Processing method and computer program for information processing device

Abstract

To provide a method for measuring a consciousness level of an object body, using a learned artificial neural network, and a computer program.SOLUTION: A method for measuring a consciousness level of a patient comprises: a step of extracting a component of one or more frequency bands from a first section of a brain wave; a step of calculating a first index for each component of the one or more frequency bands, that is a step of calculating the first index on the basis of a degree at which, magnitude of each component of the one or more frequency bands with respect to magnitude of a prescribed standard component in the first section, exceeds a prescribed critical value; a step of calculating a probability value for one or more states of each patient, on the basis of the first index with respect to each component of the one or more frequency bands, using a learned artificial neural network; and a step of determining a consciousness level of the patient, on the basis of the probability value with respect to the calculated one or more states of each patient.SELECTED DRAWING: Figure 11

Description

[Technical Field]

[0001] The present invention relates to a method and a computer program for measuring the level of consciousness of a subject using a trained artificial neural network. [Background technology]

[0002] Generally, when a patient experiences pain during medical procedures such as surgery or treatment, anesthesia is used to block nerve transmission and eliminate or reduce the pain. Depending on the patient's symptoms and the surgical site, general anesthesia or local anesthesia may be used. In the case of general anesthesia, the patient is unable to express their own wishes, so more detailed monitoring of the patient's condition is required.

[0003] Therefore, continuous measurement of the depth of anesthesia during surgery is necessary, and methods for measuring the depth of anesthesia are broadly divided into two types: observing clinical conditions and analyzing bioelectrical signals.

[0004] One method for analyzing bioelectrical signals is to measure and analyze electroencephalograms (EEG) to assess the effects of anesthetics on the central nervous system, and prior art techniques have measured and analyzed EEG in various ways.

[0005] Currently, the most commonly used method for measuring the depth of anesthesia is the Bispectral Index (BIS) analysis method, which is characterized by quantifying the depth of anesthesia on a scale of 0 to 100.

[0006] Anesthesia depth measurement devices that use the BIS analysis method have been reported to have many problems with the accuracy of measuring a patient's anesthesia depth, and the details of the analysis algorithm installed in the equipment have not been made public, making it difficult to prove the algorithm's error.

[0007] In addition, devices for measuring the level of consciousness using the BIS analysis method have a problem in that they are slow in tracking sudden changes in the anesthesia state due to the characteristics of the method, making it difficult to accurately and quickly detect the patient's anesthesia state. Summary of the Invention [Problem to be solved by the invention]

[0008] The present invention is intended to solve the problems described above, and aims to provide accurate measurements of the depth of anesthesia even when anesthesia conditions change, and to provide information on the depth of anesthesia in a timely manner even when anesthesia or a state of consciousness changes suddenly.

[0009] Another object of the present invention is to provide information about the emotional state of a subject.

[0010] Furthermore, the present invention aims to provide a depth of anesthesia to various subjects (eg, non-human animals, etc.). [Means for solving the problem]

[0011] A method for measuring a patient's level of consciousness according to one embodiment of the present invention includes:

[0012] The method includes the steps of: extracting components of one or more frequency bands from a first section of the EEG; calculating a first index for each component of the one or more frequency bands, the first index being calculated based on the degree to which the magnitude of each component of the one or more frequency bands exceeds a predetermined critical value relative to the magnitude of a predetermined reference component in the first section; calculating a probability value for one or more states of each patient from the first index for each component of the one or more frequency bands using a trained artificial neural network; and determining the level of consciousness of the patient based on the calculated probability value for one or more states of each patient.

[0013] According to an embodiment of the present invention, a method for measuring a patient's level of consciousness may further include, before the extracting step, acquiring electroencephalograms of the patient, generating a first section of electroencephalograms including at least a portion of the acquired electroencephalograms, and removing noise from the first section of electroencephalograms. In this case, the extracting step may extract components of one or more frequency bands from the first section of electroencephalograms from which the noise has been removed.

[0014] The step of acquiring an electroencephalogram may involve acquiring an electroencephalogram of the patient sampled at a predetermined sampling frequency.

[0015] The step of generating a first section of the electroencephalogram may generate the first section of the electroencephalogram so as to include an electroencephalogram within a predetermined time interval from the time point at which the level of consciousness is determined.

[0016] The step of removing the noise includes a step of replacing a first subinterval within a first section of the EEG, the first subinterval including a point at which the magnitude of the EEG exceeds a predetermined critical size, with a second subinterval that is a different section from the first subinterval, and the first subinterval and the second subinterval may be at least a part of the first section.

[0017] The noise removal step includes a step of replacing the first section of the electroencephalogram with a second section of the electroencephalogram when a pattern in which the magnitude of the electroencephalogram in the first section exceeds a predetermined critical size corresponds to a preset pattern, and the second section may be a section different from the first section and may be at least a part of the electroencephalogram.

[0018] The extracting step can extract a first component of 0.5 to 4 Hz, a second component of 4 to 8 Hz, a third component of 8 to 16 Hz, a fourth component of 16 to 25 Hz, a fifth component of 25 to 30 Hz, a sixth component of 30 to 48 Hz, and a reference component of 0.5 to 55 Hz from the first section of the electroencephalogram.

[0019] The components of the one or more frequency bands include a magnitude of each frequency band component at one or more time points belonging to the first interval, and calculating the first index includes calculating, for each of the one or more time points, a magnitude of the component of the one or more frequency bands relative to a magnitude of the reference component, determining a time point at which the calculated magnitude exceeds the predetermined threshold as an exceedance time point, and calculating the first index based on a ratio of the number of the exceedance time points to the total number of time points belonging to the first interval, wherein the predetermined threshold may be determined based on an absolute magnitude of the reference component within the first interval.

[0020] The method for measuring a level of consciousness according to an embodiment of the present invention may further include, before the step of calculating the probability value, generating input data for the artificial neural network using a combination of first indexes for each component of the one or more frequency bands.

[0021] In accordance with an embodiment of the present invention, a method for measuring a level of consciousness includes: calculating first indices for each of N (N is a natural number) frequency bands from a first interval of a first electroencephalogram of the patient acquired through a first channel; calculating the first indices for each of the N frequency bands from a first interval of a second electroencephalogram of the patient acquired through a second channel distinct from the first channel; and generating the input data includes generating N squared pieces of first input data based on a combination of the N first indices for the first channel and the N first indices for the second channel; generating N second input data corresponding to the N first indices for the first channel; generating N third input data corresponding to the N first indices for the second channel; generating M (M is a natural number) fourth input data based on an electromyogram signal; and generating the input data including the first input data, the second input data, the third input data, and the fourth input data. In this case, N may be 6 and M may be 1.

[0022] The artificial neural network includes data reflecting EEG characteristics and data reflecting EMG characteristics, and is a neural network that has learned correlations between EEG characteristics, EMG characteristics, and patient conditions based on learning data labeled with patient condition data corresponding to the EEG characteristics and the EMG characteristics. The data reflecting EEG characteristics includes N squared pieces of first data based on a combination of N (N is a natural number) first indices for EEGs acquired through a first channel and N first indices for EEGs acquired through a second channel, N second data corresponding to the N first indices for the first channel, N third data corresponding to the N first indices for the second channel, and M (M is a natural number) fourth data based on EMG signals. The patient condition data includes K (K is a natural number) probability values ​​corresponding to each of the patient conditions.

[0023] Determining the level of consciousness includes normalizing the probability values ​​for each of the one or more patient states to calculate normalized probability values; applying a set of weights to the normalized probability values ​​corresponding to the patient state having the largest probability value among the normalized probability values; and determining the level of consciousness based on the sum of the weighted normalized probability values.

[0024] The one or more patient states include an awake state, a sedated state, a general anesthesia state, a deep anesthesia state, and a brain death state, and the applying step may apply any one of a set of weights for each of the five patient states to the normalized probability value.

[0025] The step of determining the level of awareness based on the sum of the normalized probability values ​​to which the weighted values ​​have been applied may determine the level of awareness based on a first level of awareness obtained by applying a predetermined first weight to the sum of the normalized probability values ​​to which the weighted values ​​have been applied, and a second level of awareness obtained by applying a second weight to a second level of awareness determined by a second method. [Effects of the Invention]

[0026] According to the present invention, the depth of anesthesia or state of consciousness of a patient can be measured more accurately.

[0027] In particular, the device improves upon the problems of conventional BIS analysis-based anesthesia depth measurement devices, such as a slow tracking speed and a slow response time when the level of anesthesia changes rapidly. When the patient changes from awake to hypnosis, the device responds more quickly than conventional anesthesia depth analysis devices, enabling accurate and timely understanding of the patient's condition.

[0028] Furthermore, the present invention has a simple algorithm that allows for easy real-time processing and more accurate capture of changes in conditions during anesthesia.

[0029] The present invention can also provide information about the emotional state of a subject.

[0030] The present invention can also provide a depth of anesthesia to a variety of subjects (eg, non-human animals). [Brief explanation of the drawings]

[0031] [Figure 1] 1 is a diagram illustrating a schematic configuration of a level of consciousness measurement system according to an embodiment of the present invention. [Figure 2] 10 is a diagram illustrating a method in which a control unit (120) generates a first section of an electroencephalogram (400) according to an embodiment of the present invention. [Figure 3] This is a diagram illustrating a method in which a control unit (120) according to one embodiment of the present invention removes noise from the first section (440) of an exemplary electroencephalogram and generates the first section (440FLT) of the electroencephalogram from which noise has been removed. [Figure 4]10 is a diagram illustrating a process in which a control unit (120) according to an embodiment of the present invention extracts one or more frequency band components (411, 412, 413, 414, 415, 416, 417, 418) from the first section (410FLT) of an electroencephalogram from which noise has been removed. [Figure 5] 10 is a diagram illustrating a method for a control unit (120) to calculate a first index according to an embodiment of the present invention. [Figure 6] 10 is a diagram illustrating a method for a control unit (120) to calculate a second index according to an embodiment of the present invention. [Figure 7] 5 is a diagram illustrating a process in which a control unit (120) according to an embodiment of the present invention uses a plurality of training data (510) to train an artificial neural network (520). [Figure 8] FIG. 5 is a diagram illustrating input data (531, 532, 533, 534) of an artificial neural network (520) according to one embodiment of the present invention, and the resulting output data, patient condition data (540). [Figure 9] 6 illustrates one or more exemplary sets of weights (610, 620, 630, 640, 650) in graphical form. [Figure 10] 10 is a diagram illustrating a process in which the control unit 120 determines the final depth of anesthesia for the patient 300 according to an alternative embodiment of the present invention. [Figure 11] 1 is a flow chart illustrating a method for measuring a level of consciousness performed by a user terminal (100) according to an embodiment of the present invention. BEST MODE FOR CARRYING OUT THE INVENTION

[0032] The present invention can be modified in various ways and can have various embodiments, and specific embodiments will be illustrated in the drawings and described in detail. The advantages, features, and methods of achieving the same of the present invention will become apparent from the following detailed description of the embodiments together with the drawings. However, the present invention is not limited to the embodiments disclosed below and can be embodied in various forms.

[0033] Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. When describing with reference to the drawings, the same or corresponding elements will be designated by the same reference numerals, and redundant description thereof will be omitted.

[0034] In the following examples, terms such as "first," "second," etc. are used to distinguish one component from another, not as a limitation. In the following examples, singular expressions include plural expressions unless the context clearly indicates otherwise. In the following examples, terms such as "comprise" or "have" mean the presence of a feature or component described in the specification and do not preclude the possibility that one or more other features or components may be added. In the drawings, the size of components may be exaggerated or reduced for the sake of clarity. For example, the size and shape of each component shown in the drawings are arbitrarily shown for the sake of clarity, and the present invention is not necessarily limited to those shown.

[0035] FIG. 1 is a diagram showing a schematic configuration of a level of consciousness measurement system according to an embodiment of the present invention.

[0036] A consciousness level measurement system according to an embodiment of the present invention can measure the consciousness level of a patient 300 based on biosignals of the patient 300. Here, the term "biological signals" refers to various types of signals measured directly or indirectly from the body of the patient 300, such as the patient's 300 electroencephalogram (EEG), electromyogram (EMG), and electrooculogram (EOG). The term "consciousness level" refers to the degree to which the patient 300 normally perceives and distinguishes between himself / herself and the surrounding environment, and the degree to which he / she can be awakened by a specific stimulus. In the present invention, "anesthesia depth" is used as an indicator of the consciousness level, but this is merely an example, and the concept of the present invention is not limited thereto. For example, the emotional state of the patient 300, the patient's sleep state, etc., can be used as an indicator of the consciousness level in addition to the anesthesia depth.

[0037] 1 shows a patient 300 as an example of a subject, but the concept of the present invention is not limited thereto. Therefore, the subject may be the patient 300 shown in FIG. 1, i.e., a human, or an animal.

[0038] A system for measuring a level of consciousness according to an embodiment of the present invention includes a user terminal 100 and a level of consciousness measuring device 200, as shown in FIG.

[0039] The consciousness level measuring device 200 according to one embodiment of the present invention includes a sensing unit 210 attached to the body of the patient 300 and acquiring the patient's biosignals, and a signal processing unit 220 that processes the biosignals of the patient 300 acquired by the sensing unit 210 and transmits them to the user's terminal 100.

[0040] As described above, the sensing unit 210 according to an embodiment of the present invention refers to a means attached to the body of the patient 300 to acquire the patient's biosignals. In this case, as shown in Fig. 1, the sensing unit 210 includes a reference electrode 211 for setting a reference potential, a ground electrode 212 for setting a ground potential, a first channel electrode 213 for measuring electroencephalogram and electromyogram signals, and a second channel electrode 214 for measuring electroencephalogram signals.

[0041] The plurality of electrodes 211, 212, 213, 214 are attached to the scalp of the patient 300 non-invasively or invasively by the patient 300 wearing the sensing unit 210 on his / her head, and can acquire biosignals.

[0042] 1, the shape of the sensing unit 210, the number of the electrodes 211, 212, 213, and 214 included in the sensing unit 210, and the arrangement of the electrodes 211, 212, 213, and 214 are merely examples, and the concept of the present invention is not limited thereto. Therefore, any device that is attached to the body of the patient 300 and acquires the biological signals of the patient 300 corresponds to the sensing unit 210 of the present invention.

[0043] The sensing unit 210 according to another embodiment of the present invention may be configured by omitting any one of the reference electrode 211 and the ground electrode 212. In other words, the sensing unit 210 according to another embodiment of the present invention may be configured to include the reference electrode 211, the first channel electrode 213, and the second channel electrode 214, or may be configured to include the ground electrode 212, the first channel electrode 213, and the second channel electrode 214. In this case, the reference potential and the ground potential correspond to the same potential and can be set by the reference electrode 211 or the ground electrode 212.

[0044] The signal processing unit 220 according to an embodiment of the present invention processes the biosignal acquired by the sensing unit 210 and transmits it to the user terminal 100 .

[0045] In this case, 'processing' a signal means processing the signal into a form that can be calculated by the user terminal 100. For example, this means sampling the signal or amplifying the signal.

[0046] The signal processing unit 220 according to an embodiment of the present invention may amplify the biological signal acquired by the sensing unit 210. Furthermore, the signal processing unit 220 according to an embodiment of the present invention may sample the biological signal acquired by the sensing unit 210 at a predetermined sampling frequency.

[0047] For example, the signal processing unit 220 can amplify the brain waves acquired from each of the first channel electrode 213 and the second channel electrode 214 of the sensing unit 210 at a predetermined ratio and then sample them at a sampling frequency of 250 Hz.

[0048] In addition, the signal processing unit 220 may amplify the electromyogram signal acquired by the first channel electrode 213 of the sensing unit 210 at a predetermined ratio and then sample it at a sampling frequency of 250 Hz.

[0049] The signal processor 220 according to an embodiment of the present invention can transmit the biosignal amplified and sampled according to the above-described process to the user terminal 100 via various communication methods. For example, the signal processor 220 can transmit the biosignal (amplified and sampled) to the user terminal 100 via a Bluetooth® communication method, and can transmit the biosignal to the user terminal 100 via a Wi-Fi communication method. Of course, the signal processor 220 can transmit the biosignal to the user terminal 100 using various known wired communication methods.

[0050] According to one embodiment of the present invention, another user's terminal 100 can measure the depth of anesthesia of the patient 300 based on the biological signals (amplified and sampled signals, for example, electroencephalogram, electromyogram and electrooculogram signals acquired from each of the two channels) transmitted by the signal processing unit 220.

[0051] In this case, the user terminal 100 may be a general-purpose electronic device in which an application for measuring the depth of anesthesia is installed. For example, the user terminal 100 may be a mobile phone (or tablet) in which an application for measuring the depth of anesthesia is installed. Alternatively, the user terminal 100 may be a dedicated electronic device in which only the application for measuring the depth of anesthesia is run.

[0052] 1, a user terminal 100 according to an embodiment of the present invention includes a communication unit 110, a control unit 120, a memory 130, and a display unit 140. Although not shown in the drawing, the user terminal 100 according to this embodiment further includes an input / output unit, a program storage unit, etc.

[0053] The communication unit 110 according to an embodiment of the present invention may include hardware and software necessary for the user terminal 100 to transmit and receive signals, such as control signals or data signals, via a wired or wireless connection with another device, such as the signal processing unit 220. For example, the communication unit 110 may include hardware and software for transmitting and receiving data via a Bluetooth method with the signal processing unit 220. Meanwhile, if the user terminal 100 is a general-purpose electronic device, the communication unit 110 may further include a communication modem and software for transmitting and receiving data via a general-purpose communication network (e.g., an LTE communication network, a 3G communication network, a Wi-Fi communication network, etc.).

[0054] The control unit 120 according to an embodiment of the present invention includes all types of devices capable of processing data, such as a processor. Here, the term "processor" refers to a data processing device implemented in hardware, having a physically structured circuit for performing a function expressed in accordance with code or instructions embedded in a program. Examples of such data processing devices implemented in hardware include a microprocessor, a central processing unit (CPU), a processor core, a multiprocessor, an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), etc., but the scope of the present invention is not limited thereto.

[0055] According to an embodiment of the present invention, the memory 130 temporarily or permanently stores data processed by the user terminal 100. The memory may be a magnetic storage medium or a flash storage medium, but the scope of the present invention is not limited thereto. For example, the memory 130 may temporarily and / or permanently store biosignals received from the signal processor 220. The memory 130 may also temporarily and / or permanently store weights constituting a trained artificial neural network.

[0056] The display unit 140 according to an embodiment of the present invention functions to provide the measured depth of anesthesia to the user. The display unit 140 may be implemented by various known display devices such as an LCD, an OLED, and a Micro LED. However, this is merely an example, and any device that displays a screen in response to an electrical control signal may be considered as the display unit 140 of the present invention.

[0057] The following description will focus on the method by which the control unit 120 of the user terminal 100 measures the depth of anesthesia of the patient.

[0058] The control unit 120 according to an embodiment of the present invention can acquire a biological signal of the patient 300 .

[0059] According to an embodiment of the present invention, the control unit 120 can acquire the brain waves of the patient 300. For example, the control unit 120 can acquire the brain waves by receiving the brain waves from the signal processing unit 220. In this case, the acquired brain waves may be brain waves that are amplified and sampled at a predetermined sampling frequency. In addition, the control unit 120 can acquire two brain waves acquired from different channels from the signal processing unit 220.

[0060] Furthermore, the control unit 120 according to an embodiment of the present invention can acquire an electromyogram signal in addition to the electroencephalogram signal. Of course, the acquired electromyogram signal may also be an amplified electromyogram signal sampled at a predetermined sampling frequency.

[0061] According to an embodiment of the present invention, the control unit 120 can acquire the biological signals of the patient 300 in real time. At this time, the control unit 120 can temporarily and / or permanently store the past biological signals in the memory 130 and use them to measure the depth of anesthesia of the patient.

[0062] According to an embodiment of the present invention, the control unit 120 may generate a first section of the electroencephalogram including at least a portion of the acquired electroencephalogram.

[0063] FIG. 2 is a diagram illustrating a method for generating the first section of the electroencephalogram 400 by the control unit 120 according to an embodiment of the present invention.

[0064] For ease of explanation, the following description will be given assuming that the electroencephalogram 400 of the patient 300 is the same as that shown, and that the current time point is 4 seconds, 5 seconds, and 6 seconds, depending on the case.

[0065] According to one embodiment of the present invention, when generating the first section of the electroencephalogram 400, the control unit 120 can generate the first section of the electroencephalogram so that it includes past electroencephalograms within a predetermined time interval from the time when the anesthesia depth is determined (i.e., the current time).

[0066] For example, if the current time point for determining the depth of anesthesia is assumed to be 4 seconds, the control unit 120 can generate a first section 410 of electroencephalograms including past electroencephalograms within a predetermined time interval (assumed to be 4 seconds) from the current time point (4 seconds). Also, if the current time point is 5 seconds, the control unit 120 can generate a first section 420 of electroencephalograms, and if the current time point is 6 seconds, the control unit 120 can generate a first section 430 of electroencephalograms.

[0067] In this case, the 'predetermined time interval' can be set in various ways depending on the system characteristics. For example, in a system that requires a quick response, the predetermined time interval is set relatively short. In a system that requires an accurate response, the predetermined time interval is set relatively long.

[0068] The control unit 120 according to an embodiment of the present invention may generate the first section for each of two electroencephalograms acquired through different channels according to the above-described process.

[0069] Furthermore, the control unit 120 according to an embodiment of the present invention can repeatedly generate the first section based on the current time point according to the passage of time (ie, according to a change in the time point at which the depth of anesthesia is determined). The control unit 120 according to an embodiment of the present invention can remove noise from the first section of the electroencephalogram generated according to the above-described process.

[0070] FIG. 3 is a diagram illustrating a method in which the control unit 120 according to one embodiment of the present invention removes noise from the first section 440 of an exemplary electroencephalogram and generates the first section 440FLT of the electroencephalogram from which noise has been removed.

[0071] According to one embodiment of the present invention, the control unit 120 can replace a first sub-section 444A within the first section 440 of the brain wave, in which the magnitude of the brain wave exceeds a predetermined critical size Ath, with a second sub-section 443A, which is a section different from the first sub-section.

[0072] In this case, if the magnitude of the electroencephalogram at any one time point in the first partial section 444A exceeds a predetermined threshold value Ath, the control unit 120 may determine that the magnitude of the electroencephalogram in the section 444A exceeds the predetermined threshold value Ath. In addition, the control unit 120 may substitute the section 444A by using the section 443A adjacent to the section 444A.

[0073] This allows the control unit 120 to generate a first section 440FLT of the brain wave from which noise has been removed, in which the first to third sub-sections 441B, 442B, 443B are the same as the first to third sub-sections 441A, 442A, 443A of the first section 440, and the fourth sub-section 444B is the same as the third sub-section 443A of the first section 440.

[0074] However, the length of the subinterval, the critical size Ath, and the alternative method of the subinterval are merely examples, and the concept of the present invention is not limited thereto.

[0075] In an alternative embodiment, if a pattern in which the magnitude of the brain waves in the first section 440 of the brain waves exceeds a predetermined critical size corresponds to a preset pattern, the control unit 120 may replace the first section 440 of the brain waves with a second section of the brain waves. In this case, the second section of the brain waves may be a section of the brain waves different from the first section 440 of the brain waves, for example, a section of the brain waves adjacent to the first section 440 of the brain waves.

[0076] In addition, the 'preset pattern' may be set in various ways depending on the purpose of the system. For example, the preset pattern may be a pattern in which brain waves exceeding a predetermined critical size are generated in multiple subsections within the same section, or a pattern in which brain waves exceeding a predetermined critical size are generated in two or more consecutive subsections. However, these patterns are merely examples, and the concept of the present invention is not limited thereto.

[0077] According to one embodiment of the present invention, the control unit 120 can extract components of one or more frequency bands from the first section of the electroencephalogram (or the first section of the electroencephalogram from which noise has been removed) generated according to the above-described process.

[0078] FIG. 4 is a diagram illustrating a process in which the control unit 120 according to an embodiment of the present invention extracts components 411, 412, 413, 414, 415, 416, 417, and 418 of one or more frequency bands from the first section 410FLT of the electroencephalogram from which noise has been removed.

[0079] According to one embodiment of the present invention, the control unit 120 can extract a first component 412 of 0.5 to 4 Hz, a second component 413 of 4 to 8 Hz, a third component 414 of 8 to 16 Hz, a fourth component 415 of 16 to 25 Hz, a fifth component 416 of 25 to 30 Hz, a sixth component 417 of 30 to 48 Hz, a reference component 411 of 0.5 to 55 Hz, and a BSR component 418 of 0.5 to 30 Hz from the first section 410FLT of the electroencephalogram from which noise has been removed.

[0080] According to an embodiment of the present invention, the control unit 120 may extract one or more frequency band components 411, 412, 413, 414, 415, 416, 417, and 418 from the first section 410FLT of the electroencephalogram from which noise has been removed, using frequency filters corresponding to one or more frequency bands. For example, the control unit 120 may extract the first component 412 using a bandpass filter having a passband of 0.5 to 4 Hz.

[0081] The extracted one or more frequency band components 411, 412, 413, 414, 415, 416, 417, and 418 have the same time length as the first section 410FLT of the noise-removed EEG. Also, the one or more frequency band components 411, 412, 413, 414, 415, 416, 417, and 418 may include the magnitude of each frequency band component at one or more time points belonging to the first section.

[0082] For example, if the sampling frequency of the signal processing unit 220 is 250 Hz and the length of the first section 410FLT of the noise-removed electroencephalogram is 4 seconds, the first section 410FLT of the noise-removed electroencephalogram may include 1000 time points (or 1000 sampling time points) and the magnitude of the electroencephalogram at each time point. As a result, one or more frequency band components 411, 412, 413, 414, 415, 416, 417, 418 may also include the magnitude of the component of each frequency band at each of the 1000 time points belonging to the first section.

[0083] In other words, one or more frequency band components 411, 412, 413, 414, 415, 416, 417, 418 may also include the magnitude of each frequency band component at 1000 time points belonging to the first interval.

[0084] According to an embodiment of the present invention, the control unit 120 may calculate a first index for each of the components 411, 412, 413, 414, 415, 416, 417, and 418 of one or more frequency bands.

[0085] 5 is a diagram illustrating a method for calculating a first index by the control unit 120 according to an embodiment of the present invention. For convenience of explanation, it is assumed that the graph shown in FIG. 5 is related to the first component and that the first interval for the first component includes 15 time points.

[0086] According to an embodiment of the present invention, the control unit 120 may calculate a first index based on the degree to which the magnitude of each component of one or more frequency bands relative to the magnitude of a predetermined reference component in the first interval exceeds a predetermined threshold value Rth. For example, according to an embodiment of the present invention, the control unit 120 may calculate the magnitude of components of one or more frequency bands relative to the magnitude of a reference component for one or more time points (i.e., 15 time points) belonging to the first interval.

[0087] In this case, the reference component refers to the reference component 411 extracted by the process shown in Fig. 4, and the components of one or more frequency bands refer to the components 412 to 417 extracted by the process shown in Fig. 4. For example, the control unit 120 may calculate a ratio R1 of the magnitude of the first component 412 at the first time point in the first interval to the magnitude of the reference component 411 at the first time point in the first interval. The control unit may also calculate ratios R2 to R15 of the magnitude of the first component 412 to the magnitude of the reference component 411 at the remaining time points. Of course, the control unit 120 may also calculate the ratios at each of the remaining components 413 to 417 at multiple time points using the process described above.

[0088] According to an embodiment of the present invention, the control unit 120 may determine the time point at which the calculated ratio exceeds a predetermined threshold value Rth as the exceeding time point. For example, the control unit 120 may determine the time points corresponding to R4, R12, R13, and R14 in Fig. 5 as the exceeding time points. Of course, the control unit may also determine the exceeding time points for the remaining components 413 to 417 using the same process.

[0089] According to an embodiment of the present invention, the control unit 120 may calculate a first exponent based on the ratio of the number of exceedance time points to the number of all time points belonging to the first interval. For example, the control unit 120 may calculate the first exponent for the first component 412 as 4 / 15, using 15, which is the number of all time points, and 4, which is the number of exceedance time points. In this case, the control unit 120 may normalize the calculated first exponent by multiplying the calculated first exponent by a predetermined value. Of course, the control unit may also calculate first exponents for the remaining components 413 to 417 using the same process.

[0090] Meanwhile, the control unit 120 according to an embodiment of the present invention may determine the predetermined threshold value Rth based on the absolute magnitude of the reference component within the first interval. Since the reference component is extracted from a wide frequency band, if the magnitude of the electroencephalogram within the first interval is large, the predetermined threshold value Rth may be determined to be relatively high, and if the magnitude of the electroencephalogram is small, the predetermined threshold value Rth may be determined to be relatively low.

[0091] The control unit 120 according to an embodiment of the present invention may calculate the first index for each of two electroencephalograms acquired through different channels through the above-described process. Also, the control unit 120 according to an embodiment of the present invention may repeatedly calculate the first index for the updated first interval by updating the first interval.

[0092] According to an embodiment of the present invention, the control unit 120 may calculate a second index that reflects the degree to which the electroencephalogram of the first period matches a predetermined pattern.

[0093] FIG. 6 is a diagram illustrating a method for the control unit 120 to calculate a second index according to an embodiment of the present invention.

[0094] According to an embodiment of the present invention, the control unit 120 may calculate the second index from the BSR component 418 extracted from the first section 410FLT of the electroencephalogram from which noise has been removed by the process shown in Fig. 4. In this case, the BSR component 418 may be a component of 0.5 to 30 Hz extracted from the first section 410FLT of the electroencephalogram from which noise has been removed.

[0095] According to an embodiment of the present invention, the control unit 120 may check the number of repetitions of a first type of signal (or burst signal) and a second type of signal (or suppression signal) from the BSR component 418. The first type of signal may be a type in which electroencephalograms having a magnitude (i.e., absolute value of the signal amplitude) equal to or greater than a first critical size occur during a first duration, and the second type of signal may be a type in which electroencephalograms having a magnitude equal to or less than a second critical size occur during a second duration. In this case, the first duration may be shorter than the second duration.

[0096] 6, the first type of signal may be a type in which electroencephalograms having a magnitude equal to or greater than a first critical size Ath_1_P or Ath_1_N occur during a first duration shorter than a second duration, and the second type of signal may be a type in which electroencephalograms having a magnitude equal to or less than a second critical size Ath_2_P or Ath_2_N occur during a second duration longer than the first duration. In this case, the first critical size Ath_1_P or Ath_1_N and the second critical size Ath_2_P or Ath_2_N may be preset to appropriate magnitudes. Therefore, in FIG. 6, B1, B2, and B3 may each be a first type of signal, and S1, S2, and S3 may each be a second type of signal. Meanwhile, in FIG. 6, the control unit 120 according to an embodiment of the present invention can confirm (or determine) that the number of repetitions of the first type signal and the second type signal is three.

[0097] According to an embodiment of the present invention, the control unit 120 may calculate the second index based on the ratio of the number of signal repetitions to a value corresponding to the length of the first section 410FLT of the EEG. In this case, the value corresponding to the length of the first section 410FLT of the EEG means, for example, a value proportional to the number of time points (e.g., 1,000) included in the first section 410FLT of the EEG. Therefore, the control unit 120 may determine the second index as 3 / 1,000 or a value proportional to 3 / 1,000 based on the BSR component 418 of FIG. 4.

[0098] Meanwhile, the method described in FIG. 6 is an exemplary method for determining the degree to which the electroencephalogram of the first section 410FLT matches a predetermined pattern, and the idea of ​​the present invention is not limited thereto. Any method that can determine whether a signal of a specific unit pattern is repeated in the electroencephalogram of the first section 410FLT and the number of repetitions can be used without limitation in calculating the second index in the present invention.

[0099] According to an embodiment of the present invention, the control unit 120 may calculate the third index based on at least one biological signal corresponding to the patient's condition. For example, according to an embodiment of the present invention, the control unit 120 may generate the third index based on an electromyogram signal measured by the first channel electrode 213 of the sensing unit 210.

[0100] However, using an electromyogram signal as a component for generating the third index is merely an example, and the method for calculating the third index varies depending on the configuration of the consciousness level measurement system. For example, if the consciousness level measurement system further includes a component for measuring other biosignals of the patient 300, the control unit 120 may generate the third index based on the signals measured by the component.

[0101] In an optional embodiment, the control unit 120 can generate a third index based on the patient's electrooculogram signal.

[0102] The control unit 120 according to an embodiment of the present invention can determine the depth of anesthesia of the patient 300 based on at least one of the first index, the second index, and the third index calculated by the above-described process.

[0103] According to one embodiment of the present invention, the control unit 120 can determine the depth of anesthesia of the patient 300 using a trained artificial neural network, a first index for each component of one or more frequency bands, and a third index based on the electromyogram signal.

[0104] In the present invention, the term "artificial neural network" refers to a neural network that learns the correlation between the characteristics of EEG, the characteristics of EMG, and the state of a patient based on learning data that includes data reflecting the characteristics of EEG and data reflecting the characteristics of EMG, and in which patient state data corresponding to the characteristics of EEG and the characteristics of EMG are labeled.

[0105] In other words, an artificial neural network is a neural network that has been trained to output patient condition data in response to input of data reflecting the characteristics of electroencephalograms and data reflecting the characteristics of electromyograms.

[0106] In the present invention, the artificial neural network can be realized by neural network models of various structures. For example, the artificial neural network can be realized by a CNN model, an RNN model, or an LSTM model. However, these are examples of neural network models, and any means capable of learning correlations between inputs and outputs based on training data can be used as the artificial neural network of the present invention.

[0107] FIG. 7 is a diagram illustrating a process in which the control unit 120 trains the artificial neural network 520 using a plurality of training data 510 according to an embodiment of the present invention.

[0108] As described above, the artificial neural network 520 refers to a neural network that has learned the correlation between the characteristics of the electroencephalogram, the characteristics of the electromyogram, and the state of the patient based on a plurality of training data 510 .

[0109] In this case, each of the plurality of learning data 510 may include, as described above, data reflecting the characteristics of electroencephalograms and data reflecting the characteristics of electromyograms, and may be data labeled with patient condition data corresponding to the characteristics of electroencephalograms and the characteristics of electromyograms.

[0110] For example, the first learning data 511 is data that reflects the characteristics of electroencephalograms, and may include N pieces of data 511A corresponding to first indices for each of N (N is a natural number) frequency bands calculated from the first section of the first electroencephalogram (acquired on the first channel) of patient 300, N pieces of data 511B corresponding to first indices for each of N frequency bands calculated from the first section of the second electroencephalogram (acquired on the second channel) of patient 300, and N squared pieces of data 511D generated based on a combination of the N first indices 511A for the first channel and the N first indices 511B for the second channel.

[0111] Moreover, the first learning data 511 is data that reflects the characteristics of an electromyogram, and can include M (M is a natural number) pieces of data 511C based on an electromyogram signal.

[0112] Furthermore, the first training data 511 may be labeled with patient condition data 511E including the probability that the patient 300 falls into each of the K patient conditions. In this case, N may be, for example, 6, M may be 1, and K may be 5.

[0113] The remaining training data, including the second training data 512 and the third training data 513, may also contain data similar to that of the first training data 511 and may be labeled with similar data. A specific method for generating the individual data included in the individual training data will be described later with reference to FIG. 8.

[0114] In this way, the present invention allows the artificial neural network 520 to be trained based on data generated from a patient's biological signals and training data including probabilities of each of a plurality of patient states. Thus, the artificial neural network 520 can output the probability that a patient corresponds to each of a plurality of patient states when data generated from a patient's condition signals is input. The following description is based on the assumption that the artificial neural network 520 has been trained based on the plurality of training data 510 described in FIG. 7.

[0115] FIG. 8 is a diagram illustrating input data 531, 532, 533, 534 of an artificial neural network 520 according to one embodiment of the present invention, and the resulting output data, patient condition data 540.

[0116] According to an embodiment of the present invention, the control unit 120 can generate input data 531, 532, 533, and 534 for the artificial neural network 520 to calculate a probability value that the patient 300 belongs to one or more patient states using the artificial neural network 520 trained by the above-described process. The artificial neural network 520 can include an input layer 521 including at least one input node to which the input data 531, 532, 533, and 534 is input, an intermediate layer 522 (or hidden layer) including a plurality of intermediate nodes (or hidden nodes), and an output layer 523 including at least one output node. The intermediate layer 522 can include one or more fully connected layers, as shown. If the intermediate layer 522 includes multiple layers, the artificial neural network 520 can include a function defining the relationship between each hidden layer.

[0117] For ease of explanation, the following description will be made on the assumption that the control unit 120 calculates the first index for each of the N frequency bands from the first section of the first electroencephalogram of the patient 300 acquired through the first channel through the process described in Figures 2 to 5, and that the control unit 120 calculates the first index for each of the N frequency bands from the first section of the second electroencephalogram of the patient 300 acquired through the second channel through a similar process.

[0118] Under the above-described assumptions, the control unit 120 according to an embodiment of the present invention may generate N squared first input data 533 based on a combination of N first exponents for the first channel and N first exponents for the second channel. The first input data 533 may be an item corresponding to the data 511D in FIG. 7 among the training data for training the artificial neural network 520.

[0119] For example, the control unit 120 may generate the first input data 533 to include data obtained by multiplying the first exponent of the first channel by each of the N exponents for the second channel, and data obtained by multiplying the second exponent of the first channel by each of the N exponents for the second channel. Of course, the control unit 120 may generate the first input data 533 by applying the same method as described above to the remaining exponents of the first channel.

[0120] According to an embodiment of the present invention, the control unit 120 may generate N pieces of second input data 532 corresponding to N first exponents for the first channel. Such second input data 532 may be an item corresponding to data 511A in Fig. 7 among the training data for training the artificial neural network 520. For example, the control unit 120 may generate N pieces of second input data 532 such that the N first exponents for the first channel and the N pieces of second input data 532 have the same value.

[0121] According to an embodiment of the present invention, the control unit 120 may generate N pieces of third input data 534 corresponding to the N first exponents for the second channel. Such third input data 534 may be an item corresponding to data 511B in Fig. 7 among the training data for training the artificial neural network 520. For example, the control unit 120 may generate the N pieces of third input data 534 such that the N first exponents for the second channel and the N pieces of third input data 534 have the same value.

[0122] According to an embodiment of the present invention, the control unit 120 may generate M pieces of fourth input data 531 based on the electromyogram signal. The fourth input data 531 may be an item corresponding to the data 511C in Fig. 7 among the training data for training the artificial neural network 520. For example, according to an embodiment of the present invention, the control unit 120 may generate the fourth input data 531 using the third index generated by the above-described process.

[0123] In one embodiment of the present invention, the aforementioned N may be 6 and M may be 1. Therefore, the first input data 533 may include 36 pieces of data, the second input data 532 and the third input data 534 may each include 6 pieces of data, and the fourth input data 531 may include 1 piece of data. However, the number of pieces of data included in each of the data 531 to 534 is merely an example, and the concept of the present invention is not limited to these examples.

[0124] According to one embodiment of the present invention, the control unit 120 inputs the input data 531, 532, 533, and 534 generated by the above-described process into the artificial neural network 520, and can obtain patient status data 540 including a probability value that the patient 300 belongs to one or more of each patient status.

[0125] According to an embodiment of the present invention, the patient status data 540 may include a probability value of the patient 300 being in each of K (K is a natural number) patient statuses. For example, K may be 5, and the one or more patient statuses may include an awake status (status 1), a sedation status (status 2), a general anesthesia status (status 3), a hyper or deep anesthesia status (status 4), and a brain death status (status 5).

[0126] For example, the control unit 120 can acquire patient status data 540 such as [0.81, 0.62, 0.34, 0.17, 0.01] from the artificial neural network 520. Such patient status data 540 means that there is an 81% probability that the patient 300 is in an awake state, a 62% probability that the patient 300 is in a sedated state, a 34% probability that the patient 300 is in a general anesthesia state, a 17% probability that the patient 300 is in a hyper or deep anesthesia state, and a 1% probability that the patient 300 is in a brain death state.

[0127] In another embodiment of the present invention, the K states and / or numbers may be set in various ways. For example, if the present invention is used to determine the emotional state of a patient, each of the K states may be a state indicating the patient's emotion. However, this is merely an example, and the concept of the present invention is not limited thereto.

[0128] According to one embodiment of the present invention, the control unit 120 can determine the depth of anesthesia for the patient 300 based on the probability values ​​for one or more of the patient's states.

[0129] More specifically, according to an embodiment of the present invention, the control unit 120 may calculate a normalized probability value by normalizing the probability value for each of one or more patient conditions included in the patient condition data 540. For example, the control unit 120 may calculate a normalized probability value for each of one or more patient conditions such that the sum of the probability values ​​for each of one or more patient conditions becomes 1.

[0130] Furthermore, the control unit 120 according to an embodiment of the present invention may apply a weight set corresponding to the patient condition having the largest normalized probability value to the normalized probability value.

[0131] 9 is a graphical representation of one or more exemplary weight sets 610, 620, 630, 640, and 650. A first weight set 610 corresponds to an awake state (status 1), a second weight set 620 corresponds to a sedation state (status 2), a third weight set 630 corresponds to a general anesthesia state (status 3), a fourth weight set 640 corresponds to a hyper or deep anesthesia state (status 4), and a fifth weight set 650 corresponds to a brain death state (status 5).

[0132] For example, if the normalized probability value calculated by the above process is [0.45, 0.25, 0.15, 0.1, 0.05], the probability that the patient 300 is in an awake state (status 1) is the highest, so the control unit 120 can apply the weight set 610 corresponding to the awake state (status 1) to the normalized probability value and calculate a probability such as [0.45, 0.2, 0.5, 0.01, 0]. The control unit 120 according to an embodiment of the present invention can determine the depth of anesthesia of the patient 300 based on the sum of the normalized probability values ​​to which the weighted values ​​are applied.

[0133] In an alternative embodiment, the control unit 120 according to an embodiment of the present invention can determine the final depth of anesthesia of the patient 300 by referring to the depth of anesthesia of the patient 300 determined by another method.

[0134] FIG. 10 is a diagram illustrating a process in which the control unit 120 determines the final depth of anesthesia for the patient 300 according to an alternative embodiment of the present invention.

[0135] The control unit 120 according to an alternative embodiment of the present invention may determine the final depth of anesthesia DOA_F of the patient 300 based on the sum of a first depth of anesthesia obtained by applying a predetermined weight W1 to the depth of anesthesia DOA_1 calculated by the process described with reference to Figures 2 to 9 and a second depth of anesthesia obtained by applying a predetermined weight W2 to the depth of anesthesia DOA_2 calculated by the second method. Here, the second method for calculating the depth of anesthesia DOA_2 refers to a method for calculating the depth of anesthesia that has at least a partial process different from the process described with reference to Figures 2 to 9. For example, the second depth of anesthesia may be determined based on a first index calculated from the patient's electroencephalogram, a second index reflecting the degree to which the patient's electroencephalogram matches a predetermined pattern, and a third index based on at least one biological signal corresponding to the patient's condition.

[0136] 11 is a flowchart illustrating a method for measuring the depth of anesthesia performed by the user terminal 100 according to one embodiment of the present invention. In the following, explanations that overlap with the contents explained in FIGS. 1 to 10 will be omitted, and explanations will be given with reference to FIGS. 1 to 10. The user terminal 100 according to an embodiment of the present invention can train an artificial neural network based on a plurality of training data (S910).

[0137] In the present invention, the term "artificial neural network" refers to a neural network that learns the correlation between the characteristics of EEG, the characteristics of EMG, and the state of a patient based on learning data that includes data reflecting the characteristics of EEG and data reflecting the characteristics of EMG, and in which patient state data corresponding to the characteristics of EEG and the characteristics of EMG are labeled.

[0138] In other words, an artificial neural network is a neural network that has been trained to output patient condition data in response to input of data reflecting the characteristics of electroencephalograms and data reflecting the characteristics of electromyograms.

[0139] In the present invention, the artificial neural network can be realized by neural network models of various structures. For example, the artificial neural network can be realized by a CNN model, an RNN model, or an LSTM model. However, these neural network models are merely examples, and any means capable of learning the correlation between input and output based on training data can be used as the artificial neural network of the present invention.

[0140] FIG. 7 is a diagram illustrating a process in which a user terminal 100 according to an embodiment of the present invention trains an artificial neural network 520 using a plurality of training data 510.

[0141] As described above, the artificial neural network 520 refers to a neural network that has learned the correlation between the characteristics of the electroencephalogram, the characteristics of the electromyogram, and the state of the patient based on a plurality of training data 510 .

[0142] In this case, each of the plurality of learning data 510 may include, as described above, data reflecting the characteristics of electroencephalograms and data reflecting the characteristics of electromyograms, and may be data labeled with patient condition data corresponding to the characteristics of electroencephalograms and the characteristics of electromyograms.

[0143] For example, the first learning data 511 is data that reflects the characteristics of electroencephalograms, and may include N pieces of data 511A corresponding to first indices for each of N (N is a natural number) frequency bands calculated from the first section of the first electroencephalogram (acquired on the first channel) of patient 300, N pieces of data 511B corresponding to first indices for each of N frequency bands calculated from the first section of the second electroencephalogram (acquired on the second channel) of patient 300, and N squared pieces of data 511D generated based on a combination of the N first indices 511A for the first channel and the N first indices 511B for the second channel.

[0144] Moreover, the first learning data 511 is data that reflects the characteristics of an electromyogram, and can include M (M is a natural number) pieces of data 511C based on an electromyogram signal.

[0145] Furthermore, the first training data 511 may be labeled with patient condition data 511E including the probability that the patient 300 falls into each of the K patient conditions. In this case, N may be, for example, 6, M may be 1, and K may be 5.

[0146] The remaining training data, including the second training data 512 and the third training data 513, may also contain data similar to that of the first training data 511 and may be labeled with similar data.

[0147] In this way, the present invention allows the artificial neural network 520 to be trained based on data generated from a patient's biosignals and training data including probabilities of each of a plurality of patient states. Thus, the artificial neural network 520 can output the probability that a patient corresponds to each of a plurality of patient states when data generated from a patient's condition signals is input. The following description is based on the assumption that the artificial neural network 520 has been trained based on the plurality of training data 510 described in FIG. 7.

[0148] The user terminal 100 according to an embodiment of the present invention can acquire a biological signal of the patient 300 (S920).

[0149] The user terminal 100 according to an embodiment of the present invention can acquire brain waves of the patient 300. For example, the user terminal 100 can acquire brain waves by receiving brain waves from the signal processing unit 220 described above. In this case, the acquired brain waves may be brain waves that are amplified and sampled at a predetermined sampling frequency. In addition, the user terminal 100 can acquire two brain waves acquired from different channels from the signal processing unit 220.

[0150] Meanwhile, the user terminal 100 according to an embodiment of the present invention can further acquire an electromyogram signal together with an electroencephalogram. Of course, the acquired electromyogram signal may also be an amplified electromyogram signal sampled at a predetermined sampling frequency.

[0151] The user terminal 100 according to an embodiment of the present invention can acquire the biological signals of the patient 300 in real time. At this time, the user terminal 100 can temporarily and / or permanently store the past biological signals in the memory 130 and use them to measure the depth of anesthesia of the patient.

[0152] The user terminal 100 according to an embodiment of the present invention may generate a first section of the electroencephalogram including at least a portion of the acquired electroencephalogram (S930).

[0153] FIG. 2 is a diagram for explaining a method in which the terminal 100 of the other user generates the first section of the electroencephalogram 400 in one embodiment of the present invention.

[0154] In the following, for the sake of convenience, the electroencephalogram 400 of the patient 300 is shown, and the explanation will be given assuming that the current time point is 4 seconds, 5 seconds, and 6 seconds, depending on the case.

[0155] In one embodiment of the present invention, the user terminal 100 can generate the first section of the electroencephalogram 400 so that it includes past electroencephalograms within a predetermined time interval from the time when the anesthesia depth is determined (i.e., the current time).

[0156] For example, if the current time point for determining the depth of anesthesia is assumed to be 4 seconds, the user terminal 100 can generate a first electroencephalogram section 410 including past electroencephalograms within a predetermined time interval (assumed to be 4 seconds) from the current time point (4 seconds). Furthermore, if the current time point is 5 seconds, the user terminal 100 can generate a first electroencephalogram section 420, and if the current time point is 6 seconds, the user terminal 100 can generate a first electroencephalogram section 430.

[0157] In this case, the 'predetermined time interval' can be set in various ways depending on the system characteristics. For example, in a system that requires a quick response, the predetermined time interval can be set relatively short. In a system that requires an accurate response, the predetermined time interval can be set relatively long.

[0158] The user terminal 100 according to an embodiment of the present invention can generate the first section for each of two electroencephalograms acquired through different channels through the above-described process. Also, the user terminal 100 according to an embodiment of the present invention can repeatedly generate the first section based on the current time point according to the flow of time (i.e., according to a change in the time point at which the anesthesia depth is determined).

[0159] The user terminal 100 according to an embodiment of the present invention can remove noise from the first section of the electroencephalogram generated through the above-described process (S940).

[0160] FIG. 3 is a diagram illustrating a method in which a user terminal 100 according to one embodiment of the present invention removes noise from an exemplary first section 440 of an electroencephalogram and generates a noise-removed first section 440FLT of an electroencephalogram.

[0161] According to one embodiment of the present invention, a user terminal 100 can replace a first sub-section 444A within a first section 440 of an electroencephalogram, in which the magnitude of the electroencephalogram exceeds a predetermined critical size Ath, with a second sub-section 443A, which is a section different from the first sub-section.

[0162] In this case, if the magnitude of the electroencephalogram at any one time point in the first partial section 444A exceeds a predetermined threshold size Ath, the user terminal 100 may determine that the magnitude of the electroencephalogram in the corresponding section 444A exceeds the predetermined threshold size Ath. Also, the user terminal 100 may substitute the corresponding section 444A by using the section 443A adjacent to the corresponding section 444A.

[0163] As a result, the user's terminal 100 can generate the first section 440FLT of the electroencephalogram from which noise has been removed, in which the first to third sub-sections 441B, 442B, 443B are identical to the first to third sub-sections 441A, 442A, 443A of the first section 440, and the fourth sub-section 444B is identical to the third sub-section 443A of the first section 440.

[0164] However, the length of the subinterval, the critical size Ath, and the subinterval alternative method are merely examples, and the concept of the present invention is not limited thereto.

[0165] In an alternative embodiment, if a pattern in which the magnitude of the brain waves in the first section 440 of the brain waves exceeds a predetermined critical size corresponds to a preset pattern, the user terminal 100 may replace the first section 440 of the brain waves with a second section of the brain waves. In this case, the second section of the brain waves may be a section of the brain waves different from the first section 440 of the brain waves, for example, a section of the brain waves adjacent to the first section 440 of the brain waves.

[0166] In addition, the 'preset pattern' can be set in various ways depending on the purpose of the system. For example, the preset pattern may be a pattern in which brain waves exceeding a predetermined critical size are generated in multiple subsections within the same section, or a pattern in which brain waves exceeding a predetermined critical size are generated in two or more consecutive subsections. However, these patterns are merely examples, and the concept of the present invention is not limited thereto.

[0167] According to one embodiment of the present invention, the user terminal 100 can extract components of one or more frequency bands from the first section of the electroencephalogram (or the first section of the electroencephalogram from which noise has been removed) generated by the above-mentioned process (S950).

[0168] FIG. 4 is a diagram illustrating a process in which the user terminal 100 according to an embodiment of the present invention extracts components 411, 412, 413, 414, 415, 416, 417, and 418 of one or more frequency bands from the first section 410FLT of the electroencephalogram from which noise has been removed.

[0169] A user terminal 100 according to one embodiment of the present invention can extract a first component 412 of 0.5 to 4 Hz, a second component 413 of 4 to 8 Hz, a third component 414 of 8 to 16 Hz, a fourth component 415 of 16 to 25 Hz, a fifth component 416 of 25 to 30 Hz, a sixth component 417 of 30 to 48 Hz, a reference component 411 of 0.5 to 55 Hz, and a BSR component 418 of 0.5 to 30 Hz from a first section 410FLT of an electroencephalogram from which noise has been removed.

[0170] According to an embodiment of the present invention, the user terminal 100 can extract one or more frequency band components 411, 412, 413, 414, 415, 416, 417, and 418 from the first section 410FLT of the electroencephalogram from which noise has been removed using frequency filters corresponding to one or more frequency bands. For example, the user terminal 100 can extract the first component 412 using a band-pass filter having a passband of 0.5 to 4 Hz.

[0171] The extracted one or more frequency band components 411, 412, 413, 414, 415, 416, 417, and 418 may have the same time length as the first section 410FLT of the noise-removed EEG. Also, the one or more frequency band components 411, 412, 413, 414, 415, 416, 417, and 418 may include the magnitude of each frequency band component at one or more time points belonging to the first section.

[0172] For example, if the sampling frequency of the signal processing unit 220 is 250 Hz and the length of the first section 410FLT of the noise-removed electroencephalogram is 4 seconds, the first section 410FLT of the noise-removed electroencephalogram may include 1,000 time points (or 1,000 sampling time points) and the magnitude of the electroencephalogram at each time point. Thus, one or more frequency band components 411, 412, 413, 414, 415, 416, 417, and 418 may also include the magnitude of the component of each frequency band at each of the 1,000 time points belonging to the first section.

[0173] In other words, one or more frequency band components 411, 412, 413, 414, 415, 416, 417, 418 may also include the magnitude of each frequency band component at 1000 time points belonging to the first interval.

[0174] According to an embodiment of the present invention, the user terminal 100 may calculate a first index for each of the components 411, 412, 413, 414, 415, 416, 417, and 418 of one or more frequency bands (S960).

[0175] 5 is a diagram illustrating a method for calculating a first index by a user terminal 100 according to an embodiment of the present invention. For convenience of explanation, it is assumed that the graph shown in FIG. 5 is related to the first component and that the first interval for the first component includes 15 time points.

[0176] The user terminal 100 according to an embodiment of the present invention may calculate a first index based on the degree to which the magnitude of each component of one or more frequency bands relative to the magnitude of a predetermined reference component in the first interval exceeds a predetermined threshold value Rth. For example, the user terminal 100 according to an embodiment of the present invention may calculate the magnitude of components of one or more frequency bands relative to the magnitude of a reference component for one or more time points (i.e., 15 time points) belonging to the first interval.

[0177] In this case, the reference component refers to the reference component 411 extracted by the process described in FIG. 4, and the components of one or more frequency bands refer to the components 412 to 417 extracted by the process described in FIG. 4. For example, the user terminal 100 may calculate a ratio R1 of the magnitude of the first component 412 at the first time point in the first interval to the magnitude of the reference component 411 at the first time point in the first interval. In addition, the control unit may calculate ratios R2 to R15 of the magnitude of the first component 412 to the magnitude of the reference component 411 at the remaining time points. Of course, the user terminal 100 may also calculate the ratios at multiple time points for the remaining components 413 to 417 by the process described above.

[0178] According to an embodiment of the present invention, the user terminal 100 may determine the time point at which the calculated ratio exceeds a predetermined threshold value Rth as the exceeding time point. For example, the user terminal 100 may determine the time points corresponding to R4, R12, R13, and R14 in Fig. 5 as the exceeding time points. Of course, the control unit may also determine the exceeding time points for the remaining components 413 to 417 using the same process.

[0179] According to an embodiment of the present invention, the user terminal 100 may calculate a first exponent based on the ratio of the number of overage time points to the number of all time points belonging to the first interval. For example, the user terminal 100 may calculate the first exponent as 4 / 15 for the first component 412 using 15, which is the number of all time points, and 4, which is the number of overage time points. In this case, the user terminal 100 may normalize the calculated first exponent by multiplying the calculated first exponent by a predetermined value. Of course, the control unit may also calculate first exponents for the remaining components 413 to 417 using the same process.

[0180] Meanwhile, the user terminal 100 according to an embodiment of the present invention may determine the predetermined threshold value Rth based on the absolute magnitude of the reference component within the first period. Since the reference component is extracted from a wide frequency band, if the magnitude of the EEG within the first period is large, the predetermined threshold value Rth may be determined to be relatively high, and if the magnitude of the EEG is small, the predetermined threshold value Rth may be determined to be relatively low.

[0181] The user terminal 100 according to an embodiment of the present invention can calculate the first index for each of two electroencephalograms acquired through different channels using the above-described process. Also, the user terminal 100 according to an embodiment of the present invention can repeatedly calculate the first index for the updated first interval by updating the first interval.

[0182] According to one embodiment of the present invention, the user terminal 100 can generate input data 531, 532, 533, and 534 for the artificial neural network 520 to calculate the probability value that the patient 300 belongs to one or more patient states using the artificial neural network 520 trained by the above-mentioned process (S970).

[0183] FIG. 8 is a diagram illustrating input data 531, 532, 533, 534 of an artificial neural network 520 according to one embodiment of the present invention, and the resulting output data, patient condition data 540.

[0184] For ease of explanation, the following description will be made on the assumption that the user terminal 100 calculates the first index for each of the N frequency bands from the first section of the first electroencephalogram of the patient 300 acquired through the first channel according to the process described in Figures 2 to 5, and that the user terminal 100 calculates the second index for each of the N frequency bands from the first section of the second electroencephalogram of the patient 300 acquired through the second channel according to the same process.

[0185] Under the above assumptions, the user terminal 100 according to an embodiment of the present invention can generate N squared first input data 533 based on a combination of N first exponents for the first channel and N first exponents for the second channel. Such first input data 533 may be an item corresponding to data 511D in FIG. 7 among the training data for training the artificial neural network 520.

[0186] For example, the user terminal 100 may generate the first input data 533 to include data obtained by multiplying the first exponent of the first channel by each of the N exponents for the second channel, and data obtained by multiplying the second exponent of the first channel by each of the N exponents for the second channel. Of course, the user terminal 100 may generate the first input data 533 by applying the same method as described above to the remaining exponents of the first channel.

[0187] According to an embodiment of the present invention, the user terminal 100 may generate N pieces of second input data 532 corresponding to N first exponents for the first channel. Such second input data 532 may be an item corresponding to data 511A in Fig. 7 among the training data for training the artificial neural network 520. For example, the user terminal 100 may generate N pieces of second input data 532 such that the N first exponents for the first channel and the N pieces of second input data 532 have the same value.

[0188] According to an embodiment of the present invention, the user terminal 100 may generate N pieces of third input data 534 corresponding to the N first exponents for the second channel. Such third input data 534 may be an item corresponding to data 511B in Fig. 7 among the training data for training the artificial neural network 520. For example, the user terminal 100 may generate N pieces of third input data 534 such that the N first exponents for the second channel and the N pieces of third input data 534 have the same value.

[0189] The user terminal 100 according to an embodiment of the present invention may generate M pieces of fourth input data 531 based on the electromyogram signal. The fourth input data 531 may be an item corresponding to data 511C in Fig. 7 among the training data for training the artificial neural network 520. For example, the user terminal 100 according to an embodiment of the present invention may generate the fourth input data 531 using the third index generated by the above-described process.

[0190] In one embodiment of the present invention, the aforementioned N may be 6 and M may be 1. Therefore, the first input data 533 includes 36 pieces of data, the second input data 532 and the third input data 534 each include 6 pieces of data, and the fourth input data 531 includes 1 piece of data. However, the number of pieces of data included in each of the data 531 to 534 is merely an example, and the concept of the present invention is not limited to these examples.

[0191] According to one embodiment of the present invention, the user terminal 100 inputs the input data 531, 532, 533, 534 generated by the above-described process into the artificial neural network 520, and can obtain patient condition data 540 including a probability value that the patient 300 belongs to one or more patient conditions (S980).

[0192] According to an embodiment of the present invention, the patient status data 540 may include a probability value of the patient 300 being in each of K (K is a natural number) patient statuses. For example, K may be 5, and the one or more patient statuses may include an awake status (status 1), a sedation status (status 2), a general anesthesia status (status 3), a hyper or deep anesthesia status (status 4), and a brain death status (status 5).

[0193] For example, the user terminal 100 can acquire patient status data 540 such as [0.81, 0.62, 0.34, 0.17, 0.01] from the artificial neural network 520. Such patient status data 540 means that there is an 81% probability that the patient 300 is in an awake state, a 62% probability that the patient 300 is in a sedated state, a 34% probability that the patient 300 is in a general anesthesia state, a 17% probability that the patient 300 is in a hyper or deep anesthesia state, and a 1% probability that the patient 300 is in a brain death state.

[0194] In another embodiment of the present invention, the K states and / or the number of states may be set in various ways. For example, when the present invention is used to determine the emotional state of a patient, each of the K states may be a state indicating the patient's emotion. However, this is merely an example, and the concept of the present invention is not limited thereto.

[0195] According to one embodiment of the present invention, the user terminal 100 can determine the depth of anesthesia for the patient 300 based on the probability values ​​for one or more of the patient's conditions (S990).

[0196] More specifically, the user terminal 100 according to an embodiment of the present invention may calculate a normalized probability value by normalizing the probability value for each of one or more patient states included in the patient state data 540. For example, the user terminal 100 may calculate a normalized probability value for each of one or more patient states such that the sum of the probability values ​​for each of one or more patient states is 1.

[0197] Furthermore, the user terminal 100 according to an embodiment of the present invention may apply a weight set corresponding to the patient condition having the largest normalized probability value to the normalized probability value.

[0198] 9 is a graphical representation of one or more exemplary weight sets 610, 620, 630, 640, and 650. A first weight set 610 corresponds to an awake state (status 1), a second weight set 620 corresponds to a sedation state (status 2), a third weight set 630 corresponds to a general anesthesia state (status 3), a fourth weight set 640 corresponds to a hyper or deep anesthesia state (status 4), and a fifth weight set 650 corresponds to a brain death state (status 5).

[0199] For example, if the normalized probability value calculated by the above process is [0.45, 0.25, 0.15, 0.1, 0.05], the probability that the patient 300 is in an awake state (status 1) is the highest, so the user terminal 100 can apply the weighting value set 610 corresponding to the awake state (status 1) to the normalized probability value and calculate a probability such as [0.45, 0.2, 0.5, 0.01, 0].

[0200] The user terminal 100 according to an embodiment of the present invention can determine the depth of anesthesia of the patient 300 based on the sum of the normalized probability values ​​to which the weighted values ​​are applied.

[0201] In an alternative embodiment, the user terminal 100 according to an embodiment of the present invention can determine the final depth of anesthesia for the patient 300 by referring to the depth of anesthesia for the patient 300 determined by another method.

[0202] FIG. 10 is a diagram illustrating a process in which the user terminal 100 determines the final depth of anesthesia for the patient 300 according to an alternative embodiment of the present invention.

[0203] The user terminal 100 according to an alternative embodiment of the present invention can determine the final depth of anesthesia DOA_F of the patient 300 based on the sum of a first depth of anesthesia obtained by applying a predetermined weight W1 to the depth of anesthesia DOA_1 calculated by the process described in Figures 2 to 9 and a second depth of anesthesia obtained by applying a predetermined weight W2 to the depth of anesthesia DOA_2 calculated by the second method. In this case, the second method for calculating the depth of anesthesia DOA_2 refers to a method for calculating the depth of anesthesia that has at least a partial process different from the process described in Figures 2 to 9. For example, the second depth of anesthesia may be determined based on a first index calculated from the patient's electroencephalogram, a second index reflecting the degree to which the patient's electroencephalogram matches a predetermined pattern, and a third index based on at least one biological signal corresponding to the patient's condition.

[0204] The above-described embodiments of the present invention may be embodied in the form of a computer program that can be executed by various components on a computer, and such a computer program may be recorded on a computer-readable medium. In this case, the medium may store the computer-executable program. Examples of the medium include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical recording media such as CD-ROMs and DVDs, magneto-optical media such as floptical disks, ROMs, RAMs, flash memories, etc., which may be configured to store program instructions.

[0205] On the other hand, the computer program may be specially designed and constructed for the purposes of the present invention, or it may be one that is well known and available to those skilled in the art of computer software. Examples of computer programs include not only machine code, such as produced by a compiler, but also high-level language code that can be executed by a computer using an interpreter, etc.

[0206] The specific implementations described in the present invention are merely examples and are not intended to limit the scope of the present invention in any way. For the sake of brevity, descriptions of conventional electronic configurations, control systems, software, and other functional aspects of the systems may be omitted. Wire connections or connecting members between components shown in the drawings are illustrative of functional connections and / or physical or circuit connections, and may be replaced or illustrated as various additional functional connections, physical connections, or circuit connections in an actual device. Furthermore, unless specifically referred to as "essential" or "important," a component may not necessarily be required for application of the present invention.

[0207] Therefore, the spirit of the present invention is not limited to the above-described embodiments, and not only the scope of the claims described below, but also all scopes equivalent to or modified equivalently to the claims belong to the scope of the spirit of the present invention. [Explanation of symbols]

[0208] 100: User's device 110: Communications Department 120: Control unit 130: Memory 140: Display section 200: Consciousness level measuring device 210: Sensing part 220: Signal processing unit 211~214: Electrode 300:Patient

Claims

1. A processing method for an information processing device, comprising: extracting components of one or more frequency bands from a first section of an electroencephalogram by the information processing device; a step of calculating a first index for each component of the one or more frequency bands by the information processing device, the first index being calculated based on a degree to which a magnitude of each component of the one or more frequency bands relative to a magnitude of a predetermined reference component in the first section exceeds a predetermined critical value; calculating, by the information processing device, a probability value for each of one or more patient states from a first index for each component of one or more frequency bands using the trained artificial neural network; and determining, by the information processing device, a level of consciousness of the patient based on the calculated probability values ​​for each of the one or more states of the patient; The extracting step includes: The processing method of an information processing device, wherein the information processing device extracts a first component of 0.5 to 4 Hz, a second component of 4 to 8 Hz, a third component of 8 to 16 Hz, a fourth component of 16 to 25 Hz, a fifth component of 25 to 30 Hz, a sixth component of 30 to 48 Hz, and a reference component of 0.5 to 55 Hz from a first section of the electroencephalogram.

2. The processing method of the information processing device, prior to the extracting step, The information processing device acquires an electroencephalogram of the patient; generating a first section of electroencephalograms including at least a portion of the acquired electroencephalograms by the information processing device; a step of removing noise in a first section of the electroencephalogram by the information processing device; further comprising The processing method of claim 1 , wherein the extracting step comprises extracting components of one or more frequency bands from the first interval of the electroencephalogram from which the noise has been removed.

3. The step of acquiring an electroencephalogram comprises:

3. The processing method according to claim 2, wherein the information processing device acquires the brain waves of the patient sampled at a predetermined sampling frequency.

4. The step of generating a first section of the electroencephalogram comprises: The processing method of claim 2 , wherein the information processing device generates the first section of the electroencephalogram so as to include an electroencephalogram within a predetermined time interval from the time point at which the level of consciousness is determined.

5. The step of removing noise includes: the information processing device includes a step of replacing a first partial interval, which includes a time point in the first interval of the electroencephalogram where the magnitude of the electroencephalogram exceeds a predetermined critical size, with a second partial interval which is an interval different from the first partial interval, The processing method of claim 2 , wherein the first partial interval and the second partial interval are at least a part of the first interval.

6. The step of removing noise includes: If a pattern in which the magnitude of the electroencephalogram in the first section of the electroencephalogram exceeds a predetermined critical size corresponds to a preset pattern, the information processing device includes a step of replacing the first section of the electroencephalogram with the second section of the electroencephalogram, The processing method of claim 2 , wherein the second interval is a interval different from the first interval and is at least a part of the electroencephalogram.

7. the components of the one or more frequency bands include a magnitude of the component of each frequency band at one or more time points belonging to the first section; The step of calculating the first index includes: a step of calculating, by the information processing device, a magnitude of a component of the one or more frequency bands relative to a magnitude of the reference component for each of the one or more time points; a step of determining, by the information processing device, a time point at which the calculated magnitude exceeds the predetermined critical value as an exceedance time point; The information processing device calculates the first index based on a ratio of the number of exceedance points to the number of all points belonging to the first interval; The processing method of claim 1 , further comprising:

8. The processing method of claim 7 , wherein the predetermined threshold value is determined by the information processing device based on an absolute magnitude of the reference component within the first interval.

9. The processing method of the information processing device comprises: before the step of calculating the probability value, The method of claim 1 , further comprising: generating input data for the artificial neural network using a first exponent for each component of the one or more frequency bands.

10. The processing method of the information processing device comprises: the information processing device calculates a first index for each of N (N is a natural number) frequency bands from a first section of a first electroencephalogram of the patient acquired through a first channel; the information processing device calculates the first index for each of the N frequency bands from a first section of a second electroencephalogram of the patient acquired through a second channel distinct from the first channel; The step of generating input data includes: generating N squared pieces of first input data based on a combination of N first exponents for the first channel and N first exponents for the second channel; generating N pieces of second input data corresponding to N first exponents for the first channel by the information processing device; generating N pieces of third input data corresponding to the N first exponents for the second channel by the information processing device; generating M (M is a natural number) pieces of fourth input data based on an electromyogram signal by the information processing device; generating the input data including the first input data, the second input data, the third input data, and the fourth input data by the information processing device; The processing method of claim 9 , further comprising:

11. 11. The processing method of the information processing apparatus according to claim 10, wherein N is 6 and M is 1.

12. The artificial neural network The data includes data reflecting characteristics of electroencephalograms and data reflecting characteristics of electromyograms, a neural network that learns correlations between the characteristics of the electroencephalogram, the characteristics of the electromyogram, and the state of the patient based on learning data in which state data of the patient corresponding to the characteristics of the electroencephalogram and the characteristics of the electromyogram are labeled; The data reflecting the characteristics of the brain waves is N (N is a natural number) first indices for the electroencephalograms acquired through the first channel and N squared first data based on a combination of N first indices for the electroencephalograms acquired through the second channel; N second data corresponding to the N first indexes for the first channel; N third data corresponding to the N first indexes for the second channel; and M (M is a natural number) pieces of fourth data based on electromyogram signals; The patient condition data includes:

2. The processing method of claim 1, wherein the information processing device includes K (K is a natural number) probability values ​​corresponding to each of the patient's states.

13. The step of determining the level of consciousness includes: the information processing device normalizing the probability value for each of the one or more patient states to calculate a normalized probability value; applying a set of weights to the normalized probability values ​​corresponding to the patient condition having the largest probability value; determining the level of consciousness based on a sum of normalized probability values ​​to which the set of weights has been applied; The processing method of claim 1 , further comprising:

14. The one or more patient conditions include: Includes awake state, sedation state, general anesthesia state, deep anesthesia state, and brain death state, The applying step comprises:

14. The processing method of claim 13, wherein the information processing device applies one of a set of weights for each of the five patient conditions to the normalized probability value.

15. determining the level of consciousness based on a sum of normalized probability values ​​to which the set of weights has been applied, The processing method of claim 13, wherein the information processing device determines the awareness level based on the sum of a first awareness level to which a predetermined first weighting value is applied and a second awareness level to which a second weighting value is applied to a second awareness level determined by a second method.

16. Using a computer, A computer program stored on a medium for carrying out the method according to any one of claims 1 to 15.

Images