100 communicate with each other, and can be configured regardless of communication mode. there is. For example, it can be composed of various communication networks such as Personal Area Network (PAN), Local Area Network (LAN), Metropolitan Area Network (MAN), and Wide Area Network (WAN). However, it is not limited to this. Although not specifically shown, the first IoT device 200 and the second IoT device 300 may each communicate with the platform 100 through the network 500 via a separate gateway (not shown).

Figure 2 is a block diagram schematically showing the configuration of a platform according to an embodiment of the present invention.

Referring to Figure 2, the platform 100 according to one embodiment includes a management unit 110, a data collection unit 120, a data pre-processing unit 130, a data set creation unit 140, a correlation analysis unit 150, It may include a learning data generation unit 160, a monitoring unit 170, and a database unit 180.

The management unit 110 may transmit and receive data with the first IoT device 200 and the second IoT device 300 according to a communication method suitable for each of the first IoT device 200 and the second IoT device 300. For example, the management unit 110 supports various wireless communications such as fifth generation technology standard (5G), Long Term Evolution (LTE), Wireless Fidelity (WiFi), Bluetooth, LoRa, NB-IoT, and Zigbee. Data may be transmitted and received between the platform 100 and the first IoT device 200 and the second IoT device 300 through at least one of the methods.

Additionally, the management unit 110 may communicate data with the external linking system 400 by wire or wirelessly and collect data from the external linking system 400. For example, the management unit 110 communicates data with the external linkage system 400 through at least one of the various wireless communication methods or a wired Internet communication method supporting TCP/IP (Transmission Control Protocol/Internet Protocol), etc. can do.

The management unit 110 may collect data from the external linkage system 400 by scraping data according to preset data classification. Here, the preset data classification is a data classification set by an administrator (not shown) who manages the platform 100, and includes data such as various environmental health-related data, data on hazardous factors in the living environment, information on environmental pollution areas, and regional weather information. It may be meta information, but is not limited thereto.

The preset data classification may be updated based on meta information generated through the data collection unit 120. For example, the preset data classification may be geographical information and/or temporal information among meta information about at least one piece of data generated by the data collection unit 120. Accordingly, the management unit 110 may collect data from the external linkage system 400 by scraping data according to geographical information and/or temporal information as meta information for the at least one piece of data.

The management unit 110 may transmit data received from the first IoT device 200, the second IoT device 300 or collected from the external linkage system 400 to the data collection unit 120.

The management unit 110 may provide at least one of a website, an application, or an application program for providing data stored in the database unit 180 and monitoring data of the monitoring unit 170 to a user terminal (not shown).

Here, the user terminal may be a device that can be run by accessing a website provided by the management unit 110 through a wired/wireless communication network such as the Internet and/or an intranet, or by installing an application or application program. For example, the user terminal may be a mobile terminal such as a laptop, a handheld device, a smartphone, a tablet PC, a desktop computer, or any device that uses or is directly or indirectly connected to such a device.

The data collection unit 120 may analyze data transmitted from the management unit 110 and generate meta information of the data. The data collection unit 120 may generate meta information including data type, geographic information, temporal information including creation date and time, and personal information for each of the plurality of data.

For example, the data collection unit 120 first classifies (or, primary classification information) can be created. The data collection unit 120 may secondarily classify the primary classified data into detailed types (or generate secondary classification information). As a specific example, data classified as environmental hazards may be secondary classified into fine dust/ultrafine dust data, temperature data, humidity data, acoustic data, carbon dioxide data, carbon monoxide data, TVOC data, hydrogen sulfide data, etc.

In this specification, the secondary classification method of the data collection unit 120 uses data primarily classified as environmental hazards as a specific example, but also includes health impact factors, environmental health data, life-welcome hazard factor data, and environmental pollution area information. Data can also be classified into detailed types.

The geographical information and creation date and time refer to the date and time at which the corresponding data was generated, respectively, if the data is sensing data about environmental hazards or health-impacting factors received from the first IoT device 200 or the second IoT device 300. It may mean (or sensed) geographical information and creation date and time. In addition, if the data is data collected from the external linkage system 400, the geographical information and creation date and time are not only the geographical information and creation date and time at which the data was generated, but also the geographical information and May include timely information.

If the data is received from the first IoT device 200 or the second IoT device 300, the personal information is user information stored in conjunction with each device and includes the user's age, gender, occupation, disease, etc. It can be included. Additionally, if the data is collected from the external linkage system 400, the personal information may be user information that created the data or may include personal information included in the data.

The data collection unit 120 may classify a plurality of collected data. For example, the data collection unit 120 classifies each of the plurality of data into meta information, that is, by data type (e.g., by environmental hazards, by health-impacting factors), or by geographic information (e.g., by region, by facility, by residential environment). It can be classified according to time information (e.g., by month, day of the week, time of day), or personal information (by age, gender).

Additionally, the data collection unit 120 may use a classification method to classify a plurality of data based on correlation analysis data generated by the correlation analysis unit 150, which will be described later. Accordingly, the data collection unit 120 may classify a plurality of data by reflecting the continuous correlation analysis data, and the effectiveness and utility of the plurality of data classification methods performed by the data collection unit 120 may be improved.

If the data collected from the external linkage system 400 is unstructured data, the data collection unit 120 may formalize the unstructured data. The data collection unit 120 can standardize unstructured data through data mining by type. For example, the data collection unit 120 may data mine text or images included in a plurality of unstructured data and manage them by labeling the data according to the data type.

The data collection unit 120 may structure a plurality of data classified into at least one meta information based on the meta information and store them in the database unit 180. Additionally, the data collection unit 120 may transmit the plurality of data to the data pre-processing unit 130. In this case, the data collection unit 120 may transmit classification information for each of the plurality of data to the data pre-processing unit 130.

The data pre-processing unit 130 can filter and clean a plurality of data. The data preprocessor 130 can remove redundancy and errors in a plurality of data, and process missing values where there is no data or is incomplete.

Hereinafter, a method by which the data preprocessor 130 processes missing values of data received from the first IoT device 200 will be described as a specific example, but the data preprocessor 130 processes missing values of data received from the first IoT device 200. The method of processing missing values for sensing data as received data or data received from the external linking system 400 may be applied substantially in the same way.

The data transmitted from the first IoT device 200 to the platform 100 is a periodic and aperiodic communication error due to dynamic changes in the communication method, and the sensor module's error that can occur when the measured value of the sensing data changes suddenly or is out of the measurement range. It can contain four or more types of missing values, including errors. Additionally, the first IoT device 200 includes a plurality of environmental sensor modules, and data collected from the first IoT device 200 may be collected in a time-series, continuous manner.

When processing missing values of data transmitted from the first IoT device 200, the data preprocessor 130 may use an imputation method to replace missing values with estimated values. In addition, the data preprocessor 130 performs univariate time series imputation using time dependence as a method of imputing missing values in consideration of the above-described first IoT device 200 and the specification of data collected from the first IoT device 200. Multivariate imputation methods that utilize correlations between methods and variables can be used. That is, the data preprocessor 130 processes missing values by considering the time-series autocorrelation of each of the plurality of environmental sensor modules and the correlation between data (environmental harmful factors) collected from at least two of the plurality of environmental sensor modules. can do.

As a specific example, the data preprocessor 130 linearly estimates the missing value according to the linear distance using the values of both end points to impute missing values according to the univariate imputation method (or univariate time series imputation method). Linear interpolation (LI), spline interpolation (SI), which divides into subintervals to estimate missing values using a low-order polynomial, and last, which estimates missing values using data collected just before the missing values occur. Last observation carried forward (LOCF), moving average (MA), which estimates missing values as the average of a window of a certain size centered on missing values, and Kalman smoothing The Kalman imputation method to estimate missing values can be applied.

In addition, the data preprocessor 130 uses the K-nearest neighbors imputation method (k-Nearest neighbors, KNN) to replace missing values using the K value closest to the missing value in order to replace missing values according to the multivariate imputation method. ), Multiple linear regression, which fits a multiple linear regression model and uses it to impute missing values, and Landon Forest regression, which uses the average prediction of multiple decision trees to impute missing values. forest regression), support vector regression, which uses a support vector machine (SVM) to impute missing values, and miss, which is used to impute missing values using a random forest in an iterative manner. Miss forest can be applied.

In this case, the data preprocessor 130 uses the Pearson correlation coefficient and/or Spearman correlation coefficient to identify the dependency, that is, the correlation, between a plurality of variables (or data). Thus, the correlation between multiple variables can be analyzed. Additionally, the data pre-processing unit 130 may use the data analyzed by the correlation analysis unit 150 to confirm and analyze dependencies between a plurality of variables and generate dependency data between a plurality of variables.

For example, the data preprocessor 130 analyzes a plurality of data collected from the first IoT device 200 and calculates carbon monoxide (CO) and temperature, nitrogen oxides (NOx) and temperature, carbon dioxide (CO2) and TVOC (Total). It can be analyzed that there is a high correlation between Volatile Organic Compounds) and nitrogen oxides (NOx) and carbon monoxide (CO).

The data preprocessing unit 130 can use statistical techniques and machine learning techniques as an ensemble learning method to simultaneously consider the univariate time series imputation method and the multivariate imputation method.

The data pre-processing unit 130 may use the Weighted Average Method, which derives the final prediction result by setting a weighted average according to the weight using a statistical technique. Here, a high weight is given to methods with good performance, and the weight can be set in inverse proportion to the value measured for each substitution method using an evaluation method that will be described later. The equation for deriving the final prediction result according to the weighted average algorithm is as shown in Equation 1.

here, is the final result prediction vector, class is a result vector predicted through one of the univariate imputation methods and one of the multivariate imputation methods, respectively, e1 may be a weight for the univariate imputation method, and e2 may be a weight for the multivariate imputation method.

Additionally, the data pre-processing unit 130 may use a stacking method that performs prediction again based on prediction data of individual algorithms using a machine learning technique. The data preprocessor 130 can collect missing values by one of the univariate imputation methods and missing values by one of the multivariate imputation methods, and derive the final prediction result through meta learning. As a non-limiting example, the data preprocessor 130 may use a linear regression model to derive the final prediction result.

Thereafter, the data preprocessor 130 may evaluate the final prediction result by the weighted average method and the final prediction result by the stacking method. The data preprocessor 130 calculates the final prediction result by the weighted average method and the final prediction result by the stacking method using the Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) methods, respectively. It can be evaluated using .

The data preprocessor 130 may use pre-stored data and pre-stored data patterns as reference data to calculate the mean absolute error and root mean square error. Here, the pre-stored data and pre-stored data patterns may be data received from the external linkage system 400 or data injected by an administrator who manages the platform 100.

Additionally, the pre-stored data and pre-stored data patterns may include data and data patterns processed as missing values by the data pre-processing unit 130. That is, the data pre-processing unit 130 learns data patterns for processing missing values by repeating the missing value processing process, and uses the learned data as reference data for processing missing values, thereby improving the reliability of the data. .

The data preprocessor 130 applies the prediction result with the best evaluation result among the 25 final prediction results to which one of the five univariate imputation methods and one of the five multivariate imputation methods were applied, respectively, to remove missing values. It can be handled. Here, the excellent prediction result value is a value in which the values evaluated according to each of the mean absolute error and the root mean square error are small, and/or the difference between the values evaluated according to each of the mean absolute error and the root mean square error is small. It can mean value.

In this way, the data preprocessor 130 can be advantageous in increasing the reliability of data by replacing missing values using an ensemble method that considers time dependence and correlation between data collected through different sensor modules.

As described above, the data pre-processing unit 130 determines the types of missing values such as periodic and non-periodic communication errors, errors in the sensor module that may occur when the measured value of the sensing data changes suddenly or is outside the measurement range, etc. Divided into four types, and learn and apply appropriate methods corresponding to each type in deriving the final prediction result to replace missing values by ensemble method, increasing the performance of the algorithm, further improving processing speed and data reliability. The load can be reduced while increasing.

Additionally, the data pre-processing unit 130 may provide a missing value processing method suitable for each of the first IoT device 200 and/or the second IoT device 300 through the management unit 110. In this case, the data preprocessor 130 may provide a method for processing missing values for each of a plurality of variables. The data preprocessor 130 may provide dependency data between a plurality of sensed data (or variables) for each of the first IoT device 200 and the second IoT device 300.

The data pre-processing unit 130 can convert a plurality of data into a form that is easy to analyze and store the data in the database unit 180. For example, the data pre-processing unit 130 may convert the plurality of data transmitted from the data collection unit 120 into a form that facilitates data analysis according to the classification information of each, but is not limited thereto.

Additionally, the data preprocessor 130 may process a plurality of data according to the purpose of data analysis. The data pre-processing unit 130 may process each data based on classification information of each of the plurality of data.

For example, the data pre-processing unit 130 provides information on environmental hazards, health impact factors, environmental health data, living environment hazard factor data, environmental pollution local information data, etc. according to the purpose of analyzing the correlation between environmental hazards and health impact factors. Multiple data classified by data type meta information can be processed to expand the space and time range. The data preprocessing unit 130 expands temporal units (e.g., seconds, minutes, hours, days, months, years) and/or geographical units according to the temporal and geographical scope of the correlation analysis between environmental hazards and health influencing factors. (e.g. location, town, town, town, district, etc.) can be expanded.

The data pre-processing unit 130 may transmit a plurality of data including converted and/or processed data and data classification information to the data set generating unit 140.

The data set generator 140 may generate a data set by linking a plurality of data received from the data preprocessor 130. For example, the data set generator 140 may link data by using at least one of the meta information (or classification information) for each of the plurality of data as an identifying variable or variable of interest. there is.

For example, the data set generator 140 generates geographical information (e.g., by region, facility, residential environment), temporal information (e.g., by month, day of the week, time zone), adjacent to or identical to at least two pieces of data, and at least one data set can be created by linking data based on at least one of personal information (age, gender).

The data set generator 140 may provide the generated data set to the correlation analysis unit 150 and the learning data generator 160.

The correlation analysis unit 150 may analyze the correlation between a plurality of data and generate correlation analysis data. The correlation analysis unit 150 may analyze the correlation between environmental hazards and health-impacting factors based on the data set received from the data set generator 140. In addition, the correlation analysis unit 150 variableizes a plurality of data to determine the correlation between a plurality of environmental harmful factors, the correlation between a plurality of health influencing factors, and the correlation between at least one environmental harmful factor and at least one health influencing factor. Correlation can be analyzed.

For example, the correlation analysis unit 150 may analyze the correlation between environmental hazards and health-impacting factors and/or the correlation between a plurality of data sets using the Pearson correlation coefficient and the Pierceman correlation coefficient.

As a specific example, the correlation analysis unit 150 uses a plurality of independent variables such as indoor area, air conditioning, indoor smoking, air freshener, type of building, ventilation cycle, and height (or floor) from the ground as indoor environmental factors with respect to environmental harmful factors. , and temperature, humidity, TVOC, carbon dioxide, carbon monoxide, ammonia, radon, nitrogen oxides, etc. can be set as multiple dependent variables.

In addition, the correlation analysis unit 150 sets gender, age, occupation, smoking status, drinking status, BMI, etc. as human factors regarding health influencing factors as a plurality of independent variables, and sets heart rate, breathing, lung sounds, and brain waves (EEG) as multiple independent variables. ), coral saturation, sleep patterns, etc. can be set as multiple dependent variables.

The correlation analysis unit 150 uses canonical correlation analysis (CCA) to determine the independent variables and dependent variables included in environmental hazards, the independent variables and dependent variables included in health impact factors, and the environmental hazards. You can analyze the correlation between data sets and data sets included in health impact factors.

In this way, the correlation analysis unit 150 determines linearity and monotonicity between the variables included in each of the environmental hazards and health-impacting factors, or between the variables included in the environmental hazards and the variables included in the health-impacting factors. By checking, high-quality correlation analysis data can be generated.

The correlation analysis unit 150 can perform machine learning to analyze temporal and spatial correlations between environmental harmful factors, health-impacting factors, and environmental harmful factors and health-impacting factors.

For example, the correlation analysis unit 150 uses at least one of temporal information (e.g., year, month, day, time, etc.), geographical information, weather information, and environmental harmful factors for machine learning to analyze the correlation between environmental harmful factors. Any one independent variable can be set as an input variable, and at least one dependent variable among environmental harmful factors can be set as an output variable. According to a specific example, the machine learning equation for analyzing the correlation between environmental harmful factors in the correlation analysis unit 150 may be as shown in Equation 2.

In Equation 2, the concentration of environmental harmful factors is a dependent variable of at least one of the environmental harmful factors, , , Each may be an independent variable of at least one of the environmental harmful factors.

In this way, in the case of Equation 2, the correlation analysis unit 150 analyzes the correlation between environmental harmful factors (for example, between independent variables and dependent variables included in environmental harmful factors) and simultaneously analyzes temporal trends, You can learn trends based on climate and geographical information such as altitude, land use classification, and surrounding pollution sources.

In addition, the correlation analysis unit 150 uses at least one of the environmental harmful factors, humanities and social information (e.g., population, education, etc.) for machine learning to analyze the correlation between health influencing factors and the correlation between health influencing factors and environmental harmful factors. , income, etc.), geographic information, and health influencing factors, at least one independent variable may be set as an input variable, and at least one dependent variable among the health influencing factors may be set as an output variable. According to a specific example, the machine learning equation for analyzing the correlation between health influencing factors and between health influencing factors and environmental harmful factors in the correlation analysis unit 150 may be as shown in Equation 3.

In Equation 3, the health influencing factor indicator is a dependent variable of at least one of the health influencing factors, and the environmental harmful factor concentration is a dependent variable of at least one of the environmental harmful factors, , , Each may be an independent variable of at least one of the health influencing factors.

In this way, in the case of Equation 3, the correlation analysis unit 150 determines the correlation between health influencing factors (for example, between independent variables and dependent variables included in health influencing factors) and between health influencing factors and environmental harmful factors. At the same time as analyzing, you can learn trends based on humanities and social information including population, education, and income, as well as trends based on geographic information such as altitude, land use classification, and surrounding pollution sources. Furthermore, by setting the concentration of environmental harmful factors as an input variable in Equation 3, the correlation analysis unit 150 will be advantageous in improving the accuracy of correlation analysis through mutual iterative machine learning between environmental pollutants and health-impacting factors. You can.

The correlation analysis unit 150 predicts future environmental harmful factors, health influencing factors, and the correlation between health influencing factors and environmental harmful factors, and generates environmental harmful factor trend prediction data, health influencing factor trend prediction data, and correlation prediction data. You can.

Since health influencing factors and environmental harmful factors each have the characteristic of judging and predicting trends based on changes in past data, the correlation analysis unit 150 uses machine learning to process data with an order that changes over time. Memory (Long Short-Term Memory, LSTM) and/or gated recurrent unit (GRU) can be used to predict environmental harmful factors, health influencing factors, and the correlation between health influencing factors and environmental harmful factors.

The correlation analysis unit 150 may transmit the correlation analysis data and/or correlation prediction data to the learning data generation unit 160, the monitoring unit 170, and the database unit 180. Additionally, the correlation analysis unit 150 may provide correlation analysis data to the data collection unit 120.

The learning data generator 160 may generate learning data to advance at least one artificial intelligence (AI) used for data collection, missing value processing, and correlation analysis. The learning data generator 160 may generate learning data based on data type, geographic information, temporal information, and personal information.

For example, the learning data generator 160 may generate learning data based on the data set provided from the data set generator 140. Additionally, by updating the learning data by reflecting the correlation analysis data and/or correlation prediction data provided from the correlation analysis unit 150, more reliable and effective learning data can be generated.

The monitoring unit 170 performs time-series monitoring for each of a plurality of environmental harmful factors and a plurality of health-impacting factors, and the time-series data, real-time monitoring data, and correlation generated by the correlation analysis unit 150 for each environmental harmful factor and health-impacting factor. Analysis data and correlation prediction data can be provided visually through the user interface (UI). The monitoring unit 170 may provide visualized data to a user terminal (not shown) through at least one of a website, an application, or an application.

For example, the monitoring unit 170 is a geographic information system (GIS) and can provide a plurality of data including a plurality of environmental hazards, a plurality of health-impacting factors, etc. as spatiotemporal information. In addition, the monitoring unit 170 may provide environmental harmful factor trend prediction data, health influencing factor trend prediction data, and correlation prediction data predicted by the correlation analysis unit 150 according to time and space.

In addition, the monitoring unit 170 analyzes a plurality of data to classify areas vulnerable to environmental pollution, major influencing factors for each area vulnerable to environmental pollution (e.g., environmental harmful factors that have the greatest impact on environmental pollution), and spatiotemporal information visualizing them. can be provided.

For example, the monitoring unit 170 may derive characteristics of each of a plurality of regions based on data collected from the external linkage system 400. Here, the characteristics of each of the plurality of regions may include the presence or absence of a coal-fired power plant, nuclear power plant, or internal combustion power plant located in the region, the presence or absence of an industrial complex, the presence or absence of an industrial or abandoned mine, traffic density, and whether it is an undeveloped area.

The monitoring unit 170 may set an area affected by environmental pollution as an area of interest according to the characteristics of each of the plurality of areas. Additionally, when multiple areas of interest are set, the monitoring unit 170 can calculate the ranking of the plurality of areas of interest. The monitoring unit 170 may assign search priority to the area of interest. In this case, the correlation analysis unit 150 may assign weights when analyzing the correlation between environmental hazards and health-impacting factors for the region of interest.

Accordingly, in providing the spatio-temporal information, the monitoring unit 170 can provide correlation analysis data and correlation prediction data between environmental hazards, health-impacting factors, environmental hazards and health-impacting factors according to regional characteristics. .

Additionally, the monitoring unit 170 may provide user-specific health information (or health influencing factors), health abnormality sign information, and relationship data between environmental harmful factors and health abnormality symptom information.

For example, the management unit 110 collects environmental harmful factors from the first IoT device 200 placed in the space where at least one user lives, and collects environmental harmful factors from the second IoT device worn by the at least one user ( Health information can be collected from 300).

In this case, the correlation analysis unit 150 analyzes and predicts the relationship between environmental harmful factors and health information and/or health abnormality symptom information regarding at least one user, and the monitoring unit 170 analyzes and predicts the relationship between the website and application. Alternatively, health information, relationship analysis data between environmental harmful factors and health information, and information on signs of health abnormalities may be provided to the at least one user through at least one of the application programs.

A method of providing relationship data between specific user-specific health information, environmental harmful factors, and health abnormality symptom information by the monitoring unit 170 will be described later with reference to FIGS. 6 to 14.

The database unit 180 may store data collected and generated by the platform 100.

The database unit 180 according to one embodiment includes an environmental hazard DB 181, a health impact factor DB 182, a data set DB 183, a correlation analysis DB 184, a learning data DB 185, and a monitoring DB. (186), user DB (187), and device DB (188). For example, the database unit 180 includes an environmental hazard DB 181, a health impact factor DB 182, a data set DB 183, a correlation analysis DB 184, a learning data DB 185, and a monitoring DB ( 186), the user DB 187, and the device DB 188 can be managed as a database cluster built on a plurality of servers.

The environmental hazard DB 181 may store data received from the first IoT device 200 through the management unit 110. When storing environmental hazards data, the environmental hazards DB 181 may also store meta information generated by the data collection unit 120. In some embodiments, the data collection unit 120 may classify and store a plurality of environmental harmful factor data according to the meta information.

Additionally, the environmental hazard DB 181 is data collected from the external linkage system 400 and can store environmental hazards and/or data closely related to environmental hazards. In this case, the database unit 180 collects environmental hazards and Alternatively, data related to this may be stored.

The health impact factor DB 182 may store data received from the second IoT device 200 through the management unit 110. When storing health influencing factor data, the health influencing factor DB 171 may also store meta information generated by the data collection unit 120. In some embodiments, the data collection unit 120 may classify and store a plurality of health impact factor data according to the meta information.

Additionally, the health influencing factor DB 171 is data collected from the external linkage system 400 and can store health influencing factors and/or data closely related to health influencing factors. In this case, the database unit 180 collects environmental hazards and Alternatively, data related to this may be stored.

The data set DB 183 may store the data set generated by the data set generator 140. In this case, the data set DB 183 may classify and store a plurality of data sets based on meta information (or classification information) that is the basis for data set creation.

The correlation analysis DB 184 may store correlation analysis data and/or correlation prediction data generated by the correlation analysis unit 150. The correlation analysis DB 184 may store correlation analysis data and/or correlation prediction data according to the type of data that is the subject of correlation analysis (e.g., environmental hazards, health impact factors, or environmental hazards and health impact factors). .

The learning data DB 185 may store the learning data generated by the learning data generator 160. In some embodiments, when the learning data generator 160 updates the learning data, the learning data DB 185 may accumulate, classify, and store the learning data generated according to a preset cycle.

The user DB 186 may store information on at least one user. Here, the user is a user wearing the second IoT device 200, and the user information may include the user's personal information (eg, age, gender, occupation, disease, etc.). Additionally, the user information may include a user ID, password, etc. for the user to access any one of the website, application, or application program provided by the management unit 110.

In some embodiments, when storing a plurality of user information, the user DB 186 may set and store at least two or more user information as one group. For example, the user DB 186 may store information on at least two or more users as a family group.

The device DB 187 may store information about at least one first IoT device 200 and at least one second IoT device 200. For example, the device DB 187 may store device information for each of the first IoT device 200 and the second IoT device 300. Here, the device information stores the device management number, location information, and information about sensing data (e.g., type of measured sensing data, measurement cycle, etc.) of each of the first IoT device 200 and the second IoT device 300. You can. Additionally, the device DB 187 may classify and store the device information according to user information linked to each of the first IoT device 200 and the second IoT device 300.

In this way, the platform 100 according to one embodiment uses time dependence and different sensor modules in processing missing values of sensing data collected from the first IoT device 200 and/or the second IoT device 300. It can be advantageous to increase the reliability of data by replacing missing values using an ensemble method that considers the correlation between data collected through data collection.

In addition, the platform 100 provides effective linkage data, correlation analysis data, and/or correlation prediction data between environmental harmful factors and health influencing factors by analyzing the correlation between collected environmental harmful factors and health influencing factors, and provides environmental It can be advantageous to improve the accuracy of correlation analysis through mutual iterative machine learning between harmful factors and health-impacting factors.

In addition, the platform 100 sets areas of interest that are vulnerable to environmental pollution according to the characteristics of each of the plurality of areas, and assigns weights when analyzing the correlation between environmental hazards and health-impacting factors in the area of interest, thereby determining the level of environmental health vulnerability in areas. It can be advantageous for predicting situations or managing the impact of exposure to environmental hazards in areas vulnerable to environmental health. Furthermore, the platform 100 provides user-specific health information, health abnormality symptom information, and relationship data between environmental harmful factors and health abnormality symptom information, so that each user can quickly recognize the factors of his or her health abnormality and activity environment accordingly. It can be advantageous in resolving health abnormalities related to environmental harmful factors by improving .

Figure 3 is a block diagram schematically showing the configuration of a first IoT device according to an embodiment of the present invention.

Referring to FIG. 3, the first IoT device 200 may include a first communication unit 210, a first control unit 220, a first sensor unit 230, and a first data correction unit 240.

The first communication unit 210 can communicate data with the platform 100 by wire or wirelessly. For example, the first communication unit 210 may use a wired Internet communication method supporting TCP/IP (Transmission Control Protocol/Internet Protocol), or WCMDA (Wideband Code Division Multiple Access), LTE (Long Term Evolution), and WiFi ( Data may be transmitted and received between the first IoT device 200 and the platform 100 through at least one of various wireless communication methods such as Wireless Fidelity (Wireless Fidelity). The first communication unit 210 selects a communication method that ensures stability by examining the edge portion through an FPGA (Field Programmable Gate Array) among the various wireless communication methods, and transmits and receives data to the platform 100 accordingly. can do.

The first control unit 220 can manage and link each component of the first IoT device 200. For example, the first control unit 220 receives control commands from the gateway (not shown) and/or the platform 100 through the first communication unit 210 to control each configuration of the first IoT device 200. You can.

The first control unit 220 may generate sensing cycles of a plurality of sensor modules included in the first sensor unit 230. The first control unit 220 may generate a sensing cycle to synchronize the sensing timing of the plurality of sensor modules included in the first sensor unit 230, but is not limited thereto. For example, the first control unit 220 may synchronize the sensing cycle for each sensor module that senses dependent variables based on dependency data between a plurality of variables received from the platform 100.

In addition, when at least one sensor module among the plurality of sensor modules included in the first sensor unit 230 continuously senses, the first control unit 220 determines the period of transmitting the corresponding sensing data to the platform 100. The transmission cycle can be generated or changed according to the control command.

The first control unit 220 may transmit the sensing data of the first sensor unit 230 and/or the sensing data corrected by the first data correction unit 240 to the platform 100 through the first communication unit 210. there is.

The first sensor unit 230 may include a plurality of environmental sensor modules. For example, the first sensor unit 230 includes a temperature sensor module, a humidity sensor module, a fine dust/ultrafine dust sensor module, an acoustic sensor module, a Total Volatile Organic Compounds (TVOC) sensor module, and a carbon dioxide (CO ₂ ) sensor module. , carbon monoxide (CO) sensor module, hydrogen sulfide (H ₂ S) sensor module, formaldehyde (HCHO) sensor module, nitrogen oxides (NOx) sensor module, ammonia (NH ₃ ) sensor module, odor sensor module, and radon sensor module. may include.

The first data correction unit 240 may correct missing values for the sensing data sensed by the plurality of environmental sensor modules of the first sensor unit 230. For example, the first data correction unit 240 may process missing values of the sensing data using a missing value processing method provided by the platform 100.

In this way, the first IoT device 200 is advantageous in analyzing the dependency (or correlation) between the data collected by each environmental sensor module by including a plurality of environmental sensor modules as one device, and the platform 100 ) It can be advantageous to provide reliable sensing data by correcting raw data, such as processing missing values, before transmitting the sensing data to ).

Figure 4 is a block diagram schematically showing the configuration of a second IoT device according to an embodiment of the present invention.

Referring to FIG. 4 , the second IoT device 300 may include a second communication unit 310, a second control unit 320, and a first sensor unit 330. In another embodiment, the second IoT device 300 may further include a second data correction unit 340.

When the second IoT device 300 further includes a second data correction unit 340, the function and operating method of the second data correction unit 340 are those of the first IoT device 200 described with reference to FIG. 3. Since it may be substantially the same as the first data correction unit 240, detailed description thereof will be omitted.

The second communication unit 310 can communicate data to the platform 100 through short-distance wireless communication. For example, the second communication unit 310 may perform data communication with the platform 100 through a gateway (eg, a user terminal), but is not limited thereto. The second communication unit 310 may communicate with the gateway using a short-range wireless communication method using Bluetooth, but is not limited thereto.

The second control unit 320 can manage and link each component of the second IoT device 300. For example, the second control unit 220 receives control commands from the gateway (not shown) and/or the platform 100 through the first communication unit 310 to control each configuration of the second IoT device 300. You can. Since the second control unit 320 may be substantially the same as the first control unit 220 of the first IoT device 200, detailed description thereof will be omitted.

The second sensor unit 330 may include a plurality of health sensor modules. For example, the second sensor unit 330 includes a lung sound sensor module, a blood oxygen saturation sensor module, a respiration measurement sensor module, an electromyography sensor module, a heart rate measurement sensor module, an electrocardiogram sensor module, a pulse sensor module, a temperature sensor module, and/ Alternatively, it may include an electroencephalography (EEG) measurement sensor module.

Figure 6 is a flowchart illustrating a GIS-based environmental health monitoring method according to an embodiment of the present invention.

Referring to FIG. 6, first, the platform 100 collects sensing data from the first IoT device 200 and the second IoT device 300 (S100).

The management unit 110 may collect sensing data from the first IoT device 200 and the second IoT device 300 according to a communication method suitable for each of the first IoT device 200 and the second IoT device 300. .

The management unit 110 may collect environmental harmful factor data from the first IoT device 200. For example, the management unit 110 receives temperature data, humidity data, fine dust/ultrafine dust data, sound data, TVOC (Total Volatile Organic Compounds) data, carbon dioxide (CO ₂ ) data, Carbon monoxide (CO) data, hydrogen sulfide (H ₂ S) data, formaldehyde (HCHO) data, nitrogen oxides (NOx) data, ammonia (NH ₃ ) data, odor data, and/or radon data may be collected. In some embodiments, the management unit 110 may further collect illuminance data from the first IoT device 200.

The management unit 110 may collect health impact factor data from the second IoT device 300. For example, the management unit 110 may include lung sound data, blood oxygen saturation data, respiration measurement data, electromyography data, heart rate data, electrocardiogram data, pulse data, etc. sensed by a plurality of sensor modules of the second IoT device 300. Body temperature data and/or electroencephalography (EEG) measurement data may be collected.

Additionally, the platform 100 may collect data from the external linkage system 400 (S200).

The management unit 110 may collect data from the external linkage system 400 by scraping data according to preset data classification. Here, the preset data classification is meta information set by an administrator (not shown) who manages the platform 100 and/or generated by the data collection unit 120, such as environmental health-related data and living environment hazard factors. It may be meta information about data such as data, environmental pollution area information, regional weather information, etc., but is not limited thereto.

The preset data classification may be updated based on meta information generated through the data collection unit 120. For example, the preset data classification may be geographical information and/or temporal information among meta information about at least one piece of data generated by the data collection unit 120. Accordingly, the management unit 110 may collect data from the external linkage system 400 by scraping data according to geographical information and/or temporal information as meta information for the at least one piece of data.

And, the platform 100 can classify a plurality of data (S300).

The data collection unit 120 may analyze data transmitted from the management unit 110 to generate meta information of the data and classify it. The data collection unit 120 may generate meta information including data type, geographic information, temporal information including creation date and time, and personal information for each of the plurality of data.

For example, the data collection unit 120 first classifies (or, primary classification information) can be created. The data collection unit 120 may secondarily classify the primary classified data into detailed types (or generate secondary classification information). As a specific example, data classified as environmental hazards may be secondary classified into fine dust/ultrafine dust data, temperature data, humidity data, acoustic data, carbon dioxide data, carbon monoxide data, TVOC data, hydrogen sulfide data, etc.

The data collection unit 120 may classify a plurality of collected data. For example, the data collection unit 120 classifies each of the plurality of data into meta information, that is, by data type (e.g., by environmental hazards, by health-impacting factors), or by geographic information (e.g., by region, by facility, by residential environment). It can be classified according to time information (e.g., by month, day of the week, time of day), or personal information (by age, gender).

When the management unit 110 collects health impact factor data from two or more second IoT devices 300 for one user, the data collection unit 120 collects health impact factor data from each second IoT device 300. Data can be linked and classified.

Additionally, the data collection unit 120 may reflect the correlation analysis data provided by the correlation analysis unit 150 as a classification method for classifying a plurality of data. Accordingly, the data collection unit 120 may classify a plurality of data by reflecting the continuous correlation analysis data, and the effectiveness and utility of the plurality of data classification methods performed by the data collection unit 120 may be improved.

If the data collected from the external linkage system 400 is unstructured data, the data collection unit 120 may formalize the unstructured data. The data collection unit 120 can standardize unstructured data through data mining by type. For example, the data collection unit 120 may data mine text or images included in a plurality of unstructured data and manage them by labeling the data according to the data type.

The data collection unit 120 may structure a plurality of data classified into at least one meta information based on the meta information and store them in the database unit 180. Additionally, the data collection unit 120 may transmit the plurality of data to the data pre-processing unit 130. In this case, the data collection unit 120 may transmit classification information for each of the plurality of data to the data pre-processing unit 130.

Afterwards, the platform 100 can preprocess a plurality of data (S400).

The data preprocessing unit 130 can preprocess a plurality of data. For example, the data preprocessor 130 may filter and cleanse a plurality of data. The data preprocessor 130 can remove redundancy and errors in a plurality of data, and process missing values where there is no data or is incomplete. Since the missing value processing method of the data pre-processing unit 130 is substantially the same as that described with reference to FIG. 2, detailed description thereof will be omitted.

Additionally, the data pre-processing unit 130 may provide a missing value processing method suitable for each of the first IoT device 200 and/or the second IoT device 300 through the management unit 110. In this case, the data preprocessor 130 may provide a method for processing missing values for each of a plurality of variables. The data preprocessor 130 may provide dependency data between a plurality of sensed data (or variables) for each of the first IoT device 200 and the second IoT device 300.

The data pre-processing unit 130 can convert a plurality of data into a form that is easy to analyze and store the data in the database unit 180. For example, the data pre-processing unit 130 may convert the plurality of data transmitted from the data collection unit 120 into a form that facilitates data analysis according to the classification information of each, but is not limited thereto.

Additionally, the data preprocessor 130 may process a plurality of data according to the purpose of data analysis. The data pre-processing unit 130 may process each data based on classification information of each of the plurality of data.

For example, the data pre-processing unit 130 may process environmental hazards and health-impacting factors to expand the spatial and temporal range in order to analyze the correlation between the indoor environment and sleep quality and efficiency. The data preprocessing unit 130 expands temporal units (e.g., seconds, minutes, hours, days, months, years) and/or geographical units according to the temporal and geographical scope of the correlation analysis between environmental hazards and health influencing factors. (e.g. location, town, town, town, district, etc.) can be expanded.

The data pre-processing unit 130 may transmit a plurality of data including converted and/or processed data and data classification information to the data set generating unit 140.

Next, the platform 100 can create a data set by linking a plurality of data (S500).

The data set generator 140 may generate a data set by linking a plurality of data received from the data preprocessor 130. For example, the data set generator 140 may link data by using at least one of the meta information (or classification information) for each of the plurality of data as an identifying variable or variable of interest. there is.

For example, the data set generator 140 generates geographical information (e.g., by region, facility, residential environment), temporal information (e.g., by month, day of the week, time zone), adjacent to or identical to at least two pieces of data, and at least one data set can be created by linking data based on at least one of personal information (age, gender).

The data set generator 140 may provide the generated data set to the correlation analysis unit 150 and the learning data generator 160.

Then, the platform 100 can analyze the correlation between environmental hazards and health-impacting factors and generate correlation analysis data (S600).

The correlation analysis unit 150 may analyze the correlation between a plurality of data and generate correlation analysis data. For example, the correlation analysis unit 150 may analyze the correlation between environmental hazards and health-impacting factors and/or the correlation between a plurality of data sets using the Pearson correlation coefficient and the Pierceman correlation coefficient.

As a specific example, the correlation analysis unit 150 uses a plurality of independent variables such as indoor area, air conditioning, indoor smoking, air freshener, type of building, ventilation cycle, and height (or floor) from the ground as indoor environmental factors with respect to environmental harmful factors. , and temperature, humidity, TVOC, carbon dioxide, carbon monoxide, ammonia, radon, nitrogen oxides, etc. can be set as multiple dependent variables.

In addition, the correlation analysis unit 150 sets gender, age, occupation, smoking status, drinking status, BMI, etc. as human factors regarding health influencing factors as a plurality of independent variables, and sets heart rate, breathing, lung sounds, and brain waves (EEG) as multiple independent variables. ), coral saturation, sleep patterns, etc. can be set as multiple dependent variables.

The correlation analysis unit 150 uses canonical correlation analysis (CCA) to determine the independent variables and dependent variables included in environmental hazards, the independent variables and dependent variables included in health impact factors, and the environmental hazards. You can analyze the correlation between data sets and data sets included in health impact factors.

In this way, the correlation analysis unit 150 determines linearity and monotonicity between the variables included in each of the environmental hazards and health-impacting factors, or between the variables included in the environmental hazards and the variables included in the health-impacting factors. By checking, high-quality correlation analysis data can be generated.

The correlation analysis unit 150 can predict the correlation between health influencing factors and environmental harmful factors and generate correlation prediction data. Since health influencing factors and environmental harmful factors each have the characteristic of judging and predicting trends based on changes in past data, the correlation analysis unit 150 uses machine learning to process data with an order that changes over time. Using memory (Long Short-Term Memory, LSTM) and/or gated recurrent unit (GRU), the correlation between health influencing factors and environmental harmful factors can be determined and predicted.

The correlation analysis unit 150 can perform machine learning to analyze temporal and spatial correlations between environmental harmful factors, health-impacting factors, and environmental harmful factors and health-impacting factors. The machine learning method of the correlation analysis unit 150 may be substantially the same as that described with reference to FIG. 2. Therefore, for convenience of explanation, a detailed description of the machine learning method of the correlation analysis unit 150 will be omitted.

The correlation analysis unit 150 may transmit the correlation analysis data and/or correlation prediction data to the learning data generation unit 160 and the database unit 180. Additionally, the correlation analysis unit 150 may provide correlation analysis data to the data collection unit 120.

Furthermore, the platform 100 may generate learning data (S700).

The learning data generator 160 may generate learning data to advance at least one artificial intelligence (AI) used for data collection, missing value processing, and correlation analysis. The learning data generator 160 may generate learning data based on data type, geographic information, temporal information, and personal information.

For example, the learning data generator 160 may generate learning data based on the data set provided from the data set generator 140. Additionally, by updating the learning data by reflecting the correlation analysis data and/or correlation prediction data provided from the correlation analysis unit 150, more reliable and effective learning data can be generated.

Finally, the platform 100 may provide monitoring data (S800).

The monitoring unit 170 performs time-series monitoring for each of a plurality of environmental harmful factors and a plurality of health-impacting factors, and the time-series data, real-time monitoring data, and correlation generated by the correlation analysis unit 150 for each environmental harmful factor and health-impacting factor. Analysis data and correlation prediction data can be provided visually through the user interface (UI).

For example, the monitoring unit 170 is a geographic information system (GIS) and can provide a plurality of data including a plurality of environmental hazards, a plurality of health-impacting factors, etc. as spatiotemporal information. In addition, the monitoring unit 170 may provide environmental harmful factor trend prediction data, health influencing factor trend prediction data, and correlation prediction data predicted by the correlation analysis unit 150 according to time and space.

In addition, the monitoring unit 170 may analyze a plurality of data to classify areas vulnerable to environmental pollution and provide major influencing factors for each area vulnerable to environmental pollution and spatiotemporal information visualizing them.

For example, the monitoring unit 170 may derive characteristics of each of a plurality of regions based on data collected from the external linkage system 400. Here, the characteristics of each of the plurality of regions may include the presence or absence of a coal-fired power plant, nuclear power plant, or internal combustion power plant located in the region, the presence or absence of an industrial complex, the presence or absence of an industrial or abandoned mine, traffic density, and whether it is an undeveloped area. Additionally, the characteristics of each of the plurality of regions may include the population distribution of the region (eg, average age, occupation, etc. of the population residing in the region).

The monitoring unit 170 may set an area affected by environmental pollution as an area of interest according to the characteristics of each of the plurality of areas. Additionally, when multiple areas of interest are set, the monitoring unit 170 can calculate the ranking of the plurality of areas of interest. The monitoring unit 170 may assign search priority to the area of interest. For example, the monitoring unit 170 may assign weights when analyzing the correlation between environmental hazards and health-impacting factors for the region of interest.

Accordingly, in providing the spatio-temporal information, the monitoring unit 170 can provide correlation analysis data and correlation prediction data between environmental hazards, health-impacting factors, environmental hazards and health-impacting factors according to regional characteristics. .

Additionally, the monitoring unit 170 may provide user-specific health information (or health influencing factors), health abnormality sign information, and relationship data between environmental harmful factors and health abnormality symptom information.

In this way, the platform 100 according to one embodiment uses time dependence and different sensor modules in processing missing values of sensing data collected from the first IoT device 200 and/or the second IoT device 300. It can be advantageous to increase the reliability of data by replacing missing values using an ensemble method that considers the correlation between data collected through data collection.

In addition, the platform 100 provides effective linkage data, correlation analysis data, and/or correlation prediction data between environmental harmful factors and health influencing factors by analyzing the correlation between collected environmental harmful factors and health influencing factors, and provides environmental It can be advantageous to improve the accuracy of correlation analysis through mutual iterative machine learning between harmful factors and health-impacting factors.

In addition, the platform 100 sets areas of interest that are vulnerable to environmental pollution according to the characteristics of each of the plurality of areas, and assigns weights when analyzing the correlation between environmental hazards and health-impacting factors in the area of interest, thereby determining the level of environmental health vulnerability in areas. It can be advantageous for predicting situations or managing the impact of exposure to environmental hazards in areas vulnerable to environmental health.

6 to 14 are diagrams showing screens output by a monitoring unit through a user interface according to another embodiment of the present invention.

Referring to FIGS. 6 to 15 , the monitoring unit 170 may provide user-specific health information, health abnormality sign information, and correlation data between environmental harmful factors and health abnormality symptom information.

The monitoring unit 170 may provide various monitoring data through at least one of a website, an application, or an application program provided by the management unit 110 to at least one user terminal (not shown).

First, as shown in FIG. 6, the monitoring unit 170 includes the status and status of each of the plurality of first IoT devices 200 and the plurality of second IoT devices 300 registered in the platform 100. It can provide device monitoring data. In this case, the device monitoring data may be linked to the user data of at least one of a plurality of users, and the monitoring unit 170 may provide the device monitoring data to a user terminal linked to the device monitoring data. You can.

The monitoring unit 170 may provide a plurality of sensing data measured by at least one first IoT device 200 in the form of daily reports and monthly reports, as shown in FIG. 7 . For example, the monitoring unit 170 calculates daily and monthly average values for each of the plurality of environmental harmful factors measured by the first IoT device 200, and calculates the daily and monthly average values for each of the plurality of environmental harmful factors based on the pre-registered evaluation criteria. Evaluation information can be provided. Here, the pre-registered evaluation standard is an evaluation standard registered according to the numerical range of each of a plurality of environmental harmful factors, and the evaluation information is information set in response to the numerical range, for example, very good, good, average, bad. , very bad, etc.

In addition, as shown in FIG. 8, the monitoring unit 170 may provide comprehensive evaluation information for each of the plurality of sensing data measured by the first IoT device 200 in response to a specific period according to the user's settings. . For example, if the user sets the specific period from 0:00 on November 1, 2021 to 14:00 on November 5, 2021, the monitoring unit 170 is monitored by the first IoT device 200 during the specific period. The average value and evaluation information for each of the multiple measured environmental harmful factors can be provided.

In addition, as shown in FIG. 9, the monitoring unit 170 may provide trend information for each of a plurality of sensing data measured by the first IoT device 200 in response to a specific period according to the user's settings. . For example, the monitoring unit 170 may graph data on each of a plurality of environmental harmful factors measured during the specific period and provide visual data of changes in each of the plurality of environmental harmful factors during the specific period. .

The monitoring unit 170 may provide sensing data measured from the second IoT device 300 worn by at least one user. For example, the monitoring unit 170 provides a plurality of health information (or health impact factors) measured from a plurality of second IoT devices 300, as shown in FIGS. 10 and 11, or one A plurality of health information measured by the second IoT device 300 may be provided according to time or date.

Furthermore, the monitoring unit 170 may provide information on abnormal health symptoms for each user and data on the relationship between environmental harmful factors and information on abnormal health symptoms. For example, as shown in FIGS. 12 and 13, the monitoring unit 170 collects health abnormality sign information for each of a plurality of users, the health abnormality symptom collection history, and the average value of each of a plurality of environmental harmful factors during the collection period. and evaluation information can be provided.

In this specification, information on health abnormalities may include, but is not limited to, high blood pressure, low blood pressure, stress, atrial fibrillation, and blood oxygen. For example, information on health abnormalities may further include respiratory diseases, sleep disorders, etc.

Additionally, as shown in FIG. 14, the monitoring unit 170 may provide relationship data between a plurality of health abnormality sign information and a plurality of environmental harmful factors. For example, the monitoring unit 170 converts each of a plurality of information on health abnormalities occurring in a specific period (e.g., weekly, monthly) into a graph to provide visualized data, and provides visualized data for each of the plurality of health abnormalities. The average value and evaluation information of multiple environmental harmful factors corresponding to a specific period can be provided.

In this way, the monitoring unit 170 according to one embodiment provides user-specific health information, health abnormality sign information, and relationship data between environmental harmful factors and health abnormality symptom information, so that each user can determine the factors of his or her health abnormality. It can be advantageous to quickly recognize and improve the activity environment accordingly to eliminate health abnormalities related to environmental harmful factors.

So far, the present invention has been examined in detail, focusing on the preferred embodiments shown in the drawings. These embodiments are not intended to limit the invention but are merely illustrative and should be considered from an illustrative rather than a limiting perspective. The true technical protection scope of the present invention should be determined by the technical spirit of the appended claims rather than the foregoing description. Although specific terms are used in this specification, they are used only for the purpose of explaining the concept of the present invention and are not used to limit the meaning or scope of the present invention described in the claims. Each step of the present invention does not necessarily have to be performed in the order described, but may be performed in parallel, selectively, or individually. Those skilled in the art to which the present invention pertains will understand that various modifications and other equivalent embodiments are possible without departing from the essential technical spirit of the present invention as claimed in the patent claims. Equivalents should be understood to include not only currently known equivalents but also equivalents developed in the future, that is, all components invented to perform the same function regardless of structure.

100: service server 200: first IoT device
300: Second IoT device 400: External linkage system
110: Management Department 120: Data Collection Department
130: data preprocessing unit 140: data set generation unit
150: correlation analysis unit 160: learning data generation unit
170: monitoring unit 180: database unit
210: first communication unit 220: first control unit
230: first sensor unit 240: first data correction unit

Claims (18)

A platform that collects and monitors data on environmental hazards and health influencing factors by linking with a first IoT device including a plurality of environmental sensor modules and a second IoT device including a plurality of health sensor modules,
a management unit that collects environmental hazards and health-impacting factors from the first IoT device and the second IoT device, respectively;
a data collection unit that analyzes a plurality of data including the environmental harmful factors and the health-impacting factors to generate meta information for each of the plurality of data, and classifies the plurality of data based on the meta information;
a data set generator that generates a data set by linking the plurality of data based on at least one of the meta information;
A correlation analysis unit that analyzes the correlation between the environmental harmful factors, the correlation between the health-impacting factors, and the correlation between the environmental harmful factors and the health-impacting factors based on the data set, and generates correlation analysis data; and
A platform comprising; a monitoring unit providing spatiotemporal information visualizing the plurality of data and the correlation analysis data to a user terminal. According to claim 1,
A platform wherein the correlation analysis unit performs machine learning for temporal and spatial correlation analysis when generating the correlation analysis data. According to clause 2,
When performing machine learning to analyze the correlation between the environmental harmful factors, the correlation analysis unit sets at least one of temporal information, geographic information, meteorological information, and environmental harmful factors as an input variable, and sets at least one of the environmental harmful factors. A platform characterized by setting it as an output variable. According to clause 2,
When the correlation analysis unit performs machine learning to analyze the correlation between the health influencing factors and the correlation between the health influencing factors and the environmental harmful factors, at least one of the environmental harmful factors, humanities and social information, geographical information, and health impact A platform characterized by setting at least one of the factors as an input variable and setting at least one of the health impact factors as an output variable. According to claim 1,
A platform characterized in that the correlation analysis unit generates correlation prediction data between the environmental harmful factors and the health influencing factors. According to claim 1,
The management department collects data from an external linkage system that provides any one of environmental health-related data, living environment hazard data, and environmental pollution area information,
The plurality of data includes data collected from the external linkage system,
A platform wherein the monitoring unit analyzes data collected from the external linkage system to derive characteristics of each of a plurality of regions and classifies areas vulnerable to environmental pollution. According to clause 6,
A platform wherein the correlation analysis unit assigns weights when analyzing the correlation between the environmental harmful factors and the health-impacting factors for the area vulnerable to environmental pollution. According to claim 1,
The monitoring unit is a platform characterized in that it provides user-specific health information, plurality of abnormal health sign information, and relationship data between the environmental harmful factors and the plurality of abnormal health symptom information. According to clause 8,
The monitoring unit converts each of the plurality of health abnormality sign information generated in a period set by the user into a graph to provide visualized data, and provides the data corresponding to the period set by the user for each of the plurality of health abnormality signs. A platform characterized by providing average values and evaluation information of environmental harmful factors. A platform comprising: collecting environmental harmful factors and health influencing factors from a first IoT device including a plurality of environmental sensor modules and a second IoT device including a plurality of health sensor modules;
generating meta information for each of the plurality of data by analyzing a plurality of data including the environmental harmful factors and the health influencing factors, by the platform;
The platform classifying the plurality of data based on the meta information;
generating, by the platform, a data set by linking the plurality of data based on at least one of the meta information;
The platform analyzes the correlation between the environmental harmful factors, the correlation between the health influencing factors, and the correlation between the environmental harmful factors and the health influencing factors based on the data set, and generates correlation analysis data; and
An environmental health monitoring method comprising: providing, by the platform, spatiotemporal information visualizing the plurality of data and the correlation analysis data to a user terminal. According to claim 10,
The step of the platform generating the correlation analysis data is,
An environmental health monitoring method comprising the step of the platform performing machine learning for temporal and spatial correlation analysis. According to claim 11,
When the platform performs machine learning to analyze the correlation between the environmental harmful factors, at least one of temporal information, geographic information, weather information, and the environmental harmful factors is set as an input variable, and at least one of the environmental harmful factors is set as an input variable. An environmental health monitoring method characterized by setting as an output variable. According to claim 11,
When the platform performs machine learning to analyze the correlation between the health influencing factors and the correlation between the health influencing factors and the environmental harmful factors, at least one of the environmental harmful factors, humanities and social information, geographical information, and the health impact An environmental health monitoring method characterized by setting at least one of the factors as an input variable and setting at least one of the health impact factors as an output variable. According to claim 10,
The platform further includes the step of generating correlation prediction data between the environmental harmful factors and the health influencing factors. According to claim 10,
The platform further includes collecting data from an external linkage system that provides any one of environmental health-related data, living environment hazard data, and environmental pollution area information,
The plurality of data includes data collected from the external linkage system,
The platform is an environmental health monitoring method characterized in that it analyzes data collected from the external linkage system to derive characteristics of each of a plurality of regions and classifies areas vulnerable to environmental pollution. According to claim 15,
The step of the platform generating the correlation analysis data is,
The platform further includes a step of analyzing the correlation between the environmental harmful factors and the health influencing factors for the area vulnerable to environmental pollution,
An environmental health monitoring method, wherein the platform assigns weights when analyzing the correlation between the environmental harmful factors and the health influencing factors for the area vulnerable to environmental pollution. According to claim 10,
An environmental health monitoring method, characterized in that the platform further comprises providing user-specific health information, information on a plurality of abnormal health symptoms, and relationship data between the environmental harmful factors and the plurality of abnormal health symptom information. According to claim 17,
The step of the platform providing relationship data between the environmental harmful factors and the plurality of health abnormality symptom information,
The platform provides visualized data by converting each of the plurality of health abnormality symptom information generated in a period set by the user into a graph, and provides visualized data corresponding to the period set by the user for each of the plurality of health abnormality symptoms. An environmental health monitoring method characterized by providing average values and evaluation information of environmental harmful factors.