KR20150144562A

KR20150144562A - System and Method for Biometric Data Analysis

Info

Publication number: KR20150144562A
Application number: KR1020140073499A
Authority: KR
Inventors: 박광호; 김태웅; 김희철; 이상훈
Original assignee: 인제대학교 산학협력단
Priority date: 2014-06-17
Filing date: 2014-06-17
Publication date: 2015-12-28

Abstract

The present invention relates to a biometric data analysis system and method capable of processing real-time biometric data through expansion of the Hadoop platform and serving biometric data-related information on a network, and a biometric data analysis system includes a user terminal A service-oriented architecture for receiving real-time data from a network and providing a real-time biometric data service on a network, a pre-processing unit connected to a service-oriented architecture for extracting feature values of raw data, a conversion engine for converting raw data of the pre- And a Hadoop distributed file system that reads and classifies and parses the standard data of the conversion engine, and provides the service-oriented architecture with result values that are computed on the feature values.

Description

System and Method for Biometric Data Analysis [

Embodiments of the present invention relate to a biometric data analysis system and method, and more particularly, to a biometric data analysis system and method capable of processing real-time biometric data through expansion of the Hadoop platform and serving biometric data- .

The modern social health care service is changing to proactive health care oriented service centered on the consumer. The ubiquitous healthcare system using bio-signals according to these changes should be combined with technologies such as analysis and processing of bio-signals, real-time response, and storage of biometric data.

Biological signal means the instantaneous value of a signal generated in a human body. Data indicating simple characteristic values such as body temperature and blood pressure, and linear structure such as electrocardiogram, respiration, etc., As shown in Fig. Biological signals of a linear structure have a data type and quantity that are incompatible with the storage of the database. Biological signals of a linear structure need to be classified and processed as big data. In particular, bio-signals are data of linear structures that are not processed, and are not usually analyzable data.

Hadoop is the platform for handling big data. Hadoop is useful for analyzing batch type data accumulated for a certain period of time, but it is insufficient for use in a health care system that needs to process a linear structure bio signal in real time, and it is not suitable for service dimensional biometric data processing .

Accordingly, in one embodiment of the present invention, a biometric data analysis system and method capable of processing and servicing biometric signals in real time by extending the Hadoop platform.

In other words, in the embodiment of the present invention, the feature values are extracted from the bio-signals using the signal analysis algorithm, the bio-signals are converted into valid data formats and processed into analytical data, the expansion of the Hadoop platform And to provide a biometric data analysis system and method using a service-oriented architecture.

According to one aspect of the present invention, there is provided a biometric data analysis system for receiving raw data from a user terminal connected through a network and providing a real-time biometric data service on a network; A preprocessor connected to the service-oriented architecture and extracting characteristic values of the raw data; A conversion engine for converting a bio-signal including feature values into standard data; And the Hadoop distributed file system that reads the standard data of the conversion engine, classifies, parses and parses the feature values obtained by the operation, and provides the resultant values to the service-oriented architecture.

In one embodiment, the raw data comprises an acceleration signal. The acceleration signal can be extracted by the acceleration sensor of the user terminal and obtained from the service terminal of the service oriented architecture platform message from the user terminal.

In one embodiment, the preprocessor may include a motion artifact remover that removes noise from the raw data and returns a corrected value, and a controller that determines, based on each biometric signal, a corrected value of the motion artifact remover, a measurement time, And a feature value extracting unit that extracts a feature value through a bio-signal algorithm using a weight, a weight, and a sampling rate as parameters. The bio-signal algorithm corresponds to the feature value extraction algorithm.

In one embodiment, the conversion engine converts the biomedical signal into a standardized format in the aECG (Annotated ECG) format. The standardized format includes a first part representing information upon measurement of an ECG (Electrocardiogram), a second part representing information about a subject to be examined, a third part representing waveform information of the bio signal, and a fourth part representing annotations .

In one embodiment, the Hadoop distributed file system inherits the first class provided by Hadoop and implements the first class. The first file includes a first key containing the size of data from the aACC data according to the input format, An input unit for outputting a value; A map method for parsing a first key and a first value received as input parameters from an input unit and classifying the data into a data type and a characteristic value of a characteristic value and outputting a second value containing a characteristic value and a second key containing the data type; A reduction method for sorting and merging second keys and second values of the map mask with respect to the second keys and using them as parameters and outputting the result values through calculation of feature values; And an output unit for determining an output format of a result value of the redescription method and outputting a result value of the determined output format.

In one embodiment, the service-oriented architecture sends and receives messages in a mutually understandable format through a service-oriented architecture platform. The message includes a service-oriented architecture platform message having an attachment for data transmission and reporting with the biometric signal transmission module of the user terminal.

According to another aspect of the present invention, there is provided a biometric data analysis method comprising: receiving raw data from a user terminal through a service-oriented architecture connected to a user terminal through a network; Extracting feature values of raw data through a preprocessing unit connected to a service-oriented architecture; Converting the bio-signal including the feature value into standard data of a predetermined standardization format through a conversion engine connected to the preprocessing unit; Classifying and parsing standard data of a conversion engine through a Hadoop distributed file system connected to a service-oriented architecture, a preprocessing unit, and a conversion engine, and parsing and outputting a result value through an operation process on a feature value; And providing real-time biometric data related information to a user terminal on the network based on the result value in a service-oriented architecture.

In one embodiment, the raw data comprises an acceleration signal, the acceleration signal being extracted by an acceleration sensor of the user terminal and acquired from a service terminal of the service oriented architecture platform message from the user terminal.

In one embodiment, extracting the feature value of the raw data comprises: returning the corrected value through a motion artifact removal algorithm that removes noise from the raw data; And a step of extracting a feature value through a bio-signal algorithm using as parameters the return value (corrected value), the measurement time, the measurement date, the key, the body weight, and the sampling rate of the motion artifact elimination algorithm according to each bio- .

In one embodiment, the step of converting the biomedical signal (biometrics data) into the standardized format includes a step of converting the biomedical signal into a standardized format of aECG (Annotated ECG) format. The standardized format includes a first part representing information upon measurement of an ECG (Electrocardiogram), a second part representing information about a subject to be examined, a third part representing waveform information of the bio signal, and a fourth part representing annotations .

According to the present invention, it is possible to provide a biometric data analysis system and method for providing a biometric data service on a network, a web, or the like by real-time processing of a biometric signal through expansion of the Hadoop platform.

According to an embodiment of the present invention, the feature values are extracted from a bio-signal using a signal analysis algorithm, the bio-signal is converted into a valid data format and processed into analytical data, and the Hadoop platform is expanded for real- A biometric data analysis system and method using a service-oriented architecture can be provided.

1 is a block diagram of a biometric data analysis system according to an embodiment of the present invention;
Figure 2 is an exemplary diagram of a user terminal providing raw data to the biometric data analysis system of Figure 1;
Figure 3 is an exemplary diagram of raw data obtained through the user terminal of Figure 2;
Figure 4 is an exemplary diagram of the data format of the raw data collected from the user terminal of Figure 2;
FIG. 5 is a block diagram of a preprocessing section that can be employed in the biometric data analysis system of FIG.
FIG. 6 is a flowchart of an operation process of the Hadoop distributed file system that can be employed in the biometric data analysis system of FIG.
Fig. 7 is an exemplary diagram of biometric data stored in the Hadoop distributed file system of the biometric data analysis system of Fig. 1
FIG. 8 is an exemplary diagram of a real-time service of biometric data-related information of the biometric data analysis system of FIG. 1

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. It should be understood, however, that the embodiments described herein and the configurations shown in the drawings are only a preferred embodiment of the present invention, and that various equivalents and modifications may be made thereto at the time of the present application. DETAILED DESCRIPTION OF EXEMPLARY EMBODIMENTS Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the following description, well-known functions or constructions are not described in detail to avoid obscuring the subject matter of the present invention. The following terms are defined in consideration of the functions of the present invention, and the meaning of each term should be interpreted based on the contents throughout this specification. The same reference numerals are used for portions having similar functions and functions throughout the drawings.

1 is a block diagram of a biometric data analysis system according to an embodiment of the present invention.

The biometric data analysis system 10 according to the present embodiment uses an acceleration signal capable of extracting a bio-signal for motion in the body. In order to obtain a biological signal (electrocardiogram, respiration, etc.), a separate medical sensor is required. However, in the biometric data analysis system of the present embodiment, a user's biometric signal is obtained through an acceleration sensor built in a user terminal such as a smart phone.

That is, the biometric data analysis system 10 collects and analyzes an acceleration signal extracted through a built-in sensor of a user terminal from a user terminal through a predetermined message. To this end, the biometric data analysis system 10 is implemented such that interoperability HL7 (Health Level 7) conversion, MAP / REDAS analysis, and real-time service modules interact with each other.

More specifically, the biometric data analysis system 10 is implemented as an extension of the Hadoop platform including the existing Hadoop platform 13 as shown in FIG. In other words, the biometric data analysis system 10 includes a Service Oriented Architecture (SOA) 14 that combines with the Hadoop platform 13 and the Hadoop platform 13 to process and service real-time bio-signals. The biometric data analysis system 10 includes a preprocessor 15 for preprocessing a bio-signal between the SOA 14 and the Hadoop distributed file system 11 and a conversion engine 16 for converting the preprocessed biometric data into standard data. .

Hadoop is a framework that allows for distributed processing of large datasets across computer clusters using a simple programming model. Hadoop performs distributed storage of large amounts of data and parallel processing on multiple server clusters. The Hadoop platform 13 implementing the Hadoop includes a Hadoop Distributed File System (HDFS) 11 and a Map / Reduce 12. The Hadoop distributed file system 11 serves as a stable and fast storage because the files are divided and stored separately and operated as a name node and a data node.

In the present embodiment, the Hadoop distributed file system 11 reads standard data of the conversion engine 16, classifies and parses the standard data, and outputs a result value through an operation process on the feature value. The result value is provided to the service-oriented architecture 14 for real-time biometric data service.

The Hadoop distributed file system 11 also stores raw data (raw data 111), biometric signals including characteristic values, and standard data converted into a standard format. In this embodiment, the raw data is the ACC source data, and the standard data is the HL7 Annotated ECG (aECG) standardized format (HL7 aECG format) 112 data, but not limited thereto, u- care system or the like, as shown in FIG.

The MapReduce 12 is a programming model for data processing in a subproject of the Hadoop platform 13, and has a simple but inherent concurrency. Maple deuce 12 works by dividing the process into a map step and a reduction step. Each step has a key / value pair as its input and output, and its type can be defined in various ways. Usually, the map function performs a common operation on input elements, processes them, and classifies them to generate various intermediate values. The Reduce function is used to collect all elements and output a single result. The MapReduce program of the Hadoop platform can easily parallelize data processing through the implementation of the map functions and the reduction functions described above, and can be used universally.

In this embodiment, the MapReduce 12 refers to a means (software framework) for processing large-capacity data processing such as linear-structured ecological data in distributed parallel computing, or a component performing a function corresponding to such means. The MapReduce 12 analyzes the standard data transformed from the bio-signals and provides the analysis result values to the service-oriented architecture (SOA) 14.

Service Oriented Architecture (SOA) is an application that combines the functions of existing applications into functional units with business meaning, recombines them into software component units by means of standardized call interfaces, and then combines these services into business functions It is a software architecture that produces. This SOA service is a unit with standard interface-based functions that are not dependent on a specific technology or platform. Key features include loose coupling, composability, granularity, and discoverability. In addition, the service provider, the service requester, and the service repository are characterized in that they are organically linked.

In this embodiment, the SOA 14 implements a service that receives raw data from a user terminal connected through a network and provides biometric data related information on the network in real time. To this end, the SOA 14 includes a manager module 141 for managing a service-oriented architecture platform, and transmits and receives messages to and from a user terminal through a manager module (SOAP Manager) 141. The message received from the SOA 14 by the user terminal is implemented as a SOAP Message with attachment including an attached file for data transmission and reporting with the bio-signal transmission module of the user terminal. Such a message includes an acceleration signal, and the acceleration signal is classified as big data as biometric data of a linear structure.

That is, the acceleration signal included in the SOAP message is a linear structure of raw data that is not processed, and is not usually data that can be analyzed. Therefore, in the present embodiment, the raw data collected from the user terminal is regarded as the big data, the characteristic values are extracted from the raw data through the signal analysis algorithm of the preprocessing unit 15 to generate the corrected value, (Biomedical signal having a characteristic value) to the valid data format that can be analyzed. The conversion engine 16 may be an HL7 Transform Engine that supports the Health Level 7 (HL7) standard format.

Big data can be defined as the size of a dataset beyond the ability to collect, store, manage, and analyze data in existing database management tools. There are three attributes (volume, volume, speed, velocity, and variety) that are not structured data such as photos and videos. 3V). Big data also has high processing complexity, high processing and analysis flexibility, and low concurrent throughput. The classification according to the degree of stereotype of such big data is as shown in the following Table 1

In addition, the biometric data analysis system according to the present embodiment uses a service-oriented architecture platform (SOAP) -based web service (hereinafter referred to as a SOAP web service) for real-time service implementation.

SOAP Web service is a technology for sharing distributed contents in an abstract service form and sharing them in standardized form, and is a representative technology for realizing SOA concept. This web service technology aims to interoperate between applications built on heterogeneous platforms, and it is a user terminal that wants to freely use a service object or API (Application Protocol Interface) Server, etc.). The SOAP web service of this SOP is transmitted over Hypertext transfer protocol (HTTP) as a protocol that can access distributed objects without depending on specific distribution or platform. HTTP is a protocol that allows information to be exchanged on the World Wide Web ("Web").

As described above, the biometric data analysis system of the present embodiment real-time analyzes the biometric data of the linear structure through the expansion of the Hadoop platform and can provide the biometric data related information obtained through the analysis results to various user terminals on the web in real time.

2 is an exemplary view of a user terminal providing raw data, which is a biological signal, to the biometric data analysis system of FIG.

2, the biometric data analysis system according to the present embodiment receives a biometric signal based on an acceleration signal from the user terminal 21. [ The acceleration signal has three axis values of x, y and z axes. Therefore, the motion of the body can be detected using the values of the three axes. On the other hand, the signals of three axes may be different according to the position of the measuring device (user terminal, for example, smart phone, etc.) (pocket on the top, pocket on the bottom, etc.). In this paper, we implement the algorithm using Inje University 's internal documents and extract feature values from them.

We use the key, weight, and gender received from the SOAP message as parameters for algorithm analysis. The algorithm extracts information such as the number of steps, stride per step, and time per step by using the measured information and the acceleration sensor value.

For example, the process of acquiring and storing an acceleration signal is as follows.

That is, the reception of the acceleration signal is implemented by a servlet. Servlets are a core technology used to develop Web applications based on components in the Java platform, and because they operate on a thread-by-thread basis, they have the advantage of not slowing the response time, even if they accept users at the same time. In addition, it can be received by the POST method out of the GET method and the POST method, which are data transmission / reception methods of HTTP (Hypertext Transfer Protocol). With the POST method, a message is stored in a body of HTTP, so that it can be transmitted with more data than a non-GET. Accordingly, the SOA of the present embodiment can receive a SOAP message carrying a SOAP message in a body of HTTP from a user terminal, and the user terminal can receive a SOAP message via HTTP transmitted by the SOA.

The source code for acquiring and storing (sending and receiving) a SOAP message including an acceleration signal is shown in Table 2 below.

In FIG. 2, the SOAPMessageParser extracts the SOAP message part from the received HTTP message, and stores the element name and value of the SOAP body part as a key and a value of the Hashtable. This element contains the height, weight and sex of the measurer. SOAPAttachment has attachment information (such as file name). And saveLocation stores the received acceleration signal.

FIG. 3 is an exemplary view of raw data acquired through the user terminal of FIG. 2. FIG.

As shown in Fig. 3, the linearly structured bio-signal (hereinafter referred to as raw data) 31 is not data meaningful to the numerical value itself. Therefore, it is classified as untyped data with no type. In FIG. 3, the horizontal axis represents time and the vertical axis represents signal intensity.

For example, when raw data is stored at about 60 biometric signals per second, 216,000 data are accumulated in one hour measurement. The amount of such data is preferably classified as big data if viewed from a semantic standpoint. Also, when accumulating data over a long period of time, it is correct to approach the big data processing method instead of the general data processing.

Biological signals are biological signals that occur in the body and can be classified as shown in Table 3 according to their characteristics.

As shown in Table 3, the signals of various types of bio-signals are very small in size. Therefore, bio-signal measurement technology is an important measurement technology, and sensors that measure it include an oxygen saturation sensor, an electrocardiograph sensor, a blood pressure sensor, an acceleration sensor, a body temperature sensor, and a breathing sensor.

Data having simple characteristic values such as body temperature and blood pressure and data having linear structure such as electrocardiogram and breathing can be classified according to the type of bio signal to be collected. In storing these two types of data, the biometric signal having the characteristic value has a data type suitable for the database storage, but the biometric signal of the linear structure has the data type and volume which are not suitable for storing in the database I have.

In particular, a continuous linear structure biomedical signal alone can not be regarded as an indicator for determining the state of the body. It is necessary to process the data processing of the bio-signal. Therefore, in order to express the numerical value representing the state of the body, The feature value should be extracted from the bio-signal. To this end, the biometric data analysis system according to the present embodiment uses a preprocessor for extracting a specific signal characteristic value from a bio-signal of a linear structure.

4 is an illustration of a data format of raw data collected from the user terminal of FIG.

As shown in FIG. 4, in the raw data including the acceleration signal, the attachment file is represented by < Attachment >, but the actual message receives the HTTP message including the raw data. The received message is stored as an HL7 format document that does not contain the raw data as shown in Table 5 through the feature value extraction algorithm of the acceleration signal. As described above, the biometric data analysis system according to the present embodiment uses a SOAP Message with Attachment having a message structure as shown in FIG. 4 for data transmission and reporting between the biometric signal transmission module and the server of the user terminal.

SOAP is an eXtended Markup Langage (XML) message that can be guaranteed to be interoperable on any platform. SOAP Message with Attachment is a SOAP message that can include attachments to SOAP. It is a W3C note Note). With this SOAP message, one or more attachments can be sent in one SOAP message.

Binary data is also available for attachments. The way to distinguish the types of attachments is to use the MIME type, which is the same way as sending an attachment to an email. In this case, the SOAP message containing the attachment is divided into a SOAP Part and an Attachment Part. A SOAP Part is a part that contains a SOAP Evelope element. There must be one, and an Attachment Part is a part that contains Attachment.

Biometric data of a linear structure obtained from the user terminal in the SOA is shown in Table 4 below. Table 4 is an example of electrocardiogram biometric data as biometric data of a linear structure.

FIG. 5 is a block diagram of a preprocessing unit that can be employed in the biometric data analysis system of FIG. 1. FIG.

Referring to FIG. 5, the preprocessing unit 15 according to the present embodiment includes a motion artifact elimination algorithm 151 and a feature value extraction algorithm 152. The motion artifact removal algorithm 151 removes noise from the raw data and returns the corrected value. Then, the feature value extraction algorithm 152 extracts the feature value using the corrected value by the motion artifact elimination algorithm, the measurement time of the raw data, the measurement date, the key, the body weight, and the sampling rate, Extract the value. The feature value may include a maximum value (Peak), a measurement interval (Interval), and other predetermined feature values. The preprocessing unit 15 may be referred to as a feature value extracting unit.

More specifically, the raw data measured by the human body include not only biological signals, but also external signals such as noise, communication noise, and the like of the measuring device (user terminal, etc.). This problem is largely solved by the development of measuring sensors, but there are still problems with hardware and measurement environment. Therefore, in the present embodiment, the corrected values are returned through the motion artifact elimination algorithm 151 filtering out the unnecessary measured values, and the feature values extracted from the previously corrected biometric data through the feature value extraction algorithm 152 for each bio- Feature values are extracted and finally returned. At this time, the information on the correction value, measurement time, measurement date, key, weight, and sampling rate that have been previously returned is input to the feature value extraction algorithm 152 as a parameter.

It is necessary to convert the feature value and the biometric data into the standardized format in order to simultaneously store the feature value and the raw data extracted from the algorithm. To this end, the biometric data analysis system according to the present embodiment uses a conversion engine that expresses biometric data and characteristic values in a standardized format. For example, the conversion engine expresses biometric data and feature values in accordance with the HL7 Annotated ECG (aECG) format standardized format.

The HL7 aECG standard was developed by HL7's Regulatory Clinical Research Information Management (RCRIM) in response to the need for a digital ECG by the Food and Drug Administration (FDA) and was finally adopted as the standard acceptance of HL7 version 3 in January 2004, It has been approved by the American National Standards Institute (ANSI). The structure of aECG is composed of a first part representing information at the time of ECG measurement, a second part representing information about the subject to be examined, a third part representing waveform information of the measured bio signal and a fourth part being a comment representation part have. An example of aECG document is shown in Table 5.

Table 5 shows an example of the HL7 aECG structure, which expresses the r-r interval value, which is a feature value of the electrocardiogram signal.

In addition, the biometric data analysis system of this embodiment can express biometric data by redefining an aACC document. That is, a meta model of HL7 aECG is used to express a bio-signal in a standardized format.

To conveniently create a metamodel, Java code is generated from the aECG schema using the EMF (Eclipse Modeling Framework). EMF is a modeling framework and source code generation tool for building tools or applications based on a structured data model. EMF provides a mechanism for storing information in an XML file.

The Java code generated using the EMF can be expressed as a Java class in all the attribute values in representing the aECG as shown in Table 6 below. These classes can be used to generate aECG, aRESP, and aACC models.

As described above, while studies on acceleration signals are continuously performed, standardization of acceleration signals to be implemented is not supported. Nevertheless, since the acceleration signal has characteristics similar to electrocardiogram, the aECG model is used in this embodiment to express biometric data. That is, in this embodiment, biometric data is expressed by redefining the aACC document.

The generated source code of the aACC model adopted in this embodiment is shown in Table 7 below.

In Table 7, ACCParser stores each feature value and information as an object using an acceleration signal characteristic value extraction algorithm. The creator of ACCParser can be used to input the type of device that measured the bio-signal. The biometric data analysis algorithm of this embodiment supports two kinds of devices, but not limited to, "belt" and "phone". In the bio-data analysis system, raw data, gender, height, weight, measurement time, measurement time, and sampling rate are used as parameters.

Existing aECG documents store biological signal source data, but in this embodiment, raw data is stored and managed separately without storing or listing the aACC. That is, the xml document is a document for the representation of structural data, and the values represented by the raw data in the aACC document are in violation of the xml expression feature and may be unnecessary constraints in extracting feature values. Therefore, in this embodiment, the raw data is not separately displayed in the aACC document but stored separately in the Hadoop distributed file system (HDFS).

According to the present embodiment, the VitalSignModel can be inherited from a subclass by using a Hierarchical Distributed File System (HDFS) as a superclass, and can have polymorphism, reusability, and extensibility. Then, an ACCC format document can be generated from the ACCModel.

6 is a flow chart of the operation of the Hadoop distributed file system of FIG.

The Hadoop distributed file system according to the present embodiment includes an input unit, a map method unit, a reduction method unit, and an output unit.

The input unit inherits the first class provided by Hadoop, and outputs a first key containing a data size and a first value (aECG data) in an input format implemented by the Hadoop. The map method unit parses the first key and the first value received as input parameters from the input format processing unit and classifies the first key and the first value into a data type and a feature value of the feature value and outputs a second value containing the feature value and a second key containing the data type do. The reduction method part sorts and merges the second key and the second value of the map mask part by the second key, merges the second value, and outputs the resultant value through calculation of the feature value. Then, the output unit determines the output format of the result value of the reduction method unit and outputs the result value.

A detailed map / resume structure for biometric data analysis based on HL7 in the biometric data analysis process of the Hadoop distributed file system is as shown in FIG. The HL7 aECG document is an XML format. In Map / Reduce, an XML document is read, the input file is classified into key and value format data, and parsed, and the result value is output after calculating the feature value.

To this end, the Hadoop distributed file system according to the present embodiment reads the aECG.xml document stored in the HDFS first (S60, S61) as a biometric data analysis process in the map / reduce. The key is the size of the data and the value is aECG raw data (or aACC data).

Next, the key and the value are transferred to the input data of the map method, and the map method extracts the feature value from the input data (S62).

Next, the resultant feature extracted from the map method is output as a key (data type) and a value (data characteristic value) (S63). The output key corresponds to the information per step, and the value corresponds to the information feature value per step.

Next, the output values from the map method are sorted and merged (S64). In this step S64, the sorted and merged key corresponds to the information type per step, and the value may correspond to the information feature value list per step.

Next, the key and the value output in the step S64 are transmitted as input data of the reduction method, and the reduction method performs an operation according to the characteristic value (S65).

Next, the reduction method outputs the result of the operation according to the feature value as a key and a value (S66). Here, the key corresponds to the information type per step, and the value corresponds to the information feature value calculation result per step.

Finally, the Hadoop distributed file system stores the output of the reduction method as output data (S67).

The following explains the MapReduce used in the biometric data distribution system of this embodiment in more detail.

A. Implementation of Map / Reduce

In Map / Reduce, feature values are extracted by parallel processing from acceleration data stored in HDFS. Manages and operates map / reduce using the Job class. Use the Job's methods to specify the mapper / reducer class, the InputFormat, the OutputFormat class, and the redo output format class. The user executes the job by calling the waitForCompletion method, specifying the input file path and the output file path.

The source code of the ACC driver used in the map / resume is shown in Table 8 below.

In Table 8, the map method classifies the key (data size) and value (aACC) output from XmlInputFormat by feature value element and collects data. XmlParser is a class that finds the element whose feature value is represented in aACC. The element thus found outputs the information per step (time per step, distance, consumed calories) to key, and the information feature value per step to value.

The mapping method source code of the acceleration signal is shown in Table 9 below.

On the other hand, the key and value in Reduce are sorted by value in list by key. That is, the values are sorted into a list for information per step. In this embodiment, since the information on the exercise amount is provided, the calculation of the values for the exercise information per step should be performed. Therefore, the calculation operation is performed on the motion information per step in Reduce.

Acceleration Signal Reduce Method Source code is shown in Table 10 below.

Feature values extracted through Map / Reduce are provided to the user application according to the SOAP Message format. To this end, the SOA of this embodiment takes charge of not only the reception of a bio-signal, but also the generation and transmission of a SOAP message. For example, a SOAP message element is created using a SOAP message class, and the element can represent information such as a user ID (ID), measurement date, feature value, and the like.

7 is an exemplary diagram of biometric data including an acceleration signal measured at a user terminal. 8 is an example of a biometric data service provided to a user terminal through a web browser.

As shown in FIG. 7, biometric data is acquired using an acceleration signal extracted from the smartphone at about 50 Hz per second. The data of each row of the biometric data in Fig. 7 is composed of time information, x-axis, y-axis, and z-axis information, and represents a part of the acceleration signal (see Fig. 2).

Biometric data can be received from the user terminal using the SOAP Messages with attachment and message format as shown in Table 11 below.

In Table 11, the message can be divided into SOAP Header, Body, and Attachment, and the element of the body represents user information and measurement information. In Attachment, signal measurement time (time information), x-axis, y-axis, and z-axis values are expressed (see FIG. These raw data are stored in a HL7 format document (see Table 5) that does not contain the raw data through the preprocessor and the conversion engine.

Biometric data related information in the standardized format stored in the Hadoop Distributed File System (HDFS) can be viewed through a Web browser. The HDFS storage document may have a data format (HL7 or the like) as shown in Fig.

HL7 documents can be stored in HDFS and extracted from the map / resize and output in the format shown in Table 12. That is, the output data may include a result of the number of steps, the distance, and the step time (measurement time) with respect to the measurement result.

The result obtained from the map / reduce is provided to the user in the form of a SOAP message. Table 13 below is an example of a SOAP message for providing the user terminal with the analysis result value of the biometric data including the acceleration signal.

The predetermined application of the user terminal can receive the message of Table 13 and then extract the necessary result values by parsing the received message and provide biometric data related information (information per step) to the user based on the extracted result value.

The existing map / reduce method is currently composed of concise logic and it is still not enough to show the health index or the exercise index using the biometric data. Therefore, in this embodiment, the feature values are extracted only from the biometric data of the acceleration signal, and the biometric data related information is provided to the user.

In the above-described embodiment, the feature value is extracted only for the biometric data of the acceleration signal for the sake of convenience of explanation, and the biometric data related information is provided. However, the present invention is not limited to such a configuration, and various biometric data such as electrocardiogram, To apply biometric data analysis to the biometric data. In order to perform such research, the following three conditions are required for the expression of biometric data, the distributed processing of data, and the service items that can be provided to the user.

First, the biometric data is represented in the HL7 medical standard document format and a medical standard document is established for various biological signals as well as acceleration signals.

Second, the distributed processing of biometric data is designed as a system that predicts and processes various environments and situations.

Third, it implements a map / reduce method that can collect and analyze distributed biometric data. Here, the map / method extracts the health index or the movement index through data mining. In the case of applying to the present embodiment, the accuracy and the processing speed of data processing may be important factors.

The research that defines and processes the biometric data from the viewpoint of big data is a research that needs collaboration among experts in each field. The biometric data analysis technique of the present invention can be a cornerstone for collaboration when storing and processing biometric data as big data through collaboration research in the future.

In short, ubiquitous healthcare services are expanding with the development of medical environment, social awareness, and personal income. In ubiquitous healthcare, biometric data is an indicator of health status. Therefore, classification and storage of biometric data should be an important factor. However, there is a lack of clear standards and dissemination of biometric data storage methods. In addition, unlike data with simple characteristics such as body temperature and blood pressure, which are easy to store in the database, data of a linear structure continuously flowing into the electrocardiogram and breathing are data types and quantities not fit to the database. Accordingly, in the biometric data analyzing technology according to the present embodiment, limitations of the linear data of bio-signals as general data are listed, re-classified into big data, and the Hadoop platform is extended to fit the ubiquitous health care system.

As described above, the biometric data analysis system proposed in the present invention includes a bio-signal reception, a feature value extraction algorithm, a bio-signal standardization format conversion engine, a map / reuse for feature value extraction, and a platform for real-time service. Here, the bio-signal standard HL7 document includes the bio-signal source data, characteristic values, and other information, and the bio-information measured by this document alone can be expressed. In addition, since it is a standard document, mutual compatibility can be ensured and it can be provided to an institution such as a hospital conforming to the standard. In addition, it is possible to provide a service for extracting feature values from standard documents using Hadoop map / reduce and returning them to users.

In addition, the biometric data analysis system proposed in the above embodiment is implemented so that a user (user terminal) can receive services in any environment using the SOA structure. When SOA is applied, the processing cost of biometric data can be modularized, and development costs and time can be saved when developing other bio-signals. It can also integrate heterogeneous environments with the advantages of high reusability.

In addition, modern society is the age of data. Storage and mining technologies for vast data, which are increasing in short time, short term and exponential, are developing and their importance is increasing. These technologies should be combined with the medical field, and biometric data storage / processing as big data should be prepared. Therefore, the biometric data analysis technique according to the present embodiment can be used as a useful technique leading to the atmosphere of the times.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments, but, on the contrary, Those skilled in the art will readily appreciate that many suitable modifications and variations are possible in light of the present invention. Accordingly, all such modifications and variations as fall within the scope of the present invention should be considered.

10: Biometric data analysis system
11: Hadoop Distributed File System (HDFS)
12: Maple Deuce (Map / Reduce)
14: Service Oriented Architecture (SOA)
15:
16: The conversion engine

Claims

A service-oriented architecture that receives raw data from a user terminal connected through a network and provides a real-time biometric data service on the network;
A preprocessing unit connected to the service-oriented architecture and extracting feature values of the raw data;
A conversion engine for converting the bio-signal including the feature value into standard data; And
A Hadoop distributed file system for reading and classifying and parsing standard data of the conversion engine, outputting a result value through an operation process on the feature value, and providing the result value to the service-oriented architecture;
And a biometric data analysis system.

The method according to claim 1,
Wherein the raw data comprises an acceleration signal and the acceleration signal is extracted by an acceleration sensor of the user terminal and obtained through a service oriented architecture platform message of the service oriented architecture.

The method according to claim 1,
The preprocessing unit may include a motion artifact removing unit that removes noise from the raw data and returns a corrected value, and a correction unit that corrects the corrected value, the measurement time, the measurement date, the key, the body weight, And a feature value extracting unit for extracting a feature value through a bio-signal algorithm.

The method of claim 3,
Wherein the conversion engine converts the bio-signal into a standardized format of an ECG (Annotated ECG) format, the standardized format includes a first portion that represents information upon measurement of an ECG (Electrocardiogram), a second portion that represents information A third part representing the waveform information of the bio-signal, and a fourth part representing the annotation.

The method according to claim 1,
In the Hadoop distributed file system,
An input unit configured to inherit a first class provided by Hadoop and output a first key containing a size of data and a first value which is aECG data in an input format implemented;
A first key and a first value received as input parameters from the input unit are parsed and classified into a data type and a feature value of a feature value and a second key containing the data type and a second value containing the feature value are output method;
A reuse method for sorting and merging a second key and a second value of the map method by a second key for use as a parameter and outputting a result value through operation of the feature value; And
And an output unit for determining an output format of a result value of the reduction method and outputting the result value.

The method according to claim 1,
The service-oriented architecture sends and receives messages in a mutually understandable format through a service-oriented architecture platform, the messages comprising service-oriented architecture platforms with attached files for data transmission and reporting with the biometric signal transmission module of the user terminal A biometric data analysis system comprising a message.

Receiving raw data from the user terminal via a service-oriented architecture connected to the user terminal through a network;
Extracting feature values of the raw data through a preprocessor connected to the service-oriented architecture;
Converting the biometric signal including the feature value into standard data through a conversion engine connected to the preprocessing unit;
Reading the standard data of the conversion engine through the Hadoop distributed file system connected to the service-oriented architecture, the preprocessor and the conversion engine, classifying and parsing the standard data, and outputting the resultant through an operation process on the feature value; And
Providing real-time biometric data related information on the network based on the resultant value in the service-oriented architecture;
And analyzing the biometric data.

The method of claim 7,
Wherein the raw data comprises an acceleration signal and the acceleration signal is extracted by an acceleration sensor of the user terminal and obtained through a service oriented architecture platform message of the service oriented architecture.

The method of claim 7,
Wherein the extracting of the feature value of the raw data comprises:
Returning the corrected value through a motion artifact removal algorithm that removes noise from the raw data; And
Extracting a feature value through a bio-signal algorithm using the corrected value, the measurement time, the measurement date, the key, the body weight, and the sampling rate as parameters;
And analyzing the biometric data.

The method of claim 9,
The step of converting the biomedical signal into a standardized format may include converting the biomedical signal into a standardized format of an ECG (Annotated ECG) format, wherein the standardized format is a first A second portion that represents information on a subject to be examined, a third portion that represents waveform information of the bio-signal, and a fourth portion that expresses the annotation.