KR102267105B1 - Apparatus and method for inspecting sleep disorder based on deep-learning - Google Patents

Abstract

The present invention relates to a deep learning-based polysomnography examination technology, and a polysomnography examination apparatus according to an embodiment receives the collected polysomnography data and converts the format of the received data to an artificial neural network (Artificial Neural Network). A data conversion unit that converts into an applicable format as an input of a neural network), a decision model generator that generates a sleep state determination model by performing artificial neural network-based machine learning with the converted data as an input, and the generated determination model It may include a result determination unit that determines the polysomnography test results based on the results.

Description

Deep learning-based polysomnography and method {APPARATUS AND METHOD FOR INSPECTING SLEEP DISORDER BASED ON DEEP-LEARNING}

The present invention relates to a polysomnography test apparatus and method, and more particularly, to a technical idea for automating the analysis of polysomnia test results.

Polysomnography is a test for diagnosing sleep disorders. It comprehensively measures EEG, eye movement, muscle movement, respiration, and electrocardiogram during sleep. It is used to diagnose sleep-related diseases by analyzing the records and to set treatment policies.

The above-described polysomnography test can diagnose symptoms such as sleep apnea, sleep disorder, and sleep gait. As an index for determining these diseases, sleep stage, apnea-hypopnea index (AHI) , upper respiratory tract resistance syndrome (Respiratory effort-related arousals; RERA) index and the like are used.

On the other hand, polysomnography test uses a manual sleep scoring method in which experts judge the above-mentioned indices by combining the patient's biometric data measured through various sensors.

In other words, in the current polysomnography test, sleep scoring is performed according to the subjective judgment standard of each professional manpower, so there is a problem in that the accuracy and reliability of the scoring result are lowered.

Specifically, the polysomnography test seems to have very clear criteria for scoring sleep, but the shape and size of biosignals are different for each patient, and due to atypicalities appearing in biosignals, there are many errors even among specialists. In fact, it is known that when experts perform sleep scoring with the same biometric data of the same patient, there is a scoring deviation of about 10%.

In addition, as sleep scoring is performed manually by a professional manpower, there is a problem that the required time is too long. In fact, it is known that it takes about 3 to 4 hours for a skilled professional to perform sleep scoring for one patient.

Korean Patent Registration No. 10-1395197 "Method and device for automatic detection of sleep state and wake state"

An object of the present invention is to provide a polysomnia test apparatus and method capable of minimizing the time required for polysomnia test by automating analysis of polysomnia test results through a sleep state determination model.

Another object of the present invention is to provide a polysomnography examination apparatus and method capable of improving the accuracy and reliability of polysomnia examination by generating a sleep state determination model through artificial neural network-based machine learning.

The polysomnography test apparatus according to an embodiment receives polysomnography data, and converts a format of the received polysomnography data into a format applicable as an input of an artificial neural network. It may include a conversion unit and a result determination unit that determines the polysomnography test result by applying the converted polysomnia test data to the sleep state determination model.

According to one side, the received polysomnography data is an EEG (Electroencephalogram) sensor, EOG (Electrooculography) sensor, EMG (Electromyogram) sensor, EKG (Electrokardiogramme) sensor, PPG (Photoplethysmography) sensor, Chest belt, Abderman. The user's biometric data measured through at least one sensing means of an Abdomen belt, a thermister, a flow sensor, and a microphone may be included.

According to one side, the received polysomnography data may include data recorded in the polysomnography database by the polysomnography performed in advance.

According to one side, the collected polysomnography data may further include sleep scoring data with respect to the user's biometric data.

According to one side, the data converter may match the sampling frequency for each sensor corresponding to the received polysomnography data, and extract necessary data corresponding to the sleep state determination model from the data with the sampling frequency matched. .

According to one side, the data conversion unit may divide the extracted necessary data into time units corresponding to the sleep state determination model, and label the divided necessary data with preset labels.

According to one side, the sleep state determination model may be a model for performing sleep scoring according to a polysomnography test.

According to one side, the sleep state judgment model is a sleep stage, apnea-hypopnea index (AHI), respiratory disturbance index (RDI), upper airway resistance syndrome (Respiratory effort) according to the polysomnography test. It may be a model for determining at least one of a -related arousals; RERA) index, a total sleep time (TST), an arrhythmia, and a Periodic limb movements of sleep index.

According to one side, the polysomnography examination apparatus may further include a determination model generator configured to generate a sleep state determination model by performing artificial neural network-based machine learning by inputting the converted polysomnia examination data.

According to one side, the judgment model generator may receive the determined polysomnography test result and re-perform artificial neural network-based machine learning.

According to one side, the polysomnia examination apparatus may further include a result information processing unit that converts the determined polysomnia examination result into at least one format of a graph and text and outputs the converted polysomnia examination result to the outside.

A polysomnography test method according to an embodiment receives polysomnography data from a data converter, and converts a format of the received polysomnia test data as an input to an artificial neural network. and applying the polysomnia test data converted by the result determining unit to the sleep state determination model to determine the polysomnography test result.

According to one side, the converting step includes matching the sampling frequency for each sensor corresponding to the collected polysomnography data, and extracting necessary data corresponding to the sleep state determination model from the data in which the sampling frequency is matched. and dividing the extracted necessary data into time units corresponding to the sleep state determination model, and labeling the divided necessary data with preset labels.

According to one side, the polysomnography examination method may further include generating a sleep state determination model by performing artificial neural network-based machine learning by inputting polysomnia examination data converted by the determination model generating unit.

According to one side, the polysomnia test method may further include converting the polysomnia test result determined by the result information processing unit into at least one format of a graph and text and outputting it to the outside.

According to an embodiment, by automating the analysis of polysomnography test results through the sleep state determination model, the time required for polysomnography can be minimized.

According to an embodiment, by generating a sleep state determination model through artificial neural network-based machine learning, the accuracy and reliability of polysomnography can be improved.

1 is a view for explaining a polysomnia examination apparatus according to an embodiment.
FIG. 2 is a diagram for explaining an example of polysomnia test data received by the polysomnography test apparatus according to an exemplary embodiment.
3 is a view for explaining an example of a data conversion unit provided in the polysomnia examination apparatus according to an embodiment.
4 is a view for explaining an example of a judgment model generator provided in the polysomnography examination apparatus according to an embodiment.
5A to 5B are diagrams for explaining an example of outputting a polysomnia test result from a result information processing unit provided in the polysomnography test apparatus according to an exemplary embodiment.
6 is a diagram for explaining a polysomnia examination method according to an embodiment.
7 is a diagram for explaining an example of a method of changing a format of polysomnia examination data in a polysomnography examination method according to an embodiment.

Specific structural or functional descriptions of the embodiments according to the concept of the present invention disclosed herein are only exemplified for the purpose of explaining the embodiments according to the concept of the present invention, and the embodiment according to the concept of the present invention These may be embodied in various forms and are not limited to the embodiments described herein.

Since the embodiments according to the concept of the present invention may have various changes and may have various forms, the embodiments will be illustrated in the drawings and described in detail herein. However, this is not intended to limit the embodiments according to the concept of the present invention to specific disclosed forms, and includes changes, equivalents, or substitutes included in the spirit and scope of the present invention.

Terms such as first or second may be used to describe various elements, but the elements should not be limited by the terms. The above terms are used only for the purpose of distinguishing one element from another element, for example, without departing from the scope of rights according to the concept of the present invention, a first element may be named as a second element, Similarly, the second component may also be referred to as the first component.

When a component is referred to as being “connected” or “connected” to another component, it is understood that the other component may be directly connected or connected to the other component, but other components may exist in between. it should be On the other hand, when it is mentioned that a certain element is "directly connected" or "directly connected" to another element, it should be understood that the other element does not exist in the middle. Expressions describing the relationship between elements, for example, “between” and “between” or “directly adjacent to”, etc. should be interpreted similarly.

The terminology used herein is used only to describe specific embodiments, and is not intended to limit the present invention. The singular expression includes the plural expression unless the context clearly dictates otherwise. In this specification, terms such as "comprise" or "have" are intended to designate that the described feature, number, step, operation, component, part, or combination thereof exists, and includes one or more other features or numbers, It should be understood that the possibility of the presence or addition of steps, operations, components, parts or combinations thereof is not precluded in advance.

Unless defined otherwise, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Terms such as those defined in a commonly used dictionary should be interpreted as having a meaning consistent with the meaning in the context of the related art, and should not be interpreted in an ideal or excessively formal meaning unless explicitly defined in the present specification. does not

Hereinafter, embodiments will be described in detail with reference to the accompanying drawings. However, the scope of the patent application is not limited or limited by these examples. Like reference numerals in each figure indicate like elements.

1 is a view for explaining a polysomnia examination apparatus according to an embodiment.

Referring to FIG. 1 , the polysomnography examination apparatus 110 according to an embodiment may minimize the time required for the polysomnia examination by automating the analysis of the polysomnia examination result through the sleep state determination model.

In addition, the polysomnography examination apparatus 110 according to an embodiment generates a sleep state determination model through artificial neural network-based machine learning, thereby improving the accuracy and reliability of the polysomnia examination.

To this end, the polysomnography examination apparatus 110 may include a data conversion unit 111 , a determination model generation unit 112 , a result determination unit 113 , and a result information processing unit 114 .

Specifically, the data conversion unit 111 according to an embodiment may receive polysomnography data.

For example, the collected polysomnography data may include an Electroencephalogram (EEG) sensor, an Electrooculography (EOG) sensor, an Electromyogram (EMG) sensor, an Electrokardiogramme (EKG) sensor, a Photoplethysmography (PPG) sensor, a chest belt, and Abderman. The user's biometric data measured through at least one sensing means of an Abdomen belt, a thermister, a flow sensor, and a microphone may be included.

In other words, the data conversion unit 111 includes an EEG sensor for measuring brain waves, an EOG sensor for measuring eye movement, an EMG sensor for measuring muscle movement, an EKG sensor for measuring heartbeat, and oxygen saturation. and at least one measured through a PPG sensor for measuring heart rate, a thermistor and flow sensor for measuring respiration, a chest belt and Abderman belt for measuring movement of the abdomen and chest, and a microphone for measuring snoring of the user's biometric data can be received.

According to one side, the received polysomnography data may include data recorded in the polysomnia test database 120 by the polysomnography test performed in advance.

In addition, the collected polysomnography data may further include sleep scoring data with respect to the user's biometric data.

More specifically, the data conversion unit 111 may receive polysomnia test data for generating a sleep state determination model from the polysomnia test database 120 , and sleep through a path separate from the polysomnia test database 120 . Polysomnia test data for determining a polysomnia test result may be received through the state determination model.

That is, the 'polysomnography test result' shown in FIG. 1 represents polysomnia test data for determining the polysomnia test result through the sleep state determination model.

In addition, the polysomnia test data received from the polysomnia test database 120 may include the user's biometric data derived through the previously performed polysomnia test and sleep-scored data for the user's biometric data.

Meanwhile, the data converter 111 according to an exemplary embodiment may convert a format of the received polysomnography data into a format applicable as an input of an artificial neural network.

More specifically, the data conversion unit 111 extracts necessary data corresponding to the sleep state determination model from among the received polysomnography examination data, and converts the extracted necessary data into a format having a format and a time interval suitable for the sleep state determination model. can do.

According to one side, the data conversion unit 111 matches the sampling frequency of each sensor corresponding to the collected polysomnography data, and the required data corresponding to the sleep state determination model among the data with the sampling frequency matched. can be extracted.

Next, the data conversion unit 111 divides the extracted necessary data into time units corresponding to the sleep state determination model, and labels the divided necessary data with preset labels, thereby receiving polysomnography test data format can be converted into a format applicable to the artificial neural network input.

For example, the preset label may be a label extracted through sleep scored data.

For a more specific example, when the sleep state determination model is a model for determining the sleep stage, the data conversion unit 111 manually displays the divided necessary data W (Wake stage), N1 by sleep scoring experts in advance. (Sleep stage 1), N2 (sleep stage 2), N3 (sleep stage 3), and R (REM sleep stage) can be labeled with at least one label.

A detailed configuration of the data conversion unit 111 according to an embodiment will be described in more detail later with reference to FIG. 3 according to the embodiment.

Next, the decision model generating unit 112 according to an embodiment performs artificial neural network-based machine learning by inputting polysomnographic examination data whose format is converted through the data conversion unit 111 to obtain a sleep state determination model. can create

According to one side, the sleep state determination model may be a model for performing sleep scoring according to a polysomnography test.

In other words, the sleep state determination model may be a model that automatically performs sleep scoring on the input polysomnia examination data by receiving as an input polysomnia examination data for determining the polysomnia examination result through the sleep state determination model. have.

In addition, the sleep state judgment model is a sleep stage, apnea-hypopnea index (AHI), respiratory disturbance index (RDI), upper airway resistance syndrome (Respiratory effort-related) according to the polysomnography test. It may be a model for determining at least one of arousals; RERA) index, total sleep time (TST), arrhythmia, and periodic limb movements of sleep index, but the sleep state according to an embodiment The judgment model is not limited to the above-described example, and may be a model for judging various symptoms that can be derived through polysomnia examination.

According to one side, the decision model generating unit 112 extracts features from the converted data by using not only the data converted through the data conversion unit 111, but also a similarity calculation algorithm such as cross correlation, The extracted features can also be used as input to the artificial neural network.

In addition, the judgment model generator 112 may generate a sleep state judgment model using at least one artificial neural network of a recurrent neural network (RNN), a deep neural network (DNN), and a convolution neural network (CNN), but one The decision model generating unit 112 according to the embodiment is not limited to the above-described example, and various types of artificial neural networks may be used.

On the other hand, the decision model generator 112 uses a back propagation algorithm, which is an algorithm for updating the weight of the neural network by using the labeled data of the output layer provided in the artificial neural network. Thus, a sleep state determination model may be generated.

Next, the result determination unit 113 according to an exemplary embodiment may determine the polysomnography examination result by applying the polysomnia examination data converted through the data conversion unit 111 to the sleep state determination model.

More specifically, when polysomnography data for determining a polysomnia test result is input through the sleep state determination model, the data conversion unit 111 converts the format of the input polysomnia test data and converts the format-converted data. The result may be transmitted to the determination unit 113 .

Next, the result determination unit 113 may determine the polysomnography test result by applying the format-converted data received from the data conversion unit 111 as an input of the determination model received from the determination model generation unit 112 . have.

According to one side, the decision model generator 112 may receive the polysomnography test result determined from the result determiner 113 and re-perform the artificial neural network-based machine learning.

In other words, the judgment model generator 112 receives the determined polysomnography test result as a feedback and re-performs artificial neural network-based machine learning to generate a learned sleep state judgment model, thereby determining the polysomnography test result. The accuracy can be further improved.

According to one side, the result information processing unit 114 may convert the determined polysomnography test result into at least one format of a graph and text and output it to the outside.

In other words, the result information processing unit 114 converts the polysomnography test result determined by the result determination unit 113 into a graph format or a text format that collects and organizes only the results and may provide it to a clinician or a user, but the decision The polysomnography test results obtained are not limited to the above-described examples and may be converted into various formats desired by a clinician or a user.

FIG. 2 is a diagram for explaining an example of polysomnia test data received by the polysomnography test apparatus according to an embodiment.

Referring to FIG. 2 , reference numeral 200 denotes an example of polysomnia test data recorded in the polysomnia test database 120 described with reference to FIG. 1 .

Specifically, according to reference numeral 200, polysomnography test data recorded in the polysomnography test database according to an embodiment is an EEG (Electroencephalogram) sensor, EOG (Electrooculography) sensor, EMG (Electromyogram) sensor, EKG (Electrokardiogramme) sensor, PPG (Photoplethysmography) of at least one user measured through at least one sensing means of a sensor, a chest belt, an Abdomen belt, a thermister, a flow sensor, and a microphone. Biometric data may be included.

For example, the user's biometric data may be data continuously measured within a preset time range.

According to one side, polysomnography test data may further include sleep scoring data for the user's biometric data.

In other words, the polysomnography data recorded in the polysomnia test database may include the user's biometric data derived through the previously performed polysomnia test and sleep scoring data for the user's biometric data.

3 is a view for explaining an example of a data conversion unit provided in the polysomnia examination apparatus according to an embodiment.

In other words, FIG. 3 is a view for explaining an example of a data conversion unit provided in the polysomnia examination apparatus according to the embodiment described with reference to FIG. 1 , and a polysomnia examination according to an embodiment of the contents described later with reference to FIG. 3 . Descriptions overlapping with those described through the device will be omitted.

Referring to FIG. 3 , the data conversion unit 300 according to an embodiment may receive polysomnography data and convert the format of the received data into a format applicable as an input of an artificial neural network.

For example, the data conversion unit 300 may receive polysomnia examination data for generating a sleep state determination model from the polysomnia examination database, and the sleep state determination model through a separate path from the polysomnia examination database 120 . Polysomnia test data for determining a polysomnia test result may be received through the

To this end, the data converting unit 300 may include a data receiving unit 310 , a data extracting unit 320 , a data dividing unit 330 , and a data labeling unit 340 .

Specifically, the data receiving unit 310 may receive polysomnography test data.

For example, the data receiving unit 310 may receive the collected polysomnography test data stored in EDF (European data format) format from the polysomnia test database in order to use the collected polysomnia test data as an input to the artificial neural network. In this case, the data stored in the EDF format may be the user's biometric data measured through at least one sensing means.

Next, the data extraction unit 320 matches the sampling frequency of each sensor corresponding to the collected polysomnography data, and extracts necessary data corresponding to the sleep state determination model from among the data with the sampling frequency matched. can do.

More specifically, polysomnography data may be stored at different sampling frequencies for each sensor as shown in Table 1 below.

Data Frequency length Chin 200 Hz 6848000 Leg-L 200 Hz 6848000 EKG 200 Hz 6848000 Leg-R 200 Hz 6848000 LOC 500 Hz 17120000 ROC 500 Hz 17120000 C3-M2 500 Hz 17120000 C4-M1 500 Hz 17120000

According to Table 1, since EEG data is stored at 500 Hz, EOG data at 200 Hz, and EMG data at 200 Hz among the polysomnography data, an operation to match the number of data is required.

To this end, the data extractor 320 may perform an operation of interpolating the EOG data and the EMG data as shown in Table 2 below.

Data Frequency length Chin 500 Hz 17120000 Leg-L 500 Hz 17120000 EKG 500 Hz 17120000 Leg-R 500 Hz 17120000 LOC 500 Hz 17120000 ROC 500 Hz 17120000 C3-M2 500 Hz 17120000 C4-M1 500 Hz 17120000

According to Table 2, the data extractor 320 may match the sampling frequencies of EEG data having different sampling frequencies, EOG data, and EMG data through interpolation.

Next, the data extracting unit 320 may extract necessary data corresponding to the sleep state determination model from among the data with the matching sampling frequency.

Meanwhile, the type of data required for generating the sleep state determination model may be determined through at least one of a determination criterion for performing sleep scoring and a user input.

According to one side, the sleep state judgment model is a sleep stage, apnea-hypopnea index (AHI), respiratory disturbance index (RDI), upper respiratory tract resistance syndrome (Respiratory effort-related arousals; RERA) ) index, total sleep time (TST), arrhythmia, and periodic limb movements of sleep index may be a model for determining at least one of the index, but the sleep state determination model according to an embodiment is It may be a model for determining various symptoms that can be derived through polysomnography without being limited to the above example.

For example, when the sleep state determination model is a model for determining a sleep phase, the data extractor 320 may include EEG data, EOG data, and EMG data that are data necessary for determining a sleep phase among data with matching sampling frequencies. can be separately extracted as necessary data.

According to one side, the data extraction unit 320 may first perform a process of extracting necessary data from polysomnography data, and then perform a process of matching the sampling frequency for each sensor corresponding to the extracted necessary data. have.

Next, the data division unit 330 may divide the extracted necessary data into time units corresponding to the sleep state determination model.

For example, when the sleep state determination model is a model for determining the sleep phase, the data division unit 330 determines the sleep phase in units of 30 seconds, so the extracted necessary data may be divided into units of 30 seconds. .

Table 3 below shows an example of dividing the extracted necessary data into time units corresponding to the sleep state determination model.

According to one side, the time unit corresponding to the sleep state determination model may be determined in advance through at least one of a determination criterion and a user input for performing sleep scoring.

Next, the labeling unit 340 may label the necessary data divided through the data dividing unit 330 with preset labels.

For example, the preset label may be a label extracted through sleep scored data.

For a more specific example, the labeling unit 340, when the sleep state determination model is a model for determining the sleep stage, W (Wake stage), N1, each of the divided necessary data is manually marked in advance by sleep scoring experts. (Sleep stage 1), N2 (sleep stage 2), N3 (sleep stage 3), and R (REM sleep stage) can be labeled with at least one label.

In other words, the labeling unit 340 may convert the format of the divided necessary data into a format applicable to an artificial neural network-based input by adding a label to the necessary data divided by time units corresponding to the sleep state determination model. .

4 is a view for explaining an example of a judgment model generator provided in the polysomnography examination apparatus according to an embodiment.

Referring to FIG. 4 , reference numeral 400 denotes an example of an artificial neural network for generating a sleep state determination model by a determination model generating unit of a polysomnia examination apparatus according to an embodiment.

The artificial neural network 400 according to an embodiment may include one input layer 410 , a plurality of hidden layers 420 , and one output layer 430 .

Here, each layer consists of a plurality of nodes, and in the order of the input layer 410 , the hidden layers 420 , and the output layer 430 , the nodes of each layer may be connected by sequential directional edges.

Specifically, the artificial neural network 400 according to an embodiment receives data through the input layer 410 , calculates the data received through the hidden layer 420 , and finally calculates the calculated data through the output layer 430 . data can be output.

For example, each node constituting the input layer 410 may be format-converted data received from the data conversion unit of the polysomnography apparatus according to an embodiment.

More specifically, the sleep state determination model generated through the artificial neural network 400 is output after the operation of the hidden layer 420 is performed when data converted from the format of the sensors for obtaining a specific index is input to the input layer 410 . A similarity score or probability associated with a specific index may be output through the layer 430 , and an index may be finally calculated based on this.

For example, when the sleep state determination model is a model for determining the sleep stage, EEG data, EMG data, and EOG data are input in 30-second increments, and finally through a deep neural network, which sleep stage is determined by the user according to the input data. It can be classified by stage.

5A to 5B are diagrams for explaining an example of outputting a polysomnia test result from a result information processing unit provided in the polysomnography test apparatus according to an exemplary embodiment.

Referring to FIGS. 5A to 5B , reference numerals 510 to 520 denote examples of outputting the polysomnography test result determined through the sleep state determination model in the form of a graph.

Specifically, according to reference numerals 510 to 520, the result information processing unit according to an embodiment may convert the polysomnography test result determined by the result determination unit into a graph format and output it.

According to one side, the result information processing unit may collect the determined polysomnography test result information and convert it to a text format and output it, but the result information processing unit can display the polysomnia test result in the graph format and text format as desired by the clinician or user. It can be converted and output in various formats.

6 is a view for explaining a polysomnia examination method according to an embodiment.

In other words, FIG. 6 is a view of a test method performed by the polysomnography test apparatus according to an embodiment described with reference to FIGS. 1 to 5B , and a polysomnia test according to an embodiment of the contents described later with reference to FIG. 6 . Descriptions overlapping with those described through the device will be omitted.

Referring to FIG. 6 , in step 610, the polysomnography test method according to an embodiment receives polysomnia test data from the data converter, and converts the format of the received data as an input to an artificial neural network. It can be converted to any applicable format.

Next, in step 620, the polysomnography test method according to an embodiment may determine the polysomnia test result by applying the polysomnia test data converted by the result determining unit to the sleep state determination model.

According to one side, in step 630, the polysomnia test according to the embodiment may convert the polysomnia test result determined by the result information processing unit into at least one format of a graph and text, and output the converted polysomnia test result to the outside.

According to one side, the polysomnia examination method according to an embodiment may further include generating a sleep state determination model by performing artificial neural network-based machine learning by inputting the polysomnia examination data converted by the determination model generation unit. have.

In addition, the polysomnography test method according to an embodiment may further include receiving the polysomnia test result determined by the judgment model generator and re-performing artificial neural network-based machine learning.

7 is a diagram for explaining an example of a method of changing the format of polysomnia test data collected in the polysomnia test method according to an embodiment.

In other words, the polysomnia examination method described with reference to FIG. 7 below may be performed through the data converter in step 610 described with reference to FIG. 6 .

Referring to FIG. 7 , in step 710 , the polysomnography test method according to an embodiment may match a sampling frequency for each sensor corresponding to polysomnia test data.

Next, in step 720 , the polysomnography test method according to an embodiment may extract necessary data corresponding to the sleep state determination model from the data in which the sampling frequency is matched.

Next, in step 730, the polysomnography examination method according to an embodiment may divide the extracted necessary data into time units corresponding to the sleep state determination model.

Meanwhile, in step 740 , the polysomnia examination method according to an embodiment may label the divided necessary data with preset labels.

After all, by using the present invention, it is possible to minimize the time required for the polysomnography test by automating the analysis of the polysomnography test result through the sleep state determination model.

In addition, by generating a sleep state judgment model through artificial neural network-based machine learning, the accuracy and reliability of polysomnography can be improved.

The device described above may be implemented as a hardware component, a software component, and/or a combination of the hardware component and the software component. For example, devices and components described in the embodiments may include, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate array (FPGA). , a programmable logic unit (PLU), a microprocessor, or any other device capable of executing and responding to instructions, may be implemented using one or more general purpose or special purpose computers. The processing device may execute an operating system (OS) and one or more software applications running on the operating system. The processing device may also access, store, manipulate, process, and generate data in response to execution of the software. For convenience of understanding, although one processing device is sometimes described as being used, one of ordinary skill in the art will recognize that the processing device includes a plurality of processing elements and/or a plurality of types of processing elements. It can be seen that can include For example, the processing device may include a plurality of processors or one processor and one controller. Other processing configurations are also possible, such as parallel processors.

The software may comprise a computer program, code, instructions, or a combination of one or more thereof, which configures a processing device to operate as desired or is independently or collectively processed You can command the device. The software and/or data may be any kind of machine, component, physical device, virtual equipment, computer storage medium or device, to be interpreted by or to provide instructions or data to the processing device. , or may be permanently or temporarily embody in a transmitted signal wave. The software may be distributed over networked computer systems, and stored or executed in a distributed manner. Software and data may be stored in one or more computer-readable recording media.

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc. alone or in combination. The program instructions recorded on the medium may be specially designed and configured for the embodiment, or may be known and available to those skilled in the art of computer software. Examples of the computer-readable recording medium include magnetic media such as hard disks, floppy disks and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic such as floppy disks. - includes magneto-optical media, and hardware devices specially configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. Examples of program instructions include not only machine language codes such as those generated by a compiler, but also high-level language codes that can be executed by a computer using an interpreter or the like. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

As described above, although the embodiments have been described with reference to the limited drawings, various modifications and variations are possible from the above description by those of ordinary skill in the art. For example, the described techniques are performed in a different order than the described method, and/or the described components of the system, structure, apparatus, circuit, etc. are combined or combined in a different form than the described method, or other components Or substituted or substituted by equivalents may achieve an appropriate result.

Therefore, other implementations, other embodiments, and equivalents to the claims are also within the scope of the following claims.

110: polysomnography inspection device 111: data conversion unit
112: judgment model generation unit 113: result determination unit
114: result information processing unit 120: polysomnia examination database

Claims (15)

a data conversion unit that receives polysomnography data and converts a format of the received polysomnography data into a format applicable as an input of an artificial neural network;
a judgment model generator for generating a sleep state judgment model by performing machine learning based on the artificial neural network by inputting the converted polysomnography test data; and
A result determination unit that determines a polysomnography test result by applying the converted polysomnography data to the sleep state determination model
including,
The judgment model generation unit,
Using the back propagation algorithm, which is an algorithm for updating the weight of the artificial neural network using the labeled data of the output layer provided in the artificial neural network, the sleep state judgment model create,
The data conversion unit
a data receiving unit for receiving the polysomnography data;
For the received polysomnography test data stored at different sampling frequencies for each sensor, an interpolation operation to match the number of data is performed to match the sampling frequency for each sensor, and the sampling frequency is matched data A data extraction unit for extracting necessary data corresponding to the sleep state determination model from among
A data division unit that divides the extracted necessary data into time units corresponding to the sleep state determination model
A polysomnography test device comprising a. According to claim 1,
The received polysomnography data is
EEG(Electroencephalogram) sensor, EOG(Electrooculography) sensor, EMG(Electromyogram) sensor, EKG(Electrokardiogramme) sensor, PPG(Photoplethysmography) sensor, Chest belt, Abdomen belt, Thermister, Including the user's biometric data measured through at least one sensing means of a flow sensor and a microphone (Microphone)
polysomnography test device. 3. The method of claim 2,
The polysomnography data are
It contains data recorded in the polysomnography database by the polysomnography performed in advance.
polysomnography test device. 4. The method of claim 3,
The received polysomnography data is
The user's biometric data further includes manual data that is scored sleep (Sleep scoring).
polysomnography test device. delete According to claim 1,
The data conversion unit
Splitting the extracted necessary data into time units corresponding to the sleep state determination model, and labeling the divided necessary data with preset labels
polysomnography test device. According to claim 1,
The sleep state judgment model is
A model for performing sleep scoring according to the polysomnography test
polysomnography test device. According to claim 1,
The sleep state judgment model is
Sleep stage, Apnea-hypopnea index (AHI), Respiratory disturbance index (RDI), Respiratory effort-related arousals (RERA) index according to the polysomnography test, A model for determining at least one of total sleep time (TST), arrhythmia, and periodic limb movements of sleep indices.
polysomnography test device. delete According to claim 1,
The judgment model generation unit
Receiving the determined polysomnography test result and re-performing the artificial neural network-based machine learning
polysomnography test device. According to claim 1,
Further comprising a result information processing unit for converting the determined polysomnography test result into at least one format of a graph and text and outputting it to the outside
polysomnography test device. receiving polysomnography data from a data conversion unit, and converting the received polysomnography data into a format applicable as an input of an artificial neural network;
generating a sleep state judgment model by performing machine learning based on the artificial neural network by inputting the converted polysomnography data in the judgment model generator; and
determining the polysomnography test result by applying the converted polysomnia test data to a sleep state judgment model in the result determining unit
including,
The step of generating the sleep state determination model comprises:
Using the back propagation algorithm, which is an algorithm for updating the weight of the artificial neural network using the labeled data of the output layer provided in the artificial neural network, the sleep state judgment model steps to create
including,
The converting step is
receiving the polysomnography data;
matching the sampling frequency for each sensor by performing an interpolation operation for matching the number of data on the received polysomnography data stored at different sampling frequencies for each sensor;
extracting necessary data corresponding to the sleep state determination model from among the data in which the sampling frequency is matched;
dividing the extracted necessary data into time units corresponding to the sleep state determination model
A polysomnography test method comprising a. 13. The method of claim 12,
The converting step is
dividing the extracted necessary data into time units corresponding to the sleep state determination model; and
Labeling the divided necessary data with preset labels (Labels)
A polysomnography test method further comprising a. delete 13. The method of claim 12,
converting the determined polysomnography test result into at least one format of a graph and text by the result information processing unit and outputting it to the outside
A polysomnography test method further comprising a.

Images