WO2024101388A1 - 情報処理方法、プログラム、及び情報処理装置 - Google Patents

情報処理方法、プログラム、及び情報処理装置 Download PDF

Info

Publication number
WO2024101388A1
WO2024101388A1 PCT/JP2023/040207 JP2023040207W WO2024101388A1 WO 2024101388 A1 WO2024101388 A1 WO 2024101388A1 JP 2023040207 W JP2023040207 W JP 2023040207W WO 2024101388 A1 WO2024101388 A1 WO 2024101388A1
Authority
WO
WIPO (PCT)
Prior art keywords
sedation
information processing
user
depth
electroencephalogram
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/JP2023/040207
Other languages
English (en)
French (fr)
Japanese (ja)
Inventor
拓也 茨木
友規 矢野
雄介 依田
弘憲 砂川
崇 渡邊
学 橋本
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Vie Inc
National Cancer Center Japan
National Cancer Center Korea
Original Assignee
Vie Inc
National Cancer Center Japan
National Cancer Center Korea
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Vie Inc, National Cancer Center Japan, National Cancer Center Korea filed Critical Vie Inc
Priority to JP2024557827A priority Critical patent/JPWO2024101388A1/ja
Publication of WO2024101388A1 publication Critical patent/WO2024101388A1/ja
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/16—Devices for psychotechnics; Testing reaction times ; Devices for evaluating the psychological state
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/25—Bioelectric electrodes therefor
    • A61B5/251—Means for maintaining electrode contact with the body
    • A61B5/256—Wearable electrodes, e.g. having straps or bands
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/316—Modalities, i.e. specific diagnostic methods
    • A61B5/369—Electroencephalography [EEG]
    • A61B5/372—Analysis of electroencephalograms
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61M—DEVICES FOR INTRODUCING MEDIA INTO, OR ONTO, THE BODY; DEVICES FOR TRANSDUCING BODY MEDIA OR FOR TAKING MEDIA FROM THE BODY; DEVICES FOR PRODUCING OR ENDING SLEEP OR STUPOR
    • A61M5/00—Devices for bringing media into the body in a subcutaneous, intra-vascular or intramuscular way; Accessories therefor, e.g. filling or cleaning devices, arm-rests
    • A61M5/14—Infusion devices, e.g. infusing by gravity; Blood infusion; Accessories therefor
    • A61M5/168—Means for controlling media flow to the body or for metering media to the body, e.g. drip meters, counters ; Monitoring media flow to the body
    • A61M5/172—Means for controlling media flow to the body or for metering media to the body, e.g. drip meters, counters ; Monitoring media flow to the body electrical or electronic

Definitions

  • the present invention relates to an information processing method, a program, and an information processing device.
  • the Bispectral Index (BIS) technology which monitors electroencephalograms, has been known as a method for determining a patient's state of sedation during anesthesia.
  • BIS Bispectral Index
  • a technology that applies electrical stimulation to a living body determines the state of analgesia from the body's reaction, and outputs an index related to the patient's sedation
  • Patent Document 1 There is also a technology that uses machine learning to estimate the sedation level from electroencephalogram signals (see, for example, Patent Document 2).
  • Patent Document 2 also uses a sensor that has at least six electrodes and is attached to the skin on the forehead (e.g., Masimo's SedLine (registered trademark)), which raises concerns about the discomfort and ease of wearing it.
  • Masimo's SedLine registered trademark
  • one aspect of the disclosed technology aims to provide an information processing method, program, and information processing device that enable a simple and appropriate sedation monitoring method.
  • the information processing method involves an information processing device sequentially acquiring EEG signals from an EEG measuring device worn on a user's ear, estimating the sedation depth for the EEG signals acquired in the sequence using the sequentially acquired EEG signals and a learning model that has learned learning data including the EEG signals and the sedation depth for the EEG signals, and outputting sedation depth information related to the estimated sedation depth.
  • One aspect of the disclosed technology makes it possible to provide a simple and appropriate sedation monitoring method.
  • FIG. 1 is a diagram illustrating an example of an overview of an information processing system according to an embodiment.
  • FIG. 1 is a diagram illustrating an example of an earphone set according to an embodiment. 1 is a diagram showing an example of a schematic cross section of an earphone according to an embodiment.
  • FIG. FIG. 1 is a diagram illustrating an application example of an information processing system 1 according to an embodiment.
  • FIG. 1 shows the scores of the sedation scale RASS.
  • FIG. 13 is a diagram showing an example of data of subject A in this experiment.
  • FIG. 13 shows the classification accuracy of subject A.
  • FIG. 13 is a diagram showing the relationship between RASS and BIS throughout the measurement period in subject A and the estimated values of the estimation model.
  • FIG. 13 is a diagram showing the relationship between three electroencephalogram signals and sedation depth.
  • FIG. 13 is a diagram showing the accuracy of the estimated values of the estimation model for each subject.
  • FIG. 13 shows the results of a t-test using cross-correlation coefficients.
  • FIG. 2 is a block diagram showing an example of a server according to an embodiment.
  • FIG. 2 is a block diagram illustrating an example of a processing terminal according to the embodiment.
  • 13 is a flowchart illustrating an example of an estimation model generation process of the server according to the embodiment.
  • 13 is a flowchart illustrating an example of a process for estimating a sedation depth by the processing terminal according to the embodiment.
  • 13 is a flowchart illustrating an example of a learning model generation process of the processing terminal according to the embodiment.
  • FIG. 4 is a diagram illustrating an example of a display screen according to the embodiment.
  • a user who is to measure an electroencephalogram wears an earphone set 10 having a bioelectrode provided in an ear canal as an electroencephalogram measuring device.
  • a neck-worn earphone set 10 is used, but any earphones may be used as long as they are capable of sensing EEG signals from the ear canal.
  • earphone sets that obtain a reference signal from the earlobe, earphones that obtain a reference signal or earth signal from other positions (other positions in the ear canal), completely wireless earphones, etc. can be used.
  • the EEG measuring device is not limited to the earphone type, and may be, for example, a headphone type, with a bioelectrode provided in the earmuff portion.
  • the earphone set 10 acquires an EEG signal from the ear canal and transmits the EEG signal to the information processing device 30 or the information processing device 50 via the network N.
  • the network N includes a wired or wireless network, and may use short-range wireless communication such as Bluetooth (registered trademark).
  • the earphone set 10 may also perform a predetermined process on the EEG signal and transmit it to an information processing device 30 that acts as a server or an information processing device 50 used by the user.
  • the predetermined process includes at least one of the following processes: amplification, sampling, filtering, and differential calculation.
  • the information processing device 30 is, for example, a server, which acquires EEG signals measured by an EEG measuring device and executes various processes. For example, the information processing device 30 extracts feature data from the EEG signals of a user (e.g., a patient) receiving an anesthetic, performs machine learning using learning data in which this feature data is labeled with the user's sedation depth, and generates an estimation model that estimates the sedation depth. This makes it possible to generate an estimation model that acquires EEG signals and estimates the sedation depth while reducing the burden on the user by using a simple device.
  • the information processing device 30 may also transmit the generated estimation model to the information processing device 50.
  • the information processing device 30 may also be configured so that the information processing device 50 first acquires an EEG signal, and then acquires the EEG signal via a specified application.
  • the information processing device 30 may also calculate feature data based on the EEG signals acquired in sequence, input the feature data into the estimation model, and transmit sedation depth information including the estimated sedation depth to the information processing device 50.
  • the information processing device 50 is, for example, a processing terminal such as a mobile terminal held by a user, and sequentially acquires EEG signals from the earphone set 10.
  • the information processing device 50 may input the sequentially acquired EEG signals into an estimation model to acquire sedation depth information.
  • the information processing device 50 may also output the acquired EEG signals or feature data to the information processing device 30, and acquire sedation depth information from the information processing device 30.
  • the information processing device 50 may output an alarm, an estimated dose of anesthesia, a waveform of the sedation depth, etc., based on the estimated sedation depth to a display, or may output sound data, etc., to a sound output device (e.g., earphone set 10). This makes it possible to output information regarding the sedation depth estimated from the EEG signal, making it possible to inform other users (e.g., doctors, etc.) of the patient's sedation state.
  • a sound output device e.g., earphone set 10
  • ⁇ Earphone set composition> 2 and 3 will explain an overview of the earphone set 10 in the embodiment.
  • the earphone set 10 is not limited to the example shown in Fig. 2 and 3, and any earphones can be applied to the technology of the present disclosure as long as they are capable of sensing brain waves from the ear canal and outputting them to an external device.
  • FIG. 2 is a diagram showing an example of an earphone set 10 according to an embodiment.
  • the earphone set 10 shown in FIG. 2 has a pair of earphones 100R, 100L and a neck strap 110.
  • Each of the earphones 100R, 100L is connected to the neck strap 110 using a cable capable of signal communication, but may also be connected using wireless communication.
  • RL will be omitted when there is no need to distinguish between left and right.
  • the neck strap 110 has a central member that fits along the back of the neck, and rod-shaped members (arms) 112R, 112L that are curved along both sides of the neck. Electrodes 122, 124 for sensing EEG signals are provided on the surface of the central member that comes into contact with the neck on the back side.
  • the electrodes 122, 124 are an electrode that is connected to earth and a reference electrode, respectively. This allows the electrodes to be spaced apart from the elastic electrodes provided on the ear tips of the earphones, as described below, making it possible to acquire EEG signals with high accuracy.
  • the neck strap 110 may also have a processing unit that processes EEG signals and a communication device that communicates with the outside, but these processing units and communication units may be provided in the earphones 100.
  • the rod-shaped members 112R, 112L on both sides of the neck strap 110 are heavier at their tips than at their bases (the central member side), which allows the electrodes 122, 124 to be properly pressed against the wearer's neck.
  • weights are provided at the tips of the rod-shaped members 112R, 112L. Note that the positions of the electrodes 122, 124 are not limited to these positions.
  • FIG. 3 is a diagram showing an example of a schematic cross section of an earphone 100R according to an embodiment.
  • the earphone 100R shown in FIG. 3 may have an elastic member (e.g., urethane) 108 between the speaker 102 and the nozzle 104, for example.
  • an elastic member e.g., urethane
  • vibrations from the speaker 102 are less likely to be transmitted to the elastic electrode of the ear tip 106, preventing sound interference between the elastic electrode of the ear tip 106 and the speaker 102.
  • the ear tip 106 including the elastic electrode is located at the sound guide port, but the elasticity of the elastic electrode itself makes it possible to prevent interference from sound vibrations. Also, by using an elastic material for the housing, this elastic material makes it difficult for sound vibrations to be transmitted to the elastic electrode of the ear tip 106, making it possible to prevent interference from sound vibrations.
  • the earphones 100 may include an audio sound processor, which may be used to cut sound signals below a certain frequency (e.g., 50 Hz) that corresponds to an EEG signal.
  • the audio sound processor cuts sound signals below 30 Hz, a frequency band that is likely to produce characteristic EEG signals, but may also amplify sound signals with frequencies around 70 Hz in order to avoid compromising the bass sound.
  • the audio sound processor can be configured to cut off certain frequencies only when brainwave signals are being sensed.
  • the ear tip 106 also transmits the brainwave signal sensed from the ear canal to the contact of an electrode provided in the nozzle 104.
  • the brainwave signal is transmitted from the ear tip 106 via the contact to a biosensor (not shown) inside the earphone 100.
  • the biosensor sequentially outputs the acquired brainwave signal via a cable to a processing device provided in the neck strap 110 or transmits it to an external device.
  • the ear tip 106 may also be insulated from the housing containing the biosensor and audio sound processor.
  • Fig. 4 is a diagram showing an application example of the information processing system 1 according to the embodiment.
  • the earphone set 10 is worn by a patient, and a drug (e.g., an anesthetic) is administered during surgery. During this time, the user's brainwave signal is acquired via the electrodes of the earphone set 10.
  • a drug e.g., an anesthetic
  • the acquired EEG signal is subjected to, for example, filtering, feature calculation, and noise removal, and the sedation depth is estimated from the EEG signal after noise removal.
  • the estimated sedation depth may be output to a medical device or the like via a user interface.
  • machine learning is performed based on the user's EEG signal and the sedation depth linked to this EEG signal. This allows an estimation model to be constructed that estimates the sedation depth from the EEG signal.
  • a Nihon Kohden BIS (Bispectral Index) monitoring system the heart rate (HR), pulse rate (PR), respiratory rate (RR), arterial blood oxygen saturation (SpO2) measured using a pulse oximeter, BIS, and SQIBIS (BIS signal quality value) are obtained.
  • HR heart rate
  • PR pulse rate
  • RR respiratory rate
  • SpO2 arterial blood oxygen saturation
  • BIS blood oxygen saturation
  • SQIBIS BIS signal quality value
  • brainwave signals are obtained from the ear canals of both ears at a sampling rate of 600 Hz.
  • the reference electrode and the earth electrode are the back of the neck (electrodes 122, 124 in Fig. 2).
  • a total of three signals are used: the signal from the left or right ear, and the differential signal between the left and right ears.
  • the differential signal is used as a signal from which noise commonly mixed into the left and right electrodes has been removed.
  • RASS (described later using FIG. 5) is evaluated manually at intervals of 1 to 10 minutes. Data on medication, etc. are also entered from the system and recorded. In this experiment, for RASS, since it was not possible to speak to the subject or compress the chest, physical stimuli were applied by tapping or pinching the subject's feet, and the subject's reaction was recorded. Other events such as body movements were also recorded manually. BIS data and EEG data were synchronized with the recording time of a personal computer (PC).
  • PC personal computer
  • a 1-40 Hz bandpass filter is applied to all three signals.
  • Data is extracted from each electrode of the earphone set 10 in a 4 second time window with 1 second sliding.
  • the power (Welch's power spectral density) from 1 to 40 Hz is calculated in 0.5 Hz increments. Any signal whose average absolute amplitude value of each window exceeds 50 ⁇ V (threshold) is removed as noise.
  • the power of each frequency band is standardized within the subject and a principal component analysis is performed. Data of 1 to 100 dimensions is extracted and further standardized.
  • FIG. 5 is a diagram showing the scores of the sedation scale RASS (Richmond Agitation-Sedation Scale). As shown in FIG. 5, the RASS is expressed by a score ranging from ⁇ 5 to +4. In this experiment, in order to generate teacher data for a sedation depth estimation model (hereinafter also referred to as an “estimation model”), the RASS scores are classified into the teacher labels shown below. In this experiment, labeling is performed with moderate sedation as the boundary.
  • an index that increases as the patient becomes more awake such as the RASS
  • moderate sedation or higher means “moderate sedation” to “combative”, and less than moderate sedation means “deep sedation” and “unaroused”.
  • Teacher label 1 Indicates a state close to awakening, with a RASS score of -3 or more (moderate sedation or more).
  • Teacher label 0 Indicates deep sedation, with a RASS score of -4 or less (less than moderate sedation).
  • Metric 1 Accuracy of binary classification in test data (equal positive and negative)
  • FIG. 6 is a diagram showing an example of data of subject A in this experiment.
  • FIG. 6(A) is a diagram showing the original waveform of subject A measured by the earphone set 10.
  • the vertical axis represents the amplitude value (unit: ⁇ V), and the horizontal axis represents the measurement time (unit: minutes).
  • FIG. 6(A) shows the original waveforms of three signals indicating the right, left, and left-right differences. As shown in the example in FIG. 6(A), it can be seen that noise is introduced every time an electric scalpel is used during surgery.
  • Figure 6 (B) shows the spectral power (Z) of the differential signal.
  • the vertical axis represents the spectral power value (unit: Hz), and the horizontal axis represents the measurement time.
  • the Z value is expressed as 1 to 40 Hz.
  • Figure 6 (B) when an electric scalpel is used, it is identified as noise and is not used as learning data.
  • Figure 6 (C) is a diagram showing an example of the relationship between RASS and additional medication.
  • the left vertical axis represents the RASS score
  • the right vertical axis represents the amount of medication administered
  • the horizontal axis represents the measurement time.
  • an anesthetic drug was administered at times T1 to T5
  • an antagonist was administered at time T6.
  • Times T2, T4, and T5 are when RASS indicates -4
  • T4 and T5 are when body movement was observed.
  • FIG. 6(D) is a diagram showing an example of the relationship between BIS and vital information.
  • the left vertical axis represents the BIS score
  • the right vertical axis represents SQIBIS (unit: %)
  • the horizontal axis represents the measurement time.
  • BIS, HR, PR, RR, SpO2, and SQIBIS are measured and displayed on a single graph.
  • (Classification accuracy of subject A) 7 is a diagram showing the classification accuracy of subject A.
  • a value estimated from an EEG signal using an estimation model using the above-mentioned sparse modeling LASSO is compared with an actual RASS observed manually.
  • RASS-3 or higher is classified as moderate sedation or higher and RASS-4 or lower is classified as less than moderate sedation using the EEG signal of subject A
  • the classification accuracy is 94.5312%. This shows that the EEG signal acquired by the earphone set 10 is at a practical level.
  • FIG. 8 is a diagram showing the relationship between RASS and BIS over the entire measurement period in subject A and the estimated values of the estimation model.
  • the left vertical axis in FIG. 8 represents the RASS value
  • the right vertical axis represents the BIS value
  • the horizontal axis represents the measurement time.
  • the various conditions and the results of the cross-correlation coefficient are as follows. (Various conditions) For the value of RASS, the nearest neighbor value in the EEG time window of interest is taken. For estimates, there is a binary classification (0 or 1), but for ease of comparison, 0-(estimate) x 5 is used. A moving average filter is applied to the EEG signal using data from the previous 10 seconds.
  • Cross-correlation coefficient results Cross-correlation coefficient between RASS and estimated value: 0.86378
  • Cross-correlation coefficient between RASS and BIS 0.50357
  • Cross-correlation coefficient between the estimate and the BIS 0.58321
  • FIG. 8 it can be seen that there is a high correlation between the actual value of RASS and the estimated value of RASS.
  • Figure 9 shows the relationship between the three EEG signals and the depth of sedation.
  • the vertical axis of the graph shown in Figure 9 indicates the EEG power value (Z) obtained by subtracting data below RASS-4 from data above RASS-3, and the horizontal axis indicates the EEG frequency.
  • Z EEG power value
  • the differential signal is most likely to have power differences at each frequency, and that the higher the frequency, the more likely the power differences are to occur.
  • Fig. 10 is a diagram showing the accuracy of the estimated values of the estimation model for each subject. As described above, the accuracy of the estimation model for subject A shown in Fig. 10 was 94.53%, and the accuracy for the other nine subjects was between 66.53% and 92.07%, with the average value being 81.68%.
  • the first average value of the cross-correlation coefficient between the RASS and the estimated value is 0.67
  • the second average value of the cross-correlation coefficient between the BIS and the estimated value is 0.44
  • the third average value of the cross-correlation coefficient between the RASS and the BIS is 0.44.
  • FIG. 11 is a diagram showing the results of a t study using cross-correlation coefficients.
  • the left side shows the cross-correlation coefficient between RASS and the estimated value
  • the right side shows the cross-correlation coefficient between RASS and BIS.
  • ⁇ Server configuration example> 12 is a block diagram showing an example of an information processing device 30 according to an embodiment.
  • the information processing device 30 is, for example, a server, and may be composed of one or more devices.
  • the information processing device 30 processes an electroencephalogram signal or electroencephalogram information, and performs an estimation process of the sedation depth from the electroencephalogram signal using, for example, a learning function of artificial intelligence (AI).
  • AI artificial intelligence
  • the information processing device 30 is also referred to as a server 30. Note that the information processing device 30 does not necessarily have to be a server, and may be a general-purpose computer.
  • the server 30 includes one or more processors (CPU: Central Processing Unit) 310, one or more network communication interfaces 320, memory 330, a user interface 350, and one or more communication buses 370 for interconnecting these components.
  • processors CPU: Central Processing Unit
  • Server 30 may, for example, optionally include a user interface 350, which may include a display device (not shown) and a keyboard and/or mouse (or some other input device, such as a pointing device, not shown).
  • a user interface 350 may include a display device (not shown) and a keyboard and/or mouse (or some other input device, such as a pointing device, not shown).
  • Memory 330 may be, for example, a high-speed random access memory such as a DRAM, SRAM, DDR RAM, or other random access solid-state storage device, or may be a non-volatile memory such as one or more magnetic disk storage devices, optical disk storage devices, flash memory devices, or other non-volatile solid-state storage devices. Memory 330 may also be a non-transitory computer-readable recording medium having a program recorded thereon.
  • memory 330 may be one or more storage devices located remotely from processor 310.
  • memory 330 stores the following programs, modules and data structures, or a subset thereof:
  • the one or more processors 310 read and execute programs from the memory 330 as necessary.
  • the one or more processors 310 may configure an EEG control unit 312, an acquisition unit 313, a learning unit 314, a generation unit 315, and an output unit 316 by executing programs stored in the memory 330.
  • the EEG control unit 312 controls and processes the EEG signals acquired in sequence, and controls each of the following processes.
  • the acquisition unit 313 acquires an EEG signal measured by an EEG measuring device, for example, a bioelectrode included in the earphone set 10. For example, the acquisition unit 313 sequentially acquires EEG signals measured in the user's ear canal from an EEG measuring device worn in the user's ear. Note that the EEG measuring device is not limited to the earphone set 10.
  • the EEG signal may be, for example, an EEG signal indicating characteristic data (power vector for each frequency per unit time) of the target user's in-ear EEG.
  • the learning unit 314 performs supervised learning by inputting the EEG signals acquired in sequence into learning data including the EEG signal and the sedation depth for this EEG signal.
  • the EEG signal may be at least one of a brain half signal measured by the right ear, an EEG signal measured by the left ear, and a left-right differential signal.
  • a signal calculated from left and right signals, such as a differential signal may be used.
  • the depth of sedation may be annotated to the EEG signal and, for example, assigned to the EEG signal as a teacher label related to the level of RASS.
  • the depth of sedation may be annotated to the EEG signal and, for example, assigned to the EEG signal as a teacher label related to the level of RASS.
  • EEG signals having teacher labels of sedation depth may be used as training data.
  • a training model for example, a generalized linear model sparse modeling LASSO is used, but this is merely one example, and a training model that solves other classification problems or regression problems may also be used.
  • the generation unit 315 receives the learning data input by the learning unit 314 and generates the learned learning model as an estimated model. For example, when a learning model trained from the learning data is evaluated using test data, if the evaluation value is equal to or greater than a threshold, the generation unit 315 may use the learning model as the estimated model.
  • the output unit 316 outputs the estimation model generated by the generation unit 315.
  • the output unit 316 outputs the estimation model to another information processing device 50, such as a medical device or a processing terminal used by a user (e.g., a doctor, etc.), a personal computer, a smartphone, or a tablet terminal.
  • the other information processing device 50 estimates the sedation depth of a user wearing an EEG measuring device using the estimation model that estimates the sedation depth from an EEG signal.
  • the above processing enables the information processing device 30 to generate a sedation depth estimation model using the learning data measured by the target user. Furthermore, the information processing device 30 can generate a more general-purpose sedation depth estimation model by learning the learning data of a large number of target users.
  • Example of processing terminal configuration> 13 is a block diagram showing an example of an information processing device 50 according to an embodiment.
  • the information processing device 50 includes, for example, a medical device, a mobile terminal (such as a smartphone), a computer, a tablet terminal, etc.
  • the information processing device 50 is also referred to as a processing terminal 50.
  • the processing terminal 50 includes one or more processors (e.g., CPUs) 510, one or more network communication interfaces 520, memory 530, a user interface 550, and one or more communication buses 590 for interconnecting these components.
  • processors e.g., CPUs
  • the user interface 550 includes a display 551 and an input device (such as a keyboard and/or a mouse or some other pointing device) 552.
  • the user interface 550 may also be a touch panel.
  • Memory 530 may be, for example, a high-speed random access memory such as a DRAM, SRAM, DDR RAM, or other random access solid-state storage device, or may be a non-volatile memory such as one or more magnetic disk storage devices, optical disk storage devices, flash memory devices, or other non-volatile solid-state storage devices. Memory 530 may also be a non-transitory computer-readable recording medium that records a program.
  • a high-speed random access memory such as a DRAM, SRAM, DDR RAM, or other random access solid-state storage device
  • Memory 530 may also be a non-transitory computer-readable recording medium that records a program.
  • memory 530 may be one or more storage devices located remotely from processor 510.
  • memory 530 stores the following programs, modules and data structures, or a subset thereof:
  • the one or more processors 510 read and execute a program from the memory 530 as necessary.
  • the one or more processors 510 may configure an application control unit (hereinafter also referred to as an "app control unit") 512 by executing a program stored in the memory 530.
  • the app control unit 512 is an application that processes EEG signals, and has, for example, an acquisition unit 513, an estimation unit 514, an output unit 515, a selection unit 516, and a generation unit 517.
  • the acquisition unit 513 sequentially acquires the EEG signals measured by the EEG measuring device attached to the ear of a specific user.
  • the acquisition unit 513 may acquire the EEG signals from the earphone set 10 via the network communication interface 520, or may acquire the EEG signals that have been subjected to specific processing via the network communication interface 520.
  • the specific processing includes, for example, frequency conversion, filtering, and differential calculation of the EEG signals.
  • the acquisition unit 513 may also acquire feature data related to the EEG signals (power vector for each frequency per unit time).
  • the acquisition unit 513 may also perform a predetermined process on the EEG signal acquired at a predetermined sampling rate.
  • the predetermined process may include bandpass filtering in a predetermined range (e.g., 1 to 40 Hz), setting a time window of a predetermined time (e.g., 4 seconds), and setting a movement time of the time window of a predetermined time (e.g., 1 second).
  • the estimation unit 514 estimates the sedation depth for the EEG signals acquired in sequence by the acquisition unit 513, using a learning model that has learned learning data including the EEG signals and the sedation depth for these EEG signals.
  • the learning model may be an estimation model generated by the server 30, an estimation model generated by the generation unit 517, or an estimation model customized by the estimation unit 514.
  • the estimation unit 514 inputs the sequentially acquired EEG signals into an estimation model generated by the server 30, and outputs an estimated value of the sedation depth.
  • the sequentially acquired EEG signals may be EEG signals after a predetermined process.
  • the estimation unit 514 may also estimate the sedation depth using an estimation model generated for an individual by the generation unit 517 described later.
  • the output unit 515 outputs sedation depth information regarding the sedation depth estimated by the estimation unit 514.
  • the output unit 515 outputs sedation depth information including an estimated value of the sedation depth to a display device (display 551 or an external display device).
  • the output unit 515 may output an alarm to a sound output device (e.g., a speaker) when the estimated value satisfies a predetermined condition regarding an alarm.
  • the predetermined condition regarding an alarm includes the estimated value exceeding a threshold value.
  • the above process makes it possible to appropriately estimate the depth of sedation based on EEG signals while reducing the burden on the user by using a simple EEG measuring device that can acquire EEG signals from the ear. Furthermore, while previous research has looked at the depth of anesthesia for anesthetic drugs such as general anesthesia, there is no technology to estimate the depth of moderate sedation in endoscopic procedures, for example. In contrast, the technology disclosed herein makes it possible to appropriately estimate the depth of moderate sedation.
  • BIS performs analysis over a time range of one minute and is therefore unable to respond to sudden changes in sedation depth
  • the disclosed technology estimates sedation depth from EEG data over a recent specified time window (e.g., four seconds), allowing for rapid estimation and the ability to respond to sudden changes in sedation depth.
  • the disclosed technology uses earphone set 10 for a wearable simplified EEG meter and ear canal electrodes, making it easy to wear like regular earphones and to easily obtain EEG signals, reducing the burden on the user.
  • the EEG measuring device 50 includes both the functions of an earphone for EEG measurement and an earphone for sound output, the information processing device 50 may have the following processing.
  • the selection unit 516 selects sound data based on the estimated sedation depth. For example, when the estimated sedation depth falls below a threshold, the selection unit 516 may select one sound from a playlist that includes one or more sounds (music, binaural beats) that are generally considered to have a high sedative effect. For example, it is known that delivering audio stimuli of different frequencies from the left and right ears, known as binaural beats, to the brain enhances brain waves in the differential frequency band (Schmid, W. et al. Brainwave entrainment to minimize sedative drug doses in paediatric surgery: a randomised controlled trial. Br. J. Anaesth. 125, 330-335 (2020)).
  • the output unit 515 may output the sound data selected by the selection unit 516 to the earphone set 10.
  • the output unit 515 outputs the selected sound data to the binaural earphones 100 or the monoaural earphone 100L or right.
  • the above process makes it possible to support a calming effect by utilizing the sound output function of the brainwave measuring device.
  • the sound data to be output may also include sound data specified according to the user whose brainwaves are being measured. For example, the user whose brainwaves are being measured may be asked to select in advance soothing, healing music, or favorite music, and a playlist may be created.
  • the selection unit 516 selects sound data from the playlist for each user.
  • the above process makes it possible to select sound data from a playlist appropriate for the user undergoing EEG measurement, and to support the sedative effect with the user's preferred sound data.
  • the acquisition unit 513 may also include sequentially acquiring electroencephalogram signals when an anesthetic is administered to the user.
  • the output unit 515 may include outputting information regarding the dosage of the anesthetic based on the depth of sedation estimated by the estimation unit 514.
  • the information regarding the dosage of the anesthetic agent may include the dosage to be administered to the user, or a difference between the current dosage and the dosage to be administered.
  • the estimation unit 514 may determine the dosage of the drug to be administered to the user according to the estimated sedation depth by referring to a correspondence table between the sedation depth and the dosage to be administered. The correspondence table may be stored in the memory 530 for each drug. Furthermore, if the information processing device 50 can obtain the user's current dosage from another device, the estimation unit 514 may determine the difference between the current dosage and the dosage to be administered.
  • the above process makes it possible to output to a display device the next dosage to be administered to the user and the timing of administration according to the estimated level of sedation. As a result, doctors and others can determine the dosage and timing of administration of the drug based on objective data.
  • the generation unit 517 executes the process of generating a learning model based on an EEG signal acquired within a predetermined time period from the user whose EEG is being measured, and the user's sedation depth in response to this EEG signal. For example, the generation unit 517 may generate an estimation model suitable for this user by inputting learning data including an EEG signal acquired within a predetermined time period and teacher labels annotated to this EEG signal into the learning model.
  • the estimation unit 514 may input EEG signals acquired in sequence after a predetermined time has elapsed into the estimation model to estimate the sedation depth of the user. For example, by using the estimation model generated for each user, the estimation unit 514 becomes able to estimate the sedation depth according to the characteristics of the EEG data of that user.
  • the generating unit 517 may determine the level of the user's awake state within the depth of sedation based on an EEG signal acquired from the user before the predetermined time and before the anesthetic is administered. For example, since the power of the EEG signal in the awake state may differ for each user, the generating unit 517 may set a standard for the EEG signal in the awake state of the user (teacher label for the awake state) using the EEG signal for a predetermined time before the anesthetic is administered.
  • the generation unit 517 may set a standard for the wakefulness state using an EEG signal for a first predetermined time (e.g., 3 minutes), and generate an estimation model that estimates the sedation depth of the individual using learning data generated by annotating an EEG signal for a second predetermined time (e.g., 3 minutes) after anesthesia is administered after the first predetermined time.
  • a first predetermined time e.g. 3 minutes
  • an estimation model that estimates the sedation depth of the individual using learning data generated by annotating an EEG signal for a second predetermined time (e.g., 3 minutes) after anesthesia is administered after the first predetermined time.
  • the above processing makes it possible to generate an estimation model that estimates the sedation depth from the EEG signal according to the EEG state of each individual. As a result, it becomes possible to more appropriately estimate the sedation depth according to the characteristics of the EEG state of the EEG target user.
  • the estimation unit 514 may also remove EEG signals equal to or greater than a threshold value from among the EEG signals acquired in sequence. For example, the estimation unit 514 or acquisition unit 513 removes EEG signals (e.g., amplitude values) equal to or greater than a threshold value from values to be input to the estimation model in order to remove EEG signals containing noise. In the experiment described above, a threshold value of 50 ⁇ V was used.
  • the above processing makes it possible to remove EEG signals containing noise caused by procedures such as electrosurgical surgery, improving the reliability of the estimated sedation depth.
  • the application control unit 512 when the application control unit 512 is able to acquire body movement and vital information, it may associate this information with the EEG signal and store it in the memory 530. This makes it possible to investigate the relationship between the EEG signal, the sedation depth, and vital information.
  • an example has been described in which an individual estimation model is generated using a first predetermined time for determining the standard of the wakefulness state and a second predetermined time for generating learning data, but as the number of EEG measurement users increases, a generalized estimation model may be generated.
  • a generalized standard of the wakefulness state is set using the EEG signal for each user at the first predetermined time, and a generalized estimation model can be generated using this standard of the wakefulness state and learning data generated at the second predetermined time.
  • a generalized estimation model e.g., an estimation model generated by the server 30
  • no learning time is required, so it can be used from the start of treatment.
  • the application control unit 512 stores the administered drug in association with the EEG signal, making it possible to estimate the sedation depth for each drug. For example, it becomes possible to estimate what the future sedation depth will be depending on which drug and how much is administered. Specifically, the application control unit 512 associates an estimation model with each drug, and when the drug to be administered is determined, the estimation unit 514 performs estimation processing using the estimation model associated with that drug.
  • the information displayed on the display device or sound output device may include, for example, the following information.
  • - Estimated sedation depth calculated in real time (sedation depth estimate updated approximately every 0.1 seconds from the last 4 seconds of EEG data) in the range of 0 to 100.
  • An alert sound is issued using a specific threshold, and real-time estimates and log data of estimates are visually displayed, allowing the user to grasp the sedation trend in real time and over time.
  • Real-time noise level allowing the operator to assess the reliability of the current estimate (e.g. greying out the use of electrocautery to indicate high noise).
  • Fig. 14 is a flowchart showing an example of an estimation model generation process of the server 30 according to the embodiment.
  • a process in which the server 30 generates an estimation model for estimating a sedation depth from an EEG signal will be described.
  • the process shown in Fig. 14 can also be processed by the application control unit 512 (e.g., the generation unit 517) of the processing terminal 50.
  • step S102 the acquisition unit 313 sequentially acquires EEG signals from an EEG measuring device attached to the user's ear.
  • the acquisition unit 313 sequentially acquires EEG signals measured by bioelectrodes provided on the ear tips of earphones.
  • step S104 the acquisition unit 313 or the learning unit 314 performs processing on the acquired EEG signal.
  • the processing includes, for example, calculating the difference between the left and right EEG signals, frequency conversion, calculating the power value at each frequency, filtering, etc.
  • step S106 the learning unit 314 performs machine learning using learning data including the processed EEG signals acquired in sequence and the sedation depth (teacher label) for these EEG signals.
  • step S108 the generation unit 315 generates the trained model trained by the learning unit 314 using the training data as an estimated model.
  • the above process makes it possible to generate a sedation depth estimation model in advance using the learning data measured by the target user.
  • FIG. 15 is a flowchart showing an example of a process for estimating sedation depth by the processing terminal 50 according to an embodiment. In the example shown in FIG. 15, a process for estimating sedation depth from an electroencephalogram signal by the processing terminal 50 is described.
  • the acquisition unit 513 sequentially acquires an EEG signal from an EEG measuring device attached to the user's ear.
  • the EEG signal may mean an EEG signal on which a predetermined process such as the frequency calculation described above has been performed.
  • step S204 the estimation unit 514 estimates the sedation depth for the sequentially acquired EEG signals using a learning model (estimation model) that has learned learning data including the EEG signals and the sedation depth for these EEG signals.
  • a learning model estimation model
  • step S206 the output unit 515 outputs sedation depth information regarding the sedation depth estimated by the estimation unit 514.
  • the output unit 515 outputs the sedation depth information to a display device.
  • the above process makes it possible to estimate the depth of sedation based on EEG signals while reducing the burden on the user by using a simple EEG measuring device that can acquire EEG signals from the ear. Furthermore, the disclosed technology makes it possible to estimate and present the depth of sedation at a level that can be evaluated in real time, and is also capable of responding to sudden changes in the depth of sedation.
  • FIG. 16 is a flowchart showing an example of a learning model generation process of the processing terminal 50 according to the embodiment.
  • the process shown in FIG. 16 can be executed by the application control unit 512 of the processing terminal 50, in particular, the acquisition unit 513 and the generation unit 517.
  • the acquisition unit 513 sequentially acquires EEG signals measured by bioelectrodes in contact with the ear canal of a specific user.
  • the acquisition unit 513 sequentially acquires EEG signals measured by bioelectrodes provided on the ear tips of earphones.
  • the EEG signal may mean an EEG signal on which a specific process, such as the frequency calculation described above, has been performed.
  • step S304 the application control unit 512 determines whether or not a first predetermined time has elapsed since the start of acquisition of the EEG signal.
  • the first predetermined time is, for example, three minutes. If the first predetermined time has elapsed (step S304-YES), the process proceeds to step S306; if the first predetermined time has not elapsed (step S304-NO), the process returns to step S302.
  • step S306 the generation unit 517 uses the EEG signal measured at the first predetermined time to set a reference value for the EEG signal in the awake state of the EEG measurement user. For example, the generation unit 517 assigns a teacher label for the awake state to the average value of the EEG signal in the awake state.
  • step S308 when the drug is administered after the first predetermined time has elapsed, the acquisition unit 513 sequentially acquires the EEG signal measured by the bioelectrode in contact with the ear canal of the specified user.
  • step S310 the generator 517 assigns the sedation depth level set by the person providing the treatment (e.g., a doctor) to the EEG signal at that time and labels it.
  • the person providing the treatment e.g., a doctor
  • step S312 the application control unit 512 determines whether or not a second predetermined time has elapsed since the drug was administered to the EEG measurement user.
  • the second predetermined time is, for example, three minutes. If the second predetermined time has elapsed (step S310-YES), the process proceeds to step S314; if the second predetermined time has not elapsed (step S310-NO), the process returns to step S308.
  • step S314 the generation unit 517 performs supervised learning using the EEG signal (learning data) annotated at the second predetermined time, and generates an estimated model for the user.
  • the above process makes it possible to generate an estimation model for sedation depth from EEG signals according to the EEG state of each individual. As a result, it becomes possible to more appropriately estimate sedation depth according to the characteristics of the EEG state of the EEG target user.
  • Fig. 17 is a diagram showing an example of a display screen in the embodiment.
  • a screen D10 shown in Fig. 17 includes time-series data of the estimated value of the sedation depth of the disclosed technique, a sedation depth value (estimated value) M10, and a current drug dosage M12.
  • the estimated sedation depth values estimated using the estimation model of the disclosed technology are displayed in a time series. It is preferable that noises N10 and N12 that are mixed into this time series data during treatment with an electric scalpel, for example, are displayed so that they can be visually distinguished from normal estimated values. For example, as shown in FIG. 17, it is preferable to surround them with a frame or change the color so that it is clear at a glance that they are noise.
  • the sedation depth value M10 displays the most recent estimated sedation depth.
  • a lower sedation depth value indicates deeper sedation, and a higher sedation depth value indicates an awake state.
  • the sedation depth value may be normalized, for example, to a numerical value between 0 and 100. In this case, if the sedation depth value exceeds a threshold value indicating an awake state, or falls below a threshold value indicating an excessively deep sedated state, an alarm may be output or the color of the sedation depth value may be changed.
  • the dosage M12 indicates the current dosage of the drug.
  • the dosage M12 may also be the next dosage of the drug that is estimated based on the estimated value of the sedation depth. Note that the dosage M12 does not have to be a required display item.
  • anesthesia has been used as an example of a drug, but the disclosed technology can be applied to any drug that has a sedative effect, such as a sleeping pill or a tranquilizer.
  • a sedative effect such as a sleeping pill or a tranquilizer.
  • a user takes a sleeping pill or a tranquilizer, it is possible to learn the relationship between the user's electroencephalogram signal at the time of taking the drug and the user's sedation depth at that time.
  • the processing terminal 50 may execute the above-mentioned EEG signal acquisition process, sedation depth estimation process, output process, and the like.
  • Information processing system 10 Earphone set 30, 50 Information processing device 100 Earphone 104 Nozzle 106 Ear tip (elastic electrode) 310 Processor 312 EEG control unit 313 Acquisition unit 314 Learning unit 315 Generation unit 316 Output unit 330 Memory 510 Processor 512 Application control unit 513 Acquisition unit 514 Estimation unit 515 Output unit 516 Selection unit 517 Generation unit 530 Memory 550 User interface

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Veterinary Medicine (AREA)
  • Public Health (AREA)
  • General Health & Medical Sciences (AREA)
  • Animal Behavior & Ethology (AREA)
  • Biomedical Technology (AREA)
  • Surgery (AREA)
  • Medical Informatics (AREA)
  • Molecular Biology (AREA)
  • Pathology (AREA)
  • Biophysics (AREA)
  • Physics & Mathematics (AREA)
  • Psychology (AREA)
  • Psychiatry (AREA)
  • Developmental Disabilities (AREA)
  • Child & Adolescent Psychology (AREA)
  • Educational Technology (AREA)
  • Hospice & Palliative Care (AREA)
  • Social Psychology (AREA)
  • Vascular Medicine (AREA)
  • Anesthesiology (AREA)
  • Hematology (AREA)
  • Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
PCT/JP2023/040207 2022-11-08 2023-11-08 情報処理方法、プログラム、及び情報処理装置 Ceased WO2024101388A1 (ja)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP2024557827A JPWO2024101388A1 (https=) 2022-11-08 2023-11-08

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2022178986 2022-11-08
JP2022-178986 2022-11-08

Publications (1)

Publication Number Publication Date
WO2024101388A1 true WO2024101388A1 (ja) 2024-05-16

Family

ID=91032589

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2023/040207 Ceased WO2024101388A1 (ja) 2022-11-08 2023-11-08 情報処理方法、プログラム、及び情報処理装置

Country Status (2)

Country Link
JP (1) JPWO2024101388A1 (https=)
WO (1) WO2024101388A1 (https=)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2003290179A (ja) * 2002-04-02 2003-10-14 Japan Tobacco Inc 感覚感性評価システム
JP2018158089A (ja) * 2017-03-23 2018-10-11 富士ゼロックス株式会社 脳波測定装置及び脳波測定システム
WO2021070456A1 (ja) * 2019-10-08 2021-04-15 Ｖｉｅ Ｓｔｙｌｅ株式会社 イヤホン、情報処理装置、及び情報処理方法
WO2022107292A1 (ja) * 2020-11-19 2022-05-27 Ｖｉｅ Ｓｔｙｌｅ株式会社 プログラム、情報処理方法、及び情報処理装置

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2003290179A (ja) * 2002-04-02 2003-10-14 Japan Tobacco Inc 感覚感性評価システム
JP2018158089A (ja) * 2017-03-23 2018-10-11 富士ゼロックス株式会社 脳波測定装置及び脳波測定システム
WO2021070456A1 (ja) * 2019-10-08 2021-04-15 Ｖｉｅ Ｓｔｙｌｅ株式会社 イヤホン、情報処理装置、及び情報処理方法
WO2022107292A1 (ja) * 2020-11-19 2022-05-27 Ｖｉｅ Ｓｔｙｌｅ株式会社 プログラム、情報処理方法、及び情報処理装置

Also Published As

Publication number Publication date
JPWO2024101388A1 (https=) 2024-05-16

Similar Documents

Publication Publication Date Title
Nakamura et al. Automatic sleep monitoring using ear-EEG
Blumenthal et al. Committee report: Guidelines for human startle eyeblink electromyographic studies
EP1989998B1 (en) Methods and apparatus for monitoring consciousness
US11543879B2 (en) System for communicating sensory information with an interactive system and methods thereof
US11540769B2 (en) System and method for tracking sleep dynamics using behavioral and physiological information
EP2535000A1 (en) Method and system for quantifying anaesthesia or a state of vigilance
WO2018027141A1 (en) In-ear sensing systems and methods for biological signal monitoring
US20250000447A1 (en) Human-computer interactive device and method
JP2006514570A (ja) 麻酔および鎮静監視のシステムおよび方法
AU2001213824B2 (en) Monitoring auditory evoked potentials
JP2022520211A (ja) 非薬理的に誘発された修正された意識状態のレベルを監視するための方法およびシステム
CN107997751A (zh) 一种基于生物反馈的智能耳机系统
AU2001213824A1 (en) Monitoring auditory evoked potentials
MacDonald et al. Trial effects in single-trial ERP components and autonomic responses at very long ISIs
CN115867183A (zh) 用于监测疼痛水平的方法和系统
KR20180073114A (ko) 진정마취심도 모니터링 방법 및 그 장치
WO2024101388A1 (ja) 情報処理方法、プログラム、及び情報処理装置
US20250284333A1 (en) Program, information processing method, and information processing apparatus
CN218738908U (zh) 一种基于脑电技术的虚拟现实镇痛效果评估的系统
Van De Velde Signal validation in electroencephalography research
JP7422389B2 (ja) 情報処理装置及びプログラム
Ekpar A novel three-dimensional multilayer electroencephalography paradigm
Pazuelo Evaluating In-Ear EEG Signal for patients long-term monitoring
US20250049378A1 (en) System and method of utilizing biofeedback to identify and monitor neural synchrony and to increase parent-child neural synchrony
Uppal et al. Brain-Body Coupling in Listening to Metronomic Sounds and Music

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 23888735

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 2024557827

Country of ref document: JP

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 23888735

Country of ref document: EP

Kind code of ref document: A1