Detection method, information processing device, and computer-readable non-transitory storage medium
The use of a wearable terminal with an IMU attached to limbs or the head for sleep apnea detection addresses the limitations of existing methods by providing a convenient and accurate means to monitor respiratory events, improving detection accuracy and reducing user burden.
Patent Information
- Application Number
- PCT/JP2024/013463
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2023-09-01
- Filing Date
- 2024-04-01
- Publication Date
- 2024-11-21
AI Technical Summary
Current methods for detecting sleep apnea, such as polysomnography, are limited by high costs and restricted testing locations, and existing wearable sensors that monitor breathing movements can disrupt sleep and are cumbersome, leading to poor sleep quality and undiagnosed cases.
A detection method using a wearable terminal with an inertial measurement unit (IMU) attached to limbs or the head, which acquires and processes sensor information to detect respiratory events through acceleration and gyro data, reducing user burden and improving detection accuracy.
The method allows for more convenient and accurate detection of sleep apnea events, reducing the burden on users and potentially identifying undiagnosed cases by using a lightweight, low-power wearable device that does not disrupt sleep.
Smart Images

Figure JP2024013463_21112024_PF_FP_ABST
Abstract
Description
DETECTION METHOD, INFORMATION PROCESSING DEVICE, AND COMPUTER-READABLE NON-TRANSIENT STORAGE MEDIUM
[0001] The present disclosure relates to a detection method, an information processing device, and a computer-readable non-transitory storage medium.
[0002] In sleep apnea (SA), an apneic state in which breathing stops or a hypopneic state in which breathing becomes shallow repeatedly occurs during sleep.
[0003] Sleep apnea is divided into obstructive, central, and mixed types. In obstructive sleep apnea, apnea or hypopnea occurs due to collapse or obstruction of the upper airway. In central sleep apnea, no airway obstruction occurs. In central sleep apnea, apnea or hypopnea occurs due to the brain's cessation of breathing commands. In mixed sleep apnea, obstructive and central sleep apnea occur together.
[0004] Repeated apnea or hypopnea during sleep can lead to poor sleep quality, which can have adverse effects on social life, such as daytime sleepiness and reduced motivation for work or study.
[0005] Polysomnography (PSG) is used to diagnose sleep apnea. PSG can be used to confirm sleep apnea, for example. However, PSG has limitations in terms of testing location and cost.
[0006] To address this issue, a simpler technique for detecting sleep apnea is known, which detects abdominal and chest movements caused by breathing as changes in acceleration. This technique improves the accuracy of respiratory detection by detecting acceleration in three directions using a triaxial acceleration sensor.
[0007] JP 2007-292514 A JP 2022-542117 A International Publication No. 2017 / 098780
[0008] In conventional technology, acceleration sensors are attached to the abdomen or chest of a patient (subject) to detect breathing movements, but the acceleration sensors attached to the abdomen or chest can disturb the subject's sleep.
[0009] Patients undergoing sleep apnea diagnosis may already be experiencing poor sleep quality, and the sleep disruption caused by the accelerometer worn for the test may further worsen the patient's sleep quality.
[0010] Therefore, the present disclosure proposes a detection method, an information processing device, and a computer-readable non-transitory storage medium that can further reduce the burden on the subject.
[0011] It should be noted that the above problem or object is merely one of multiple problems or objects that can be solved or achieved by multiple embodiments disclosed in this specification.
[0012] The detection method of the present disclosure includes acquiring respiratory data of a user for a first time period from sensor information acquired by a sensor attached to at least one of the user's limbs and the user's head, and detecting respiratory event information related to a respiratory event according to a portion of the respiratory data that periodically varies in a second time period that is shorter than the first time period.
[0013] FIG. 1 is a diagram for explaining an example of a detection method according to the technology of the present disclosure. FIG. 2 is a block diagram showing an example of a configuration of a wearable terminal according to an embodiment of the present disclosure. FIG. 3 is a block diagram showing an example of a configuration of a terminal device according to an embodiment of the present disclosure. FIG. 4 is a block diagram showing an example of a configuration of an information processing device according to an embodiment of the present disclosure. FIG. 5 is a diagram showing an example of respiratory data generated by a generation unit according to an embodiment of the present disclosure. FIG. 6 is a diagram showing an example of an analysis result by an analysis unit according to an embodiment of the present disclosure. FIG. 7 is a diagram for explaining an example of detection of a respiratory event by a detection unit according to an embodiment of the present disclosure. FIG. 8 is a diagram showing an example of a detection result of a respiratory event by a detection unit according to an embodiment of the present disclosure. FIG. 9 is a flowchart showing an example of the flow of a detection process according to an embodiment of the present disclosure. FIG. 10 is a diagram showing an example of a detection result of a respiratory event according to an embodiment of the present disclosure. FIG. 11 is a diagram showing another example of a detection result of a respiratory event according to an embodiment of the present disclosure. TRT 10 illustrates the relationship between gyro and AHI for a training group according to an embodiment of the present disclosure. TRT1 illustrates the relationship between RES and AHI for a training group according to an embodiment of the present disclosure. TRT 10 illustrates the relationship between acceleration and AHI for a test group according to an embodiment of the present disclosure. TRT 10 illustrates the relationship between the gyro and AHI of a test group according to an embodiment of the present disclosure. TRT 1 illustrates the relationship between RES and AHI of a test group according to an embodiment of the present disclosure. TRT Fig. 10 is a diagram illustrating a relationship between Fig. 11 is a block diagram illustrating a configuration example of an information processing device according to another embodiment of the present disclosure Fig. 11 is a diagram illustrating an example of a hardware configuration of a device, etc.
[0014] Hereinafter, embodiments of the present disclosure will be described in detail with reference to the accompanying drawings. In this specification and drawings, components having substantially the same functional configurations are designated by the same reference numerals, and redundant description will be omitted.
[0015] In this specification and drawings, similar components of the embodiments may be distinguished by adding at least one different alphabet and / or number after the same reference numeral. However, if there is no need to particularly distinguish between the similar components, only the same reference numeral will be used.
[0016] Furthermore, although specific values are sometimes used in the present specification and drawings, these values are merely examples and other values may also be applied.
[0017] One or more embodiments (including examples, modifications, and application examples) described below can be implemented independently. However, at least a portion of the embodiments described below may be implemented in appropriate combination with at least a portion of another embodiment. These embodiments may include novel features that are different from each other. Therefore, these embodiments may contribute to solving different purposes or problems and may produce different effects from each other.
[0018] <<1. Introduction>> Sleep apnea (SA) is characterized by repeated episodes of apnea and hypopnea, and awakenings during sleep. It is said that 10% to 30% of adults experience symptoms of SA. SA can be divided into obstructive sleep apnea, which occurs due to airway collapse or obstruction; central sleep apnea, which occurs due to abnormalities in the respiratory center; and mixed sleep apnea, which is a combination of obstructive and central sleep apnea.
[0019] When SA occurs, sleep quality declines, which can lead to daytime sleepiness, reducing work and learning ability and increasing the risk of accidents, creating problems.
[0020] SA also causes repeated hypoxic states and sustained sympathetic nervous activation at night, which have been shown to increase the risk of cognitive impairment, hypertension, atrial fibrillation, stroke, and sudden nocturnal death.
[0021] Polysomnography (PSG) can be used to confirm the diagnosis of SA. For example, a patient suspected of having SA is admitted to a hospital and undergoes a PSG overnight. This allows a specialist to determine whether or not the patient actually has SA, and if so, the severity of the condition.
[0022] PSG is generally performed in hospitals using dedicated equipment, which means that there are problems such as limited locations where the test can be performed and high costs.
[0023] To solve this problem, portable testing devices that can be used at home have been developed, but there is a need for devices that can perform testing more easily and at lower cost.
[0024] Furthermore, if patients have subjective symptoms of SA, such as insomnia or daytime sleepiness, they can visit a hospital and undergo testing. However, patients who have few subjective symptoms may not have the opportunity to undergo testing and may not be able to receive treatment. In fact, it is believed that many SA patients are undiagnosed.
[0025] Making SA testing easier and more inexpensive will lead to the detection of potential SA patients. From the perspective of detecting potential SA patients, it is also desirable to make testing easier and more inexpensive.
[0026] Therefore, in the technology disclosed herein, a respiratory event is detected based on sensor information acquired by a sensor attached to a limb such as a wrist or an ankle, or to the head, for example.
[0027] (Overview of Proposed Technology) Fig. 1 is a diagram for explaining an example of a detection method according to the technology of the present disclosure. As shown in Fig. 1, the detection method according to the technology of the present disclosure is executed in, for example, an information processing system 1.
[0028] The information processing system 1 shown in FIG. 1 includes an information processing device 10, a wearable terminal 20, and a terminal device 30.
[0029] The information processing device 10 is a device that executes the detection method and detects a user's respiratory events. The information processing device 10 may be, for example, a cloud server device disposed on a network. Alternatively, the information processing device 10 may be an edge server device.
[0030] The wearable device 20 is an information processing device that is worn on, for example, the limbs or head of a user. In the example of Fig. 1, the wearable device 20 is a wristwatch-type device that is worn on the user's wrist. Note that the wearable device 20 is not limited to a wristwatch-type device, and may be a device that is worn on the ankle or head.
[0031] The wearable device 20 is equipped with a sensor and acquires sensor information. An example of the sensor is an inertial measurement unit (IMU). The IMU includes at least an acceleration sensor that acquires three-axis acceleration and an angular velocity sensor that acquires three-axis angular velocity.
[0032] The wearable terminal 20 transmits the acquired sensor information to the terminal device 30 .
[0033] The terminal device 30 is an information processing device used by a user. The terminal device 30 may be, for example, a smartphone, a tablet terminal, or a PC. The terminal device 30 transmits sensor information acquired from the wearable terminal 20 to the information processing device 10. Note that if the wearable terminal 20 and the information processing device 10 can communicate directly, the terminal device 30 may be omitted.
[0034] (Example of Detection Method) The information processing device 10 acquires sensor information acquired by a sensor of the wearable device 20 worn on the user's limbs or head (step S1). The information processing device 10 generates respiratory data of the user for a first period from the acquired sensor information (step S2). The first period is, for example, the period during which the wearable device 20 acquired sensor data (total recording time) or the user's total sleeping time.
[0035] The information processing device 10 detects information about respiratory events (here, respiratory events) in a second period of the respiratory data that is shorter than the first period, i.e., in accordance with locations that vary locally and periodically (step S3). In the example of Fig. 1, the information processing device 10 detects the four locations indicated by arrows as respiratory events of apnea or hypopnea.
[0036] In this detection method, the information processing device 10 detects respiratory events using sensors of the wearable device 20 attached to the user's limbs or head, thereby enabling the information processing device 10 to further reduce the burden on the user.
[0037] The sensors (e.g., IMU) mounted on the wearable device 20 can perform sensing for a long time with low power consumption. Furthermore, the wearable device 20 worn on the limbs or head is small and lightweight. Therefore, by acquiring the user's biometric information using the wearable device 20, the burden on the user can be reduced.
[0038] Furthermore, wristwatch-type devices are becoming popular as wearable terminals 20 equipped with an IMU. The wristwatch-type wearable terminal 20 equipped with an IMU can measure, for example, the amount of activity of the user. Some of these wristwatch-type wearable terminals are designed to be worn not only during the day but also while sleeping.
[0039] The detection method of the present disclosure can further reduce the burden on the user during the examination by acquiring respiratory data using a sensor (e.g., an IMU) mounted on a wearable terminal 20 that the user wears on their limbs or head.
[0040] 2. Configuration Example of Information Processing System>> <2-1. Configuration Example of Wearable Terminal> Fig. 2 is a block diagram showing a configuration example of a wearable terminal 20 according to an embodiment of the present disclosure. The wearable terminal 20 shown in Fig. 2 includes a communication unit 210, a sensor unit 220, a storage unit 230, a control unit 240, and an input / output unit 250.
[0041] (Communication unit 210) The communication unit 210 is a communication interface for communicating with other devices. For example, the communication unit 210 is a network interface. For example, the communication unit 210 is a LAN (Local Area Network) interface such as a NIC (Network Interface Card). Note that the communication unit 210 may be a wired interface or a wireless interface. The communication unit 210 communicates with the terminal device 30 under the control of the control unit 240.
[0042] When the communication unit 210 has a wireless interface, the communication unit 210 may be configured to connect to the terminal device 30 using a short-range wireless communication technology such as Bluetooth (registered trademark) or infrared communication. The communication unit 210 may also be configured to connect to the terminal device 30 or another communication device using a radio access technology (RAT) such as LTE (Long Term Evolution), NR (New Radio), 6G, Wi-Fi (registered trademark), etc. The communication unit 210 may be capable of connecting to a network or another communication device using a radio access technology other than the above-mentioned LTE, NR, 6G, Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.
[0043] (Sensor Unit 220) The sensor unit 220 is a detection device that detects the state of the user. The sensor unit 220 includes, for example, an IMU. The IMU includes, for example, at least one of a three-axis acceleration sensor and a three-axis gyro sensor.
[0044] The IMU is an example, and the sensor unit 220 may include a sensor other than the IMU. The sensor unit 220 may include, for example, at least one of a piezoelectric sensor, a millimeter wave sensor, and a ToF (Time of Flight) sensor, as long as it can acquire sensor information that can generate respiration data, which will be described later.
[0045] (Storage unit 230) The storage unit 230 is a data readable / writable storage device such as a dynamic random access memory (DRAM), a static random access memory (SRAM), a flash memory, or a hard disk. The storage unit 230 functions as a storage means of the wearable terminal 20. The storage unit 230 stores, for example, sensor information acquired by the sensor unit 220.
[0046] (Control Unit 240) The control unit 240 is a controller that controls each unit of the wearable terminal 20. The control unit 240 is realized by a processor such as a CPU (Central Processing Unit) or an MPU (Micro Processing Unit). For example, the control unit 240 is realized by a processor executing various programs stored in a storage device inside the wearable terminal 20 using RAM (Random Access Memory) or the like as a working area. The control unit 240 may be realized by an integrated circuit such as an ASIC (Application Specific Integrated Circuit) or FPGA (Field Programmable Gate Array). The control unit 240 may also be realized by a GPU (Graphics Processing Unit). A CPU, an MPU, an ASIC, an FPGA, and a GPU can all be considered as controllers.
[0047] The control unit 240 controls the sensor unit 220 to start detecting the user's state in accordance with instructions from the user or the terminal device 30. As a result, the sensor unit 220 starts acquiring sensor information. The control unit 240 stores the sensor information acquired by the sensor unit 220 in the storage unit 230.
[0048] The control unit 240 controls the sensor unit 220 to end detection of the user's state after a preset time has elapsed, or in accordance with an instruction from the user or the terminal device 30. This causes the sensor unit 220 to end acquisition of sensor information.
[0049] Note that the control of the start and end of sensor information acquisition by the control unit 240 is not limited to the above example. For example, if the control unit 240 has a function for detecting the start and end of the user's sleep, the control unit 240 may control the start and end of sensor information acquisition based on the detection result of this function. The start and end of sleep can be detected based on the user's location information, movements, etc.
[0050] The control unit 240 transmits the sensor information acquired by the sensor unit 220 to the terminal device 30 via the communication unit 210. The sensor information may be transmitted periodically while the sensor unit 220 is acquiring the sensor information. Alternatively, the sensor information may be transmitted when the sensor unit 220 has finished acquiring the sensor information. The sensor information may be transmitted at a predetermined timing or based on an instruction from the user or the terminal device 30.
[0051] Furthermore, the control unit 240 may transmit the sensor information for each sleep session, or may transmit the sensor information multiple times in a lump.
[0052] The control unit 240 causes the input / output unit 250 to output output information acquired from the terminal device 30. The control unit 240 also notifies the terminal device 30, via the communication unit 210, of input information acquired from the user via the input / output unit 250.
[0053] (Input / Output Unit 250) The input / output unit 250 is a user interface for exchanging information with the user. For example, the input / output unit 250 is an operation device, such as operation keys or a touch panel, that allows the user to perform various operations. Alternatively, the input / output unit 250 is a display device, such as a liquid crystal display (LCD) or an organic electroluminescence display (OLED). The input / output unit 250 may be an audio device, such as a speaker or a buzzer. The input / output unit 250 may also be a lighting device, such as an LED (Light Emitting Diode) lamp. The input / output unit 250 functions as input / output means (input means, output means, display means, operation means, or notification means) of the wearable terminal 20. For example, if the wearable terminal 20 is a sensor or the like, the input / output unit 250 (or some of its functions, such as the input function or output function) may be omitted.
[0054] 3 is a block diagram showing an example of the configuration of a terminal device 30 according to an embodiment of the present disclosure. The terminal device 30 shown in FIG. 3 includes a first communication unit 310, a second communication unit 320, a storage unit 330, a control unit 340, and an input / output unit 350.
[0055] (First communication unit 310) The first communication unit 310 is a communication interface for communicating with other devices (e.g., the wearable terminal 20). For example, the first communication unit 310 is a network interface. For example, the first communication unit 310 is a LAN (Local Area Network) interface such as a NIC (Network Interface Card). Note that the first communication unit 310 may be a wired interface or a wireless interface. The first communication unit 310 communicates with the wearable terminal 20 under the control of the control unit 340.
[0056] When the first communication unit 310 has a wireless interface, the first communication unit 310 may be configured to connect to the wearable terminal 20 using a short-range wireless communication technology such as Bluetooth (registered trademark) or infrared communication. The first communication unit 310 may also be configured to connect to the wearable terminal 20 or another communication device using a radio access technology (RAT) such as LTE (Long Term Evolution), NR (New Radio), 6G, Wi-Fi (registered trademark), etc. The first communication unit 310 may be capable of connecting to a network or another communication device using a radio access technology other than the above-mentioned LTE, NR, 6G, Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.
[0057] (Second communication unit 320) The second communication unit 320 is a communication interface for communicating with other devices. For example, the second communication unit 320 is a LAN interface such as a NIC. The second communication unit 320 may be a device connection interface such as a USB. Note that the second communication unit 320 may be a wired interface or a wireless interface. The second communication unit 320 communicates with an external device (for example, the information processing device 10) under the control of the control unit 340.
[0058] In addition, when the second communication unit 320 has a wireless interface, the second communication unit 320 may be configured to connect to a network using wireless access technologies such as LTE, NR, 6G, Wi-Fi (registered trademark), Bluetooth (registered trademark), etc. In this case, the second communication unit 320 may be configured to be able to use different wireless access technologies. For example, the second communication unit 320 may be configured to be able to use NR and Wi-Fi (registered trademark). Furthermore, the second communication unit 320 may be configured to be able to use different cellular communication technologies (e.g., LTE and NR). In addition, the second communication unit 320 may be able to connect to a network using wireless access technologies other than LTE, NR, Wi-Fi (registered trademark), and Bluetooth (registered trademark).
[0059] Here, the terminal device 30 is configured to include a first communication unit 310 that communicates with the wearable terminal 20 and a second communication unit 320 that communicates with the information processing device 10, but the number of communication units included in the terminal device 30 is not limited to two. For example, the terminal device 30 may communicate with both the wearable terminal 20 and the information processing device 10 using a single communication unit.
[0060] (Storage Unit 330) The storage unit 330 is a data readable / writable storage device such as a DRAM, an SRAM, a flash memory, a hard disk, etc. The storage unit 330 functions as a storage unit of the terminal device 30.
[0061] (Control Unit 340) The control unit 340 is a controller that controls each unit of the terminal device 30. The control unit 340 is realized by a processor such as a CPU or an MPU. For example, the control unit 340 is realized by a processor executing various programs stored in a storage device inside the terminal device 30 using RAM or the like as a work area. The control unit 340 may be realized by an integrated circuit such as an ASIC or an FPGA. The control unit 340 may also be realized by a GPU. The CPU, MPU, ASIC, FPGA, and GPU can all be considered to be controllers.
[0062] The control unit 340 transmits sensor information acquired from the wearable terminal 20 to the information processing device 10 in accordance with, for example, an instruction from the user or the information processing device 10 .
[0063] The control unit 340 causes the input / output unit 350 to output output information acquired from the information processing device 10. The control unit 340 also notifies the information processing device 10, via the second communication unit 320, of input information acquired from the user via the input / output unit 350.
[0064] (Input / Output Unit 350) The input / output unit 350 is a user interface for exchanging information with the user. For example, the input / output unit 350 is an operation device such as a keyboard, a mouse, operation keys, or a touch panel that allows the user to perform various operations. Alternatively, the input / output unit 350 is a display device such as a liquid crystal display or an organic EL display. The input / output unit 350 may be an audio device such as a speaker or a buzzer. Furthermore, the input / output unit 350 may be a lighting device such as an LED (Light Emitting Diode) lamp. The input / output unit 350 functions as input / output means (input means, output means, display means, operation means, or notification means) of the terminal device 30. For example, the input / output unit 350 (or some of its functions, such as the input function or output function) may be omitted.
[0065] 4 is a block diagram showing an example of the configuration of the information processing device 10 according to an embodiment of the present disclosure. The information processing device 10 shown in FIG. 4 includes a communication unit 110, a storage unit 120, and a control unit 130.
[0066] (Communication unit 110) The communication unit 110 is a communication interface for communicating with other devices. For example, the communication unit 110 is a LAN interface such as a NIC. The communication unit 110 may also be a device connection interface such as a USB. The communication unit 110 may also be a wired interface or a wireless interface. The communication unit 110 communicates with an external device (for example, terminal device 30) under the control of the control unit 130.
[0067] In addition, when the communication unit 110 has a wireless interface, the communication unit 110 may be configured to connect to a network using wireless access technologies such as LTE, NR, 6G, Wi-Fi (registered trademark), Bluetooth (registered trademark), etc. In this case, the communication unit 110 may be configured to be able to use different wireless access technologies. For example, the communication unit 110 may be configured to be able to use NR and Wi-Fi (registered trademark). Furthermore, the communication unit 110 may be configured to be able to use different cellular communication technologies (e.g., LTE and NR). In addition, the communication unit 110 may be able to connect to a network using wireless access technologies other than LTE, NR, Wi-Fi (registered trademark), and Bluetooth (registered trademark).
[0068] (Storage Unit 120) The storage unit 120 is a data readable / writable storage device such as a DRAM, an SRAM, a flash memory, a hard disk, etc. The storage unit 120 functions as a storage unit of the information processing device 10.
[0069] (Control unit 130) The control unit 130 is a controller that controls each unit of the information processing device 10. The control unit 130 is realized by a processor such as a CPU or an MPU. For example, the control unit 130 is realized by a processor executing various programs stored in a storage device inside the information processing device 10 using RAM or the like as a work area. The control unit 130 may be realized by an integrated circuit such as an ASIC or an FPGA. The control unit 130 may also be realized by a GPU. The CPU, MPU, ASIC, FPGA, and GPU can all be considered to be controllers.
[0070] As shown in FIG. 4 , the control unit 130 includes an acquisition unit 131, a generation unit 132, an analysis unit 133, and a detection unit 134. Each block (acquisition unit 131 to detection unit 134) constituting the control unit 130 is a functional block that indicates a function of the control unit 130. These functional blocks may be software blocks or hardware blocks. For example, each of the above-mentioned functional blocks may be a software module realized by software (including a microprogram), or may be a circuit block on a semiconductor chip (die). Of course, each functional block may be a processor or an integrated circuit. The functional blocks may be configured in any manner. The control unit 130 may be configured by functional units different from the above-mentioned functional blocks.
[0071] (Acquisition unit 131) The acquisition unit 131 acquires sensor information, for example, from the terminal device 30 via the communication unit 110. The acquisition unit 131 may acquire sensor information from a device other than the terminal device 30. The acquisition unit 131 can acquire sensor information directly from the wearable terminal 20, for example.
[0072] Alternatively, the acquisition unit 131 may acquire the sensor information from a storage device that stores the sensor information. The storage device (not shown) is, for example, an external device that collects and stores the sensor information. The storage device can collect the sensor information from at least one of the wearable terminal 20 and the terminal device 30.
[0073] As described above, the sensor unit 220 includes a three-axis acceleration sensor and a three-axis gyro sensor. The acquisition unit 131 acquires sensor information including, for example, a three-axis acceleration signal. The acquisition unit 131 outputs the acquired sensor information to the generation unit 132.
[0074] In the following, for the sake of simplicity, the sensor information will be described as a triaxial acceleration signal, but a triaxial gyro signal can also be converted into respiration data in the same manner as a triaxial acceleration signal. Alternatively, respiration data may be generated using both the triaxial acceleration signal and the triaxial gyro signal.
[0075] (Generation Unit 132) The generation unit 132 generates respiration data based on sensor information.
[0076] First, the generator 132 performs a first filtering process on the three-axis acceleration signals included in the sensor information using a band-pass filter for each axis, and extracts signals related to body movement, heart rate, and breathing (hereinafter also referred to as breathing sensor information) from the sensor information through this first filtering process.
[0077] For example, the generation unit 132 extracts a signal related to body movement from the vector using a band-pass filter of 2 Hz to 4 Hz. For example, the generation unit 132 extracts a signal related to heartbeat from the vector using a band-pass filter of 4 Hz to 33 Hz. For example, the generation unit 132 extracts a signal related to respiration from the vector using a band-pass filter of 0.1 Hz to 2 Hz.
[0078] Next, the generating unit 132 generates a scalar signal R t The scalar signal R t is calculated according to the following formula (1):
[0079]
[0080] Here, X t is the acceleration signal on the X axis after the first filtering process, in other words, included in the respiration sensor information. t is the acceleration signal in the Y axis after the first filtering process, in other words, included in the breathing sensor information. t is the acceleration signal in the Z axis after the first filtering process, in other words, included in the breathing sensor information.
[0081] The generator 132 generates a scalar signal R t A second filtering process is performed on the scalar signal R using a band-pass filter. t Respiratory data is generated from which the DC component has been removed.
[0082] 5 is a diagram illustrating an example of respiratory data generated by the generator 132 according to an embodiment of the present disclosure. The vertical axis of the graph illustrated in FIG. 5 represents a respiratory signal (arbitrary unit), and the horizontal axis represents time (seconds). As illustrated in FIG. 5, the respiratory data (respiratory signal) is generated as a discrete signal.
[0083] Returning to FIG. 4 , the generation unit 132 outputs the generated respiratory data to the analysis unit 133 .
[0084] (Analysis unit 133) The analysis unit 133 analyzes the respiratory data. For example, the analysis unit 133 calculates a high-speed envelope and a low-speed envelope of the respiratory data. The high-speed envelope follows the fast fluctuation of the respiratory data. This fast fluctuation (an example of a first period) is, for example, a periodic fluctuation of about 20 seconds or less. Note that this fast fluctuation is preferably a periodic fluctuation of about 3 seconds to about 10 seconds.
[0085] The slow envelope follows the slow fluctuations of the respiratory data. This slow fluctuation (an example of the second period) is, for example, a periodic fluctuation of about 20 seconds to about 90 seconds. Preferably, this slow fluctuation is a periodic fluctuation of about 30 seconds to 60 seconds.
[0086] For example, the analysis unit 133 calculates the amplitude of the respiratory data using a window function with a width of 3 seconds. More specifically, for example, the analysis unit 133 determines the 80th percentile value of the respiratory data included in the window period with a width of 3 seconds as the amplitude value of the fast envelope of the respiratory data at that time.
[0087] For example, the analysis unit 133 calculates the amplitude of the respiratory data using a window function with a width of 30 seconds. More specifically, for example, the analysis unit 133 determines the 80th percentile value of the respiratory data included in the window period with a width of 30 seconds as the amplitude value of the slow envelope of the respiratory data at that time.
[0088] 6 is a diagram illustrating an example of an analysis result by the analysis unit 133 according to an embodiment of the present disclosure. The vertical axis of the graph illustrated in FIG. 6 represents the respiratory signal (arbitrary unit), and the horizontal axis represents time (seconds). The solid line in FIG. 6 represents the high-speed envelope, and the dotted line represents the low-speed envelope.
[0089] The analysis unit 133 outputs the calculated high speed envelope and low speed envelope to the detection unit 134 .
[0090] (Detection Unit 134 ) The detection unit 134 detects a respiratory event (apnea or hypopnea) based on the analysis result of the analysis unit 133 .
[0091] Here, when the user's breathing is stable and no apnea or hypopnea occurs, the high-speed envelope and the low-speed envelope ideally match.
[0092] When apnea or hypopnea occurs in the user's breathing, a disturbance occurs in the breathing. The high-speed envelope decreases in size in response to the disturbance. However, since the low-speed envelope takes time to follow the disturbance, the decrease in amplitude is smaller than that of the high-speed envelope.
[0093] The detection unit 134 detects a respiratory event by detecting a drop in the high-speed envelope. A respiratory event appears as a fluctuation with a predetermined cycle within a predetermined period. For example, a respiratory event does not appear as a periodic fluctuation within a first period, e.g., the total recording time, but appears as a fluctuation with a predetermined cycle within the predetermined period. Therefore, the detection unit 134 detects a respiratory event at a point where a fluctuation with a predetermined cycle occurs within the predetermined period.
[0094] Here, the predetermined period is, for example, a period of several minutes to several tens of minutes, which is sufficiently short compared to the sleep time (several hours) (a local period, an example of a second period). Accurately measuring the user's sleep time requires, for example, detection of brain waves. When it is difficult to accurately measure the sleep time, the sleep time may be replaced with the test time (for example, the time from when sensing by the sensor unit 220 starts to when it ends) or the sleep time (for example, the time spent lying down in bed). These periods are also referred to as respiratory event detection periods (an example of a first period).
[0095] It is also known that, for example, apnea events occur multiple times (e.g., two or more times) in a localized period as a decrease in respiratory amplitude. For example, apnea occurs repeatedly at intervals of about one minute (e.g., 20 to 90 seconds). Hypopnea occurs repeatedly at intervals of about 20 seconds to 5 minutes, for example. That is, the interval (occurrence period) between respiratory events is about 20 seconds to 5 minutes. Alternatively, for example, apnea occurs repeatedly at intervals of about one minute (e.g., 10 to 120 seconds). Hypopnea occurs repeatedly at intervals of about 10 seconds to 5 minutes, for example. That is, the interval (occurrence period) between respiratory events is about 10 seconds to 5 minutes.
[0096] The amplitude of the respiratory data also changes when the arm position changes due to turning over in bed. However, this change in respiratory amplitude due to the change in arm position occurs over a longer period (global period) than the period during which an apnea event occurs.
[0097] The detection unit 134 detects, as a respiratory event, a drop in amplitude that occurs locally within a predetermined interval (period) within a second period of the respiratory data. The detection unit 134 detects a respiratory event based on the local amplitude drop (the drop in amplitude within the second period), the period during which this drop occurs (e.g., a calculation period described below), and the period during which this drop occurs. For example, the detection unit 134 detects, as a respiratory event, a point within the local period where the amplitude drop occurs, which fluctuates in a period of 20 seconds to 5 minutes or 10 seconds to 5 minutes.
[0098] This allows the detection unit 134 to avoid erroneously detecting fluctuations in respiratory amplitude caused by factors other than respiratory events, such as the position of the arm, as respiratory events, thereby enabling more accurate detection of respiratory events.
[0099] Specifically, the detection unit 134 detects respiratory events using the fast envelope and the slow envelope of the respiratory data. The slow envelope is affected by average fluctuations in respiratory amplitude. On the other hand, the fast envelope is affected by local fluctuations (declines) in respiratory amplitude. Therefore, in order to detect points where the locally changing amplitude value is lower than the globally changing amplitude value, the detection unit 134 detects points where the fast envelope is lower than the slow envelope as respiratory events.
[0100] For example, if the ratio of the difference between the slow envelope and the fast envelope to the slow envelope is equal to or greater than a threshold, the detection unit 134 determines that a respiratory event has occurred at that time.
[0101] More specifically, the detection unit 134 calculates the area of the region surrounded by the high-speed envelope and the low-speed envelope (hereinafter also referred to as the difference region) during the period in which the high-speed envelope is lower than the low-speed envelope (hereinafter also referred to as the calculation period).
[0102] Furthermore, the detection unit 134 calculates the area of the region between the low-speed envelope and a reference line (for example, a line where the respiration amplitude is zero) (hereinafter also referred to as the reference region) during the calculation period. The detection unit 134 calculates the area ratio between the difference region and the reference region, for example, the ratio of the difference region to the reference region.
[0103] If the calculated ratio is equal to or greater than the first threshold, the detection unit 134 detects a respiratory event in the calculation period.
[0104] 7 is a diagram for describing an example of detection of a respiratory event by the detection unit 134 according to an embodiment of the present disclosure. A case where the detection unit 134 detects a respiratory event between times t1 and t2 in FIG. 6 will be described with reference to FIG.
[0105] The detection unit 134 calculates the area of a differential region R1 surrounded by a dotted line and a solid line shown in Fig. 7. The detection unit 134 calculates the area of a reference region R2 between the solid line and the respiratory amplitude. The detection unit 134 calculates a respiratory event based on the ratio of the area of the differential region R1 to the area of the reference region R2.
[0106] Furthermore, the detection unit 134 may detect a respiratory event during a calculation period when the calculation period is between 10 and 90 seconds (or 5 minutes) or between 10 and 120 seconds. Furthermore, the detection unit 134 may detect a respiratory event during a calculation period when the calculation period falls within a predetermined range.
[0107] Furthermore, the detection unit 134 may detect a respiratory event in the calculation period when the calculated ratio is less than a second threshold value.
[0108] 6, the respiration amplitude becomes extremely large and the respiration signal becomes saturated around 600 seconds, which is thought to be due to large movements of the sensor unit 220 caused by turning over in bed, etc.
[0109] In particular, when the sensor unit 220 of the wearable device 20 is attached to the limbs, the sensor unit 220 moves significantly due to body movements such as turning over in bed. Therefore, the respiratory signal may also fluctuate significantly due to this body movement. In this way, even when the respiratory signal fluctuates significantly due to the influence of body movement, the ratio of the difference region to the reference region becomes large.
[0110] Therefore, the detection unit 134 may be configured not to detect the variation in respiratory amplitude due to this body movement as a respiratory event. For example, the detection unit 134 may be configured not to detect a respiratory event during a period in which the respiratory amplitude is equal to or greater than a predetermined value. Alternatively, the detection unit 134 may be configured to detect a respiratory event when the calculated ratio is equal to or greater than a first threshold value and less than a second threshold value (second threshold value).
[0111] In this way, the detection unit 134 detects a respiratory event according to the amplitude of the respiratory data, thereby preventing erroneous detection of a respiratory event and further improving the detection accuracy of a respiratory event.
[0112] 8 is a diagram illustrating an example of a result of detecting a respiratory event by the detection unit 134 according to an embodiment of the present disclosure. In Fig. 8, the respiratory events detected by the detection unit 134 in the respiratory data illustrated in Fig. 6 are indicated by arrows.
[0113] In the example of Fig. 8, the detection unit 134 detects four respiratory events occurring at a predetermined cycle (an example of a first cycle) during a period T1 (a local period, an example of a second period) that includes times t1 and t2. This predetermined cycle is, for example, 20 seconds or less. Note that this predetermined cycle is preferably 3 to 10 seconds.
[0114] Although the detection unit 134 detects a respiratory event using the high speed envelope and the low speed envelope here, the method of detecting a respiratory event by the detection unit 134 is not limited to this.
[0115] The detection unit 134 may detect a respiratory event based on, for example, a local decrease in amplitude of the respiratory data (the ratio described above), a period during which the amplitude decreases (for example, the interval between time t1 and time t2), and a cycle during which the amplitude decreases.
[0116] Furthermore, for example, the detection unit 134 may detect a period in which the locally changing value of the respiratory data (e.g., the amplitude value changing in the second period) is lower than the globally changing amplitude value (e.g., the amplitude value changing in the first period) as a period in which the amplitude of the respiratory data is locally decreasing.
[0117] As described above, in this embodiment, a respiratory event is detected based on respiratory data generated from sensor information acquired by the sensor unit 220 of the wearable device 20 attached to the limbs or head. In this case, the generated respiratory data is more susceptible to the influence of body movement than when the sensor unit 220 of the wearable device 20 is attached to the abdomen or torso.
[0118] Therefore, as described above, the detection unit 134 detects respiratory events based on the ratio of the difference between the slow envelope and the fast envelope to the slow envelope, so that the detection unit 134 can detect respiratory events with higher accuracy regardless of body movement.
[0119] Although the detection unit 134 detects respiratory events here, the detection unit 134 may detect (calculate) information related to SA other than respiratory events. For example, the detection unit 134 may calculate an AHI (Apnea Hypopnea Index). The AHI is an index of SA and is expressed as the number of respiratory events per hour of sleep. The severity of SA is classified according to the AHI. For example, SA is classified as mild if the AHI is 5 or greater but less than 15, as moderate if the AHI is 15 or greater but less than 30, and as severe if the AHI is 30 or greater.
[0120] For example, the detection unit 134 calculates the AHI by dividing the number of respiratory events detected in the detection period by the detection period.
[0121] In this way, the detection unit 134 may directly detect respiratory events, or may indirectly detect respiratory events by calculating the AHI. Here, the detection unit 134 calculates the AHI based on the number of respiratory events, but the detection unit 134 may also calculate an REI (Respiratory Event Index) instead of the AHI. In this way, the detection unit 134 detects information related to respiratory events (e.g., at least one of the respiratory events, the AHI, and the REI).
[0122] The detection unit 134 may also detect a respiratory event index (e.g., the number of apnea / hypopnea events per hour). Preferably, the detection unit 134 calculates the number of apnea / hypopnea events detected per hour during sleep, and calculates an aggregate average. Furthermore, the detection unit 134 may calculate sleep information (e.g., sleep duration), body movement information (e.g., body movement count and body movement duration), heart rate information (e.g., heart rate), and respiratory information (e.g., respiratory rate). The respiratory rate is calculated based on the respiratory data for the first period, for example.
[0123] The detection unit 134 transmits the detection result to the terminal device 30, for example, as a detection report. The detection report may include, for example, sleep information (e.g., sleep duration), body movement information (e.g., body movement count and body movement duration), heart rate information (e.g., heart rate), respiratory information (e.g., respiratory rate), respiratory event index, AHI, REI, etc. At least a portion of the detection report may be presented to the user via the input / output unit 350 of the terminal device 30. Alternatively, at least a portion of the detection report may be transmitted to the wearable terminal 20 via the terminal device 30 and presented to the user via the input / output unit 250 of the wearable terminal 20.
[0124] The detection unit 134 may evaluate the detection result. For example, the detection unit 134 may evaluate the possibility of sleep apnea syndrome, the quality of sleep, and the like based on the detection result. The detection unit 134 may evaluate a respiratory event index based on the number of respiratory events, for example. The detection unit 134 may include the evaluation result in a detection report and transmit the report to the terminal device 30.
[0125] 3. Example of Detection Processing>> Fig. 9 is a flowchart showing an example of the flow of the detection processing according to an embodiment of the present disclosure. The detection processing shown in Fig. 9 is executed by the information processing device 10 in accordance with, for example, an instruction from a user.
[0126] As shown in FIG. 9, the information processing device 10 acquires sensor information from the wearable terminal 20 via the terminal device 30 (step S101).
[0127] Next, the information processing device 10 generates respiratory data (step S102). The information processing device 10 generates the respiratory data by performing first and second filtering processes and calculation processes of a scalar signal of the respiratory sensor information.
[0128] The information processing device 10 analyzes the respiratory data (step S103). For example, the information processing device 10 analyzes the respiratory data by calculating a high-speed envelope and a low-speed envelope of the respiratory data.
[0129] The information processing device 10 detects a respiratory event from the respiratory data (step S104). For example, the information processing device 10 detects a respiratory event based on the ratio of the difference between the high speed envelope and the low speed envelope to the low speed envelope.
[0130] As described above, the information processing device 10 according to the present embodiment generates respiratory data based on sensor information acquired by sensors attached to at least one of the user's limbs and head, and detects respiratory events from the respiratory data.
[0131] This allows the information processing device 10 to detect SA while further reducing the burden on the user.
[0132] <<4. Experiment>> The present inventors conducted an experiment to detect SA using a wristwatch-type wearable device worn on the wrist of 122 subjects. Sensor information was collected by having the subjects wear a wristwatch-type wearable device on their left wrist while undergoing overnight PSG testing at a hospital to diagnose SA or confirm the effectiveness of SA treatment. In this experiment, triaxial acceleration information and triaxial gyro information were collected from the IMU installed in the wristwatch-type wearable device simultaneously with the PSG.
[0133] The three-axis acceleration information includes acceleration signals for the X, Y, and Z axes. The acceleration signals for each axis were acquired at a sampling period of 32 Hz. The three-axis gyro information includes gyro signals for the X, Y, and Z axes. The gyro signals for each axis were acquired at a sampling period of 32 Hz.
[0134] The results of the PSG tests were automatically analyzed using sleep analysis software, and then evaluated and corrected by experts.
[0135] 10 is a diagram illustrating an example of respiratory event detection results according to an embodiment of the present disclosure. Fig. 10 shows the results of respiratory event detection using three-axis acceleration information collected from a representative subject. The vertical axis of Fig. 10 shows the respiratory signal (mG) generated from the acceleration, and the horizontal axis shows the time of the test.
[0136] In the figure, the black area represents the respiratory data generated using the triaxial acceleration information. The dotted line represents the slow envelope, and the solid line represents the fast envelope. The detected respiratory events are indicated by triangles.
[0137] Here, it is assumed that a respiratory event is detected when the calculation period is between 10 seconds and 90 seconds and the calculated ratio is greater than a threshold value.
[0138] 10, no respiratory events were detected and the subject was breathing normally until about 23:38. After 23:38, respiratory events were detected.
[0139] Fig. 11 is a diagram showing another example of respiratory event detection results according to an embodiment of the present disclosure. Fig. 11 shows the results of respiratory event detection using three-axis gyro information collected from the same subject as in Fig. 10. The vertical axis of Fig. 11 shows the respiratory signal (dps) generated from the gyro signal, and the horizontal axis shows the time when the test was performed.
[0140] In the figure, the black area represents the respiratory data generated using the triaxial gyro information. The dotted line represents the slow envelope, and the solid line represents the fast envelope. The detected respiratory events are indicated by triangles.
[0141] Here, as in the case where acceleration information is used, a respiratory event is detected when the calculation period described above is between 10 seconds and 90 seconds and the calculated ratio is greater than a threshold value.
[0142] 11, no respiratory events were detected and the subject was breathing normally until about 23:39. After 23:39, respiratory events were detected.
[0143] Furthermore, the present inventor divided subjects into two groups to develop and verify the detection method (detection algorithm) according to this embodiment.
[0144] As mentioned above, AHI is known as an evaluation index of SA. AHI has several different AHI values depending on the calculation method. TST and AHI TRTThere is a saying. TST is calculated as the time average of respiratory events during total sleep time (TST). TRT is calculated as the time average of respiratory events over the total recording time (TRT). TRT This was done using
[0145] To develop and validate the detection method according to this embodiment, subjects were divided into two groups: a training group and a test group. The characteristics of the subjects in each group are shown in Table 1.
[0146]
[0147] The values in Table 1 are medians or frequencies. The values in parentheses are IQR (Interquartile Range) or percentages. P in Table 1 indicates a significant difference between the two groups. This significant difference was calculated using the chi-square test and Wilcoxon rank sum test.
[0148] AI is apnea index. BMI is body mass index. HI is hypopnea index. OAI is obstructive apnea index. MAI is mixed apnea index. REM is rapid eye movement.
[0149] The present inventors performed statistical analysis of the experimental results using SAS (Statistical Analysis System).
[0150] The present inventors have determined that the accelerations REI and AHI TRT The correlation between the gyro REI and AHI was evaluated using linear regression analysis and Pearson's correlation coefficient. TRT The associations between the two were assessed using linear regression analysis and Pearson's correlation coefficient.
[0151] Here, the acceleration REI is an REI calculated based on respiratory events detected using acceleration information, and the gyro REI is an REI calculated based on respiratory events detected using gyro information.
[0152] The present inventors also evaluated whether acceleration REI or gyro REI can predict AHITRT by multivariate analysis using multiple correlation coefficients with the REG procedure.
[0153] Here, the AHI estimated by the multiple regression model is TRT is used as a respiratory event score (RES), where RES is an index calculated based on the acceleration REI and the gyro REI.
[0154] Here, the acceleration REI and gyro REI of the subjects in the training group were highly correlated with each other and did not show any significant differences. However, the acceleration REI and gyro REI varied greatly depending on the subject.
[0155] Therefore, the present inventors have developed a model (REI) that combines the acceleration REI and the gyro REI.
[0156] RES is calculated using the regression model of the following equation (2).
[0157]
[0158] Furthermore, REI acc is the REI obtained from the acceleration information. gyr is the REI obtained from the gyro information.
[0159] The present inventors also used ROC curves and AUCs to evaluate the classification performance of the SA severity of subjects.
[0160] The present inventors have compared the calculated REI and AHI while changing the value of the first threshold. TRT Table 2 shows the correlation coefficients between REI and AHI for each first threshold (Threshold in Table 2). TRT 10 is a table showing the correlation between
[0161]
[0162] Table 2 shows the results of calculating the acceleration REI and gyro REI while changing the first threshold value between 10% and 60%. In Table 2, r indicates the ratio between REI and AHI. TRT is the correlation coefficient between the multiple regression R 2 Is AHI TRT The squared correlation coefficient between the regression and the acceleration REI and gyro REI.
[0163] In Table 2, the acceleration REI and gyro REI were calculated using the acceleration information and gyro information obtained from the subjects included in the training group. TRT was also calculated from the PSG test results of the subjects included in the training group.
[0164] As shown in Table 2, when the first threshold is 35%, the AHI TRT The results showed that the best estimation results were obtained.
[0165] FIG. 12 shows the acceleration REI and AHI of the training group according to an embodiment of the present disclosure. TRT In FIG. 12, the vertical axis represents the acceleration REI (Acc REI), and the horizontal axis represents the AHI calculated by the PSG test. TRT (PSG AHI TRT ) is shown.
[0166] The acceleration REI in Fig. 12 was calculated using acceleration information acquired from subjects included in the training group. The first threshold used to detect respiratory events was set to 35%. TRT The correlation coefficient r between the two was r=0.84.
[0167] FIG. 13 shows the gyro REI and AHI of the training group according to an embodiment of the present disclosure. TRT In FIG. 13, the vertical axis represents gyro REI (Gyr REI), and the horizontal axis represents AHI calculated by PSG examination. TRT (PSG AHITRT ) is shown.
[0168] The gyro REI in Fig. 13 was calculated using gyro information acquired from subjects included in the training group. The first threshold used to detect respiratory events was set to 35%. Gyro REI and AHI TRT The correlation coefficient r between the two was r=0.88.
[0169] FIG. 14 shows the RES and AHI of the training group according to an embodiment of the present disclosure. TRT In FIG. 14, the vertical axis represents RES, and the horizontal axis represents AHI calculated by PSG testing. TRT (PSG AHI TRT ) is shown.
[0170] The RES in Fig. 14 was calculated using acceleration information and gyro information obtained from the subjects included in the training group. The first threshold used to detect respiratory events was set to 35%. TRT The correlation coefficient r between the two was r=0.88.
[0171] FIG. 15 shows the acceleration REI and AHI of the test group according to an embodiment of the present disclosure. TRT In FIG. 15, the vertical axis represents the acceleration REI (Acc REI), and the horizontal axis represents the AHI calculated by the PSG test. TRT (PSG AHI TRT ) is shown.
[0172] The acceleration REI in Fig. 15 was calculated using acceleration information acquired from subjects included in the test group. The first threshold used to detect respiratory events was set to 35%. TRT The correlation coefficient r between the two was r=0.86.
[0173] FIG. 16 shows the gyro REI and AHI of the test group according to an embodiment of the present disclosure. TRT In FIG. 16, the vertical axis represents gyro REI (Gyr REI), and the horizontal axis represents AHI calculated by PSG examination.TRT (PSG AHI TRT ) is shown.
[0174] The gyro REI in Fig. 16 was calculated using gyro information acquired from subjects included in the test group. The first threshold used to detect respiratory events was set to 35%. Gyro REI and AHI TRT The correlation coefficient r between the two was r=0.78.
[0175] FIG. 17 shows the RES and AHI of the test group according to an embodiment of the present disclosure. TRT In FIG. 14, the vertical axis represents RES, and the horizontal axis represents AHI calculated by PSG testing. TRT (PSG AHI TRT ) is shown.
[0176] The RES in Fig. 17 was calculated using acceleration information and gyro information obtained from the subjects included in the test group. The first threshold used to detect respiratory events was set to 35%. TRT The correlation coefficient r between the two was r=0.84.
[0177] Thus, when REI or RES was calculated in the test group using the first threshold (35%) set based on the results of the training group, a high correlation was obtained between REI or RES and AHITRT, as in the case of the training group.
[0178] The present inventors also investigated the cutoff value of RES for classifying the severity of SA. TST When classifying the severity of SA using the AHI TST =15,30 is AHI TST This is the cutoff value for AHI. TST If the AHI score is less than 15, the subject is diagnosed as having mild symptoms, and if it is 15 or more, the subject is diagnosed as having moderate symptoms. TST If the score is less than 30, the subject is diagnosed as having moderate disease, and if it is 30 or more, the subject is diagnosed as having severe disease.
[0179] Table 3 shows the AHI when the RES cutoff value is set to 10.TST The cutoff values of RES used in Table 3 were the RES and AHI scores in the training group. TST This is a value determined based on the
[0180]
[0181] In Table 3, NPA is the negative predictive accuracy, and PPA is the positive predictive accuracy.
[0182] For example, in Table 3, among the subjects whose RES was less than 10, AHI TST The number of subjects with a mean score of less than 15, in other words, who were actually diagnosed with mild symptoms, was 21, and the AHI TST The number of subjects with a RES of 15 or more, in other words, actually diagnosed as moderate, was 3. When identifying subjects with a RES of less than the cutoff value (10) as mild, the test group was able to identify them with 88% accuracy.
[0183] Among subjects with a RES score of 10 or more, AHI TST The number of subjects with a score of 15 or more, in other words, who were actually diagnosed as having moderate symptoms, was 31, and the AHI TST The number of subjects with a RES score of less than 15, in other words, who were actually diagnosed as having mild symptoms, was 6. When identifying subjects with a RES score of 10 or more as having moderate symptoms, the test group was able to identify them with 84% accuracy.
[0184] When the RES cutoff value was set at 10, the severity of SA (mild or moderate) could be discriminated (classified) with 85% accuracy in the test group.
[0185] Table 4 shows the AHI when the RES cutoff value is set to 20. TST The cutoff values of RES used in Table 4 were the RES and AHI scores in the training group. TST This is a value determined based on the
[0186]
[0187] For example, in Table 4, among the subjects whose RES was less than 20, AHI TST The number of subjects with a mean score of less than 30, in other words, who were actually diagnosed as having moderate symptoms, was 37, and the AHI TST The number of subjects with a RES score of 30 or more, in other words, who were actually diagnosed as having a severe condition, was 2. When identifying subjects with a RES score below the cutoff value (20) as having a mild condition, the test group was able to identify them with 95% accuracy.
[0188] Among subjects with a RES score of 20 or more, AHI TST The number of subjects with a score of 30 or more, in other words, who were actually diagnosed as severe, was 17, and the AHI TST The number of subjects with a RES value of less than 30, in other words, actually diagnosed as having moderate disease, was 5. When identifying subjects with a RES value of 20 or more as having moderate disease, the test group was able to identify them with 77% accuracy.
[0189] When the RES cutoff value was set at 20, the severity of SA (moderate or severe) could be discriminated (classified) with 89% accuracy in the test group.
[0190] Table 5 shows the classification results of SA when the cutoff value is changed. Positive LR in Table 5 indicates the positive likelihood ratio.
[0191]
[0192] In Table 5, among subjects with RES less than 10, 10 to less than 20, 20 to less than 30, 30 to less than 40, and 40 or more, AHI TST The number of subjects with an AHI of less than 15 and TST The number of subjects with a score of 15 or more is shown.
[0193] If RES is less than 10, AHI TST The number of subjects with an AHI of less than 15 was 21, TST The number of subjects with a score of 15 or greater is 3. In this case, the positive likelihood ratio is 0.11.
[0194] Also, if RES is 10 or more but less than 20, AHI TSTThe number of subjects with a mean score of less than 15 was 4, and the AHI TST The number of subjects with a score of 15 or greater is 11. The positive likelihood ratio in this case is 2.18.
[0195] 5. Other Embodiments The above-described method for detecting respiratory events is merely an example, and respiratory events can be detected from sensor information using various methods. For example, respiratory events may be detected using machine learning.
[0196] 18 is a block diagram showing an example configuration of an information processing device 10A according to another embodiment of the present disclosure. A control unit 130A of the information processing device 10A shown in FIG. 18 does not have an analysis unit 133. In addition, the control unit 130A has a detection unit 134A instead of the detection unit 134. Except for these points, the information processing device 10A has the same components as the information processing device 10 shown in FIG. 4.
[0197] The detector 134A detects information related to respiratory events using a detection model that receives respiratory data as input. The detection model is a machine learning model generated by a learning device (not shown) that performs learning using training data. The detection model is generated using, for example, a neural network as a learning algorithm.
[0198] For example, the detection model is generated by performing learning using learning data including respiratory data and information about respiratory events (described later) as ground truth data. In this way, the detection model is generated so as to receive respiratory data as input and output information about respiratory events.
[0199] Alternatively, the detection model may be trained after being classified based on subject information of the detection subject, such as age, sex, underlying diseases, height, weight, etc. In other words, the detection model may be trained to receive respiratory data and subject information as input and output information related to respiratory events.
[0200] Note that the learning algorithm used to generate the detection model is not limited to a neural network. For example, the detection model may be generated using a learning algorithm such as a deep neural network, a support vector machine, clustering, or reinforcement learning. In other words, the detection model may be generated using any machine learning method.
[0201] The detector 134A inputs the respiratory data into a detection model and acquires information about the respiratory event output by the detection model. This information about the respiratory event may be, for example, the respiratory event itself.
[0202] In this case, for example, the detection unit 134A detects the time when a respiratory event occurred based on the output of the detection model. Alternatively, the detection unit 134A may input respiratory data for a predetermined period into the detection model and obtain an output indicating whether a respiratory event occurred during the predetermined period.
[0203] Alternatively, instead of detecting the occurrence of a respiratory event, the detection unit 134A may detect AHI or REI as information related to the respiratory event. In this case, for example, the detection model may output AHI or REI. Note that if the generation unit 132 generates respiratory data using both acceleration information and gyro information, the detection unit 134A may input the respiratory data into the detection model to detect the above-described RES instead of REI.
[0204] Note that the information about respiratory events detected by the detection unit 134A using the detection model is not limited to the above-described example (at least one of respiratory events, AHI, and REI). For example, the detection model may output a determination result of whether or not SA exists based on input respiratory data. In this case, the detection unit 134A detects whether or not the user (subject) corresponding to the respiratory data is an SA patient.
[0205] For example, the detection model may output a determination result of the severity of SA based on input respiratory data. In this case, the detection unit 134A detects the severity of the user (subject) corresponding to the respiratory data. The severity of SA here may include mild, moderate, and severe SA, as well as healthy individuals (not SA patients).
[0206] Furthermore, the input to the detection model is not limited to respiratory data. For example, sensor information may be input to the detection model. Furthermore, in addition to respiratory data, user information about the user may be input to the detection model. The user information may include, for example, at least one of age, gender, underlying disease, height, and weight. Alternatively, multiple detection models may be prepared according to the user information.
[0207] In this case, for example, the acquisition unit 131 acquires the sensor information and corresponding user information from at least one of the wearable terminal 20 and the terminal device 30. The user information may be information acquired by at least one of the wearable terminal 20 and the terminal device 30 through sensing or the like, or may be information input by the user himself / herself into, for example, the terminal device 30.
[0208] Alternatively, the user information may be information acquired by the terminal device 30 from another device (e.g., a weight scale) connected to the terminal device 30. The user information may also be information input by a medical professional or the like.
[0209] Note that the detection unit 134A can protect the privacy of the user by acquiring information about the respiratory event without using user information to input the detection model.
[0210] Furthermore, in the above-described embodiment, the information processing device 10 detects information related to respiratory events, but this information may be detected by at least one of the wearable terminal 20 and the terminal device 30. Because the sensor information acquired from the user is personal information of the user, the user's privacy can be protected by detecting respiratory events by the wearable terminal 20 or the terminal device 30. Note that when respiratory events are detected by the information processing device 10, which is a cloud server device, it is desirable that the information processing device 10 be located on a secure cloud.
[0211] 19 is a diagram showing an example of the hardware configuration of a device, etc. The information processing device 10, the wearable terminal 20, or the terminal device 30 described above is realized by, for example, a computer 1000 shown in FIG.
[0212] The computer 1000 includes a CPU 1100, a RAM 1200, a ROM (Read Only Memory) 1300, a HDD (Hard Disk Drive) 1400, a communication interface 1500, and an input / output interface 1600. The components of the computer 1000 are connected to each other via a bus 1050.
[0213] The CPU 1100 operates and controls each component based on programs stored in the ROM 1300 or the HDD 1400. For example, the CPU 1100 loads the programs stored in the ROM 1300 or the HDD 1400 into the RAM 1200 and executes processing corresponding to the various programs.
[0214] The ROM 1300 stores boot programs such as a Basic Input Output System (BIOS) executed by the CPU 1100 when the computer 1000 is started, and programs that depend on the hardware of the computer 1000 .
[0215] HDD 1400 is a computer-readable recording medium that non-temporarily records programs executed by CPU 1100 and data used by such programs. Specifically, HDD 1400 is a recording medium that records a program for the information processing method according to the present disclosure, which is an example of program data 1450.
[0216] The communication interface 1500 is an interface for connecting the computer 1000 to an external network 1550 (e.g., the Internet). For example, the CPU 1100 receives data from other devices and transmits data generated by the CPU 1100 to other devices via the communication interface 1500.
[0217] The input / output interface 1600 is an interface for connecting the input / output device 1650 and the computer 1000. For example, the CPU receives data from an input device such as a keyboard or a mouse via the input / output interface 1600. The CPU 1100 also transmits data to an output device such as a display, a speaker, or a printer via the input / output interface 1600. The input / output interface 1600 may also function as a media interface for reading programs and the like recorded on a predetermined computer-readable recording medium. Examples of the medium include optical recording media such as a DVD (Digital Versatile Disc) or a PD (Phase Change Rewritable Disc), magneto-optical recording media such as an MO (Magneto-Optical Disc), tape media, magnetic recording media, and semiconductor memories.
[0218] When the computer 1000 functions as the information processing device 10, the wearable terminal 20, or the terminal device 30 described above, the CPU 1100 of the computer 1000 executes a program loaded onto the RAM 1200 to realize the functions of the control units 130, 240, and 340. The program may be stored in the HDD 1400. The CPU 1100 reads and executes the program data 1450 from the HDD 1400, but as another example, the CPU 1100 may obtain the program from another device via an external network 1550.
[0219] Each of the above components may be configured using general-purpose materials or may be configured using hardware specialized for the function of each component. Such configurations may be changed as appropriate depending on the technical level at the time of implementation.
[0220] <<7. Conclusion>> The technologies described above can be specified, for example, as follows. One of the disclosed technologies is a detection method. As described with reference to FIG. 1 and other figures, the detection method includes acquiring respiratory data for a user's TRT (total recording time, an example of a first period) from sensor information acquired by a wearable device 20 (an example of a sensor) attached to at least one of the user's limbs and head (steps S1 and S2). The detection method also includes detecting information related to a respiratory event according to a portion of the respiratory data that periodically fluctuates during a second period that is shorter than the total recording time (step S3). This can further reduce the burden on the subject during SA testing.
[0221] 4 to 9 , the detection method includes detecting respiratory event information (respiratory event, AHI, or REI (RES)) based on the amplitude decrease in the second period of the respiratory data, the period during which the amplitude decrease occurs, and the cycle during which the amplitude decrease occurs, thereby enabling more accurate detection of respiratory event information related to the respiratory event.
[0222] As described with reference to Figures 4 to 9, the detection method includes detecting respiratory event information based on a point where the amplitude value of the second time period of the respiratory data falls below the amplitude value of the total recording time, thereby enabling more accurate detection of respiratory event information related to the respiratory event.
[0223] 4 to 9 , the detection method includes detecting, as a respiratory event, a point of fluctuation in respiratory data that fluctuates in a period of about 20 seconds to about 5 minutes or about 10 seconds to about 5 minutes during a localized period (second period) shorter than the TRT, thereby enabling more accurate detection of respiratory event information related to the respiratory event.
[0224] 4 to 9 , the detection method includes calculating a fast envelope (an example of a first envelope) that follows fluctuations in respiratory data over a first period (e.g., 3 seconds) and detecting respiratory events based on the fast envelope, thereby enabling more accurate detection of respiratory event information related to the respiratory events.
[0225] 4 to 9 , the detection method includes calculating a fast envelope according to the amplitude value of the respiratory data in a window function (an example of a first interval) corresponding to the first period, thereby enabling more accurate detection of respiratory event information related to the respiratory event.
[0226] 4 to 9 , the detection method includes calculating a slow envelope (an example of a second envelope) that follows fluctuations in respiratory data in a second period that is longer than the first period, and detecting a respiratory event based on the slow envelope, thereby enabling more accurate detection of respiratory event information related to the respiratory event.
[0227] 4 to 9 , the detection method includes calculating a slow envelope according to the amplitude value of the respiratory data in a window function (an example of a second interval) corresponding to a second period (e.g., 30 seconds), thereby enabling more accurate detection of respiratory event information related to the respiratory event.
[0228] 4 to 9 , the detection method includes detecting a respiratory event based on the area ratio of a differential region (an example of a first region) surrounded by the high-speed envelope and the low-speed envelope, and a reference region (a second region) between the high-speed envelope corresponding to the differential region and a reference line (a reference value). This allows for more accurate detection of respiratory event information related to the respiratory event.
[0229] As described with reference to Figures 4 to 9, the first period is 20 seconds or less, and the second period is 20 to 90 seconds. This allows for more accurate detection of respiratory event information related to respiratory events.
[0230] As described with reference to Fig. 2 etc., the sensor information is acquired by at least one of an acceleration sensor, a gyro sensor, a piezoelectric sensor, a millimeter wave sensor, and a Time of Flight (ToF) sensor. The detection method can detect respiratory event information using the sensor information acquired by various sensors.
[0231] 18 and other references, the detection method includes detecting respiratory event information using a detection model (an example of a machine learning model) that receives at least one of sensor information and respiratory data as input. The detection method can detect respiratory event information using machine learning.
[0232] The information processing device 10 described with reference to FIG. 1 and other figures is also one of the disclosed technologies. The information processing device 10 includes a control unit 130. The control unit 130 acquires respiratory data for the user's TRT (total recording time, an example of a first period) from sensor information acquired by a wearable device 20 (an example of a sensor) attached to at least one of the user's limbs and head. The control unit 130 also detects information related to respiratory events according to periodically varying portions of the respiratory data during a second period shorter than the total recording time. This allows the information processing device 10 to further reduce the burden on the subject during SA testing. Note that the wearable device 20 and the terminal device 30 other than the information processing device 10 may also detect respiratory event information related to respiratory events.
[0233] The computer-readable non-transitory storage medium described with reference to Figures 1 to 19 is also one of the disclosed technologies. The computer-readable non-transitory storage medium stores a program. The program causes a computer to acquire respiratory data for a user's TRT (total recording time, an example of a first period) from sensor information acquired by a wearable terminal 20 (an example of a sensor) attached to at least one of the user's limbs and head. The program also causes the computer to detect information related to respiratory events according to periodically fluctuating portions of the respiratory data during a second period that is shorter than the total recording time. This further reduces the burden on the subject during SA testing.
[0234] The effects described in this disclosure are merely examples and are not limited to the disclosed contents. Other effects may also be obtained.
[0235] Although the embodiments of the present disclosure have been described above, the technical scope of the present disclosure is not limited to the above-described embodiments, and various modifications are possible within the scope of the gist of the present disclosure. Furthermore, components of different embodiments and modifications may be combined as appropriate.
[0236] The present technology may also be configured as follows. (1) A detection method including: acquiring respiratory data of a user for a first time period from sensor information acquired by a sensor attached to at least one of the user's limbs and head; and detecting respiratory event information related to a respiratory event according to a portion of the respiratory data that periodically varies during a second time period that is shorter than the first time period. (2) The detection method according to (1), in which the respiratory event information is detected based on a decrease in amplitude of the respiratory data during the second time period, a period during which the decrease in amplitude occurs, and a period during which the decrease in amplitude occurs. (3) The detection method according to (1) or (2), in which the respiratory event information is detected according to a portion of the respiratory data where the amplitude value that changes during the second time period is lower than the amplitude value that changes during the first time period. (4) The detection method according to any one of (1) to (3), in which a portion of the respiratory data that fluctuates during a second time period that is shorter than the first time period and that varies with a period of about 10 seconds to about 5 minutes is detected as the respiratory event. (5) The detection method according to any one of (1) to (4), wherein a first envelope that follows fluctuations in the respiratory data in a first period is calculated, and the respiratory event is detected based on the first envelope. (6) The detection method according to (5), wherein the first envelope is calculated based on an amplitude value of the respiratory data in a first section corresponding to the first period. (7) The detection method according to (5) or (6), wherein a second envelope that follows fluctuations in the respiratory data in a second period longer than the first period is calculated, and the respiratory event is detected based on the second envelope. (8) The detection method according to (7), wherein the second envelope is calculated based on an amplitude value of the respiratory data in a second section corresponding to the second period. (9) The detection method according to (7) or (8), wherein the respiratory event is detected based on an area ratio of a first region surrounded by the first envelope and the second envelope, and a second region between the first envelope corresponding to the first region and a reference value. (10) The detection method according to any one of (5) to (9), wherein the first period is about 20 seconds or less.(11) The detection method according to any one of (7) to (9), wherein the second period is a period of approximately 20 seconds to approximately 90 seconds. (12) The detection method according to any one of (1) to (11), wherein the sensor information is acquired by at least one of an acceleration sensor, a gyro sensor, a piezoelectric sensor, a millimeter wave sensor, and a Time of Flight (ToF) sensor. (13) The detection method according to (1), wherein the respiratory event information is detected using a machine learning model that receives as input at least one of the sensor information and the respiratory data. (14) An information processing device comprising: a control unit that acquires respiratory data of the user for a first period from sensor information acquired by a sensor attached to at least one of the user's limbs and head, and detects respiratory event information related to a respiratory event according to a portion of the respiratory data that periodically varies during a second period that is shorter than the first period. (15) A computer-readable non-transitory storage medium storing a program that causes a computer to perform the following steps: acquire respiratory data of a user during a first period from sensor information acquired by a sensor attached to at least one of the user's limbs and head; and detect respiratory event information related to a respiratory event according to a periodically varying portion of the respiratory data during a second period that is shorter than the first period.
[0237] REFERENCE SIGNS LIST 1 Information processing system 10, 10A Information processing device 20 Wearable terminal 30 Terminal device 110, 210 Communication unit 120, 230, 330 Storage unit 130, 130A, 240, 340 Control unit 131 Acquisition unit 132 Generation unit 133 Analysis unit 134, 134A Detection unit 220 Sensor unit 250, 350 Input / output unit 310 First communication unit 320 Second communication unit
Claims
1. A detection method comprising: acquiring respiratory data of a user for a first time period from sensor information acquired by a sensor attached to at least one of the user's limbs and head; and detecting respiratory event information relating to respiratory events according to a periodically varying portion of the respiratory data for a second time period that is shorter than the first time period.
2. The detection method according to claim 1, wherein the respiratory event information is detected based on a decrease in amplitude of the respiratory data during the second period, a period during which the decrease in amplitude occurs, and a cycle during which the decrease in amplitude occurs.
3. The detection method according to claim 1, further comprising detecting the respiratory event information according to a point where the amplitude value of the respiratory data that changes during the second period falls below the amplitude value of the respiratory data that changes during the first period.
4. The detection method according to claim 1, further comprising detecting, as the respiratory event, a point of fluctuation in the respiratory data that fluctuates in a cycle of about 10 seconds to about 5 minutes during a second period that is shorter than the first period.
5. The detection method according to claim 1, further comprising the steps of: calculating a first envelope that tracks fluctuations in the respiratory data in a first period; and detecting the respiratory event based on the first envelope.
6. The detection method according to claim 5, further comprising the step of calculating the first envelope according to an amplitude value of the respiratory data in a first section according to the first period.
7. The detection method according to claim 5, further comprising the steps of: calculating a second envelope that tracks fluctuations in the respiratory data in a second period that is longer than the first period; and detecting the respiratory event based on the second envelope.
8. The detection method according to claim 7, further comprising the step of calculating the second envelope according to an amplitude value of the respiratory data in a second section according to the second period.
9. The detection method according to claim 7, wherein the respiratory event is detected according to a first region enclosed by the first envelope and the second envelope, and an area ratio of a second region between the first envelope corresponding to the first region and a reference value.
10. The method of claim 5, wherein said first period has a period of about 20 seconds or less.
11. The method of claim 7, wherein said second period is from about 20 seconds to about 90 seconds.
12. The detection method according to claim 1, wherein the sensor information is acquired by at least one of an acceleration sensor, a gyro sensor, a piezoelectric sensor, a millimeter wave sensor, and a ToF (Time of Flight) sensor.
13. The detection method of claim 1, further comprising detecting the respiratory event information using a machine learning model that uses as input at least one of the sensor information and the respiratory data.
14. The detection method according to claim 1, further comprising detecting at least one of sleep information, body movement information, and heart rate information from the sensor information.
15. The method of claim 14, further comprising generating a report including at least one of the body motion information, the heart rate information, the respiratory information, and a respiratory event index.
16. The method of claim 1, further comprising detecting a respiratory rate based on the respiratory data during the first period.
17. The method of claim 1, wherein the respiratory event is at least one of an apnea event and a hypopnea event.
18. The method of claim 1, further comprising assessing a respiratory event index based on the number of respiratory events.
19. An information processing device comprising: a control unit that acquires respiratory data of a user during a first period from sensor information acquired by a sensor attached to at least one of the user's limbs and head, and detects respiratory event information relating to respiratory events according to periodically varying parts of the respiratory data during a second period that is shorter than the first period.
20. A computer-readable non-transitory storage medium storing a program that causes a computer to perform the following operations: acquiring respiratory data of a user during a first period from sensor information acquired by a sensor attached to at least one of the user's limbs and head; and detecting respiratory event information related to respiratory events according to a periodically varying portion of the respiratory data during a second period that is shorter than the first period.
Citation Information
Patent Citations
Testing apparatus for apnea syndrome
JP2005152328A
Breathing program, recording medium, and apparatus for breath determination
JP2008110138A
Living body monitoring system
JP2014147594A
Detection of periodic breathing during cpap therapy
JP2017522059A
Device and method for measuring sleep state, phase coherence calculation device, body vibration signal measurement device, stress level mesaurement device, sleep state measurement device, and cardiac waveform extraction method
WO2017141976A1
