JP2008242687A - Method for identifying sleep and sleep observation system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a sleep observation system for identifying an emergency by distinguishing a sleeping person to be observed from pet noises or the like over a wide range of positions by means of a non-contact human sensor. <P>SOLUTION: The sleep observation system for observing the sleeping state of a person to be observed by means of a non-contact human sensor and information processing functions includes a motion sensing part for sensing the body motion of the person to be observed by means of the non-contact human sensor and binarizing the result, a data recording part for recording the binary signal and the time when the motion is sensed, a sleep identifying part for discriminating between a sleeping state and an awakened state by periodically calculating and comparing low-frequency component intensity and high-frequency component intensity from data corresponding to a predetermined period of time in the data recording part, and a communication part for notifying the result of the sleep identification to a supervisor. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

The present invention relates to a sleep monitoring system that records a binary signal and time when a body motion of a monitored person is detected by a non-contact human sensor and determines a sleeping state of the monitored person through information processing.

As one of the social problems in an aging society, it is required to watch the health and safety of the elderly. Among them, the sick elderly can be monitored in detail in a intensive care unit such as a hospital with body temperature, brain waves, pulse, camera, etc. On the other hand, for healthy elderly people, if they live with their families, they can be watched by their families, but there is no established method for watching elderly people living in a single household to become a nuclear family.

Since the number of elderly living in a single household is greater than the number of people treated in intensive care, it is impossible for the hospital to watch over the elderly in all single households. In addition, the installation of measuring equipment for monitoring the health condition, etc. also has problems in terms of the cost of equipment and technical aspects such as measurement mounting.

Therefore, various information technologies for watching healthy elderly people have been proposed. Patent Document 1 introduces an example using a wireless pendant or a security camera. However, it has been pointed out that security cameras may infringe privacy and that wireless pendants will not function due to mischief. As a method for solving this problem, an electronic pot having a communication function, a method for monitoring the use of water supply and regularly recording and transmitting it have been put into practical use. Such conventional technology is based on the premise that the monitor will wake up and take action, and watch the elderly at night, such as sleep failure and emergency conditions (such as heart failure), to help maintain their health, There has been no technology to solve this problem.

As a method of watching the sleeping state, it is possible to determine from the heart rate and the respiratory rate by the configuration of the system that watches the sleeping state at night by the restraint type sensor (sensor that is assumed to be worn on the body) as in Patent Document 2. It may be possible to estimate to sleep depth. However, for that purpose, a new problem arises in that an elderly person wears the device alone and does not remove the sensor even during sleep, and that the sensor is disturbed by sleep.

Therefore, an unconstrained sensor is preferable as a method for watching the sleeping state. Therefore, for example, Patent Document 3 discloses a technique for collecting a respiratory sound and measuring a sleep state from a sound collection level (sleeping, snoring, explosive sound, etc.) as a sleep monitoring apparatus using an unconstrained sensor. Separately, Patent Document 3 discloses a method of detecting minute body movements from reflected waves of radio waves and observing even respiration and heartbeat.

These disclosed technologies are based on the premise that the monitored person sleeps at a fixed position and there is no target to be detected other than the monitored person. However, in everyday life, there are noise factors that are inconvenient for measuring sound and radio wave reflections outside the monitored person, such as cars and electronic sounds outside the room, and electronic devices and pets indoors. Therefore, it was impossible to measure with high accuracy. Furthermore, because of these difficulties, no means for determining whether there is a sudden change in physical condition while the subject is sleeping is not provided.
Patent Publication No. 10-248093 Patent Publication 2004-89267 Patent Publication 2006-167427 Patent Publication 2006-87850

In the daily life of elderly people in a single household, it is difficult to fix a place to sleep, and pets such as dogs are dressed up and electronic devices such as fans and air conditioners are in operation. Therefore, the conventional sleep watching technology as described above cannot watch sleep. There is a problem that a monitoring method for monitoring sleep in a wider range and eliminating noise factors is required.

The object of the present invention has been made in view of the above points, and the object of the present invention is to monitor even if the monitored person goes to sleep in a wide range such as a bedroom, not bedding, and a noise factor such as a pet. It is an object of the present invention to provide a sleep monitor system that can detect a case in which there is a sudden change in physical condition while the subject is sleeping, even with high accuracy.

  The present invention relates to a system for monitoring a sleep state of a monitored person by a non-contact human sensor and an information processing function, and an operation detection unit that detects a human motion of the monitored person by the non-contact human sensor and converts it to a binary signal; The data recording unit for recording the binary signal and the sensing time, and the low frequency component intensity and the high frequency component intensity are periodically calculated from the data of the data recording unit for a predetermined time, and compared to sleep state and awakening By having a sleep determination unit that separates the state and a communication unit that communicates the result of the sleep determination to a monitor, the sleep component is detected even in a wide range, and high frequency components due to the influence of noise factors such as pets The purpose of watching the sleep time and its quality with high accuracy is achieved.

Further, according to the present invention, as the sleep determination unit of the preceding paragraph, when the low frequency component intensity WL and the high frequency component intensity WH are periodically calculated and compared from the data of the data recording unit for a predetermined time, WL exceeds WH. By making at least one determination as “sleeping state” when no sleep is exceeded and “wake state” when not exceeding, a more reliable sleeping time can be detected.

Further, the present invention provides the sleep determination unit described in the previous section, in which WL is set to WH when periodically calculating and comparing the low frequency component intensity WL and the high frequency component intensity WH from the data of the data recording unit for a predetermined time. When it exceeds, it is referred to as “sleeping state”, when it does not exceed it, it is “wakening state”, and when WH changes from “sleeping state” to 0, it is “emergency state”, and when WH changes from “wakening state” to 0 By determining at least one as “absent state”, it is possible to detect the difference between the emergency situation and the daily state.

Further, according to the present invention, when the sleep determination unit calculates and compares the low-frequency component intensity WL and the high-frequency component intensity WH from the data of the data recording unit for a predetermined time period, By setting to ˜30 minutes, it is possible to detect a detailed sleep state every 5 to 30 minutes.

Further, according to the present invention, when the sleep determination unit calculates and compares the low-frequency component intensity WL and the high-frequency component intensity WH from the data of the data recording unit for a predetermined period of time, the predetermined period of time 60 By setting to ~ 90 minutes, it is possible to detect healthy sleep with a sleep cycle of around 60 to 90 minutes.

Further, the present invention, as a sleep determination unit, when calculating and comparing the low frequency component intensity WL and the high frequency component intensity WH from the data of the data recording unit for a predetermined period of time, By using a part of the short-time discrete Fourier transform or a part of the wavelet transform, it is possible to accurately detect a sleep state in which the low-frequency component is strong.

In addition, the present invention, as a sleep determination unit, periodically calculates and compares the low-frequency component intensity WL and the high-frequency component intensity WH from the data of the data recording unit for a predetermined period of time. By comparing the wave intensity corresponding to 45 minutes to 90 minutes by the value of 1 to 20 as WH, and comparing the sum of the wave intensity corresponding to the time shorter than 30 minutes as WL, it is automatically The presence or absence can be detected.

As described above, the sleep monitor system of the present invention is not limited to a sleeping place within the size of a bedroom, and is not subject to detection errors due to noise from pets, electronic devices, etc. It is possible to detect a sleep state in the daily life of a healthy elderly person accurately and every short time, and to automatically determine an emergency situation.

Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings.

Needless to say, the present invention is not limited to this embodiment, and can be implemented in various forms without departing from the scope of the invention.

(First embodiment)
An outline of the sleep monitor apparatus of the present invention is shown in FIG. The situation where the monitored person T is sleeping on the bedding B placed at an arbitrary position in the bedroom R together with the noise factor NS that is a pet is detected by the non-contact human sensor S and is represented by a broken line as binary signal data. Is sent to the monitor device SM through the wiring shown in FIG. In the monitor device SM, data and time are recorded, data for a predetermined time is periodically determined for sleep, and information is transmitted to the monitor via the communication device SC according to the situation. Yes. An infrared non-contact human sensor or an ultrasonic differential sensor can be used as the non-contact human sensor.

FIG. 2 is a diagram illustrating the internal configuration of the monitor device SM. The non-contact human sensor S sends a binary signal of 1 to the data recording device ST as data when motion is sensed in the time-lapse TS and 0 when not sensed. The data recording device ST continuously records data and time information for each time TS in the format shown in FIG. In FIG. 3, the first item of each row indicates the value 1 or 0 of the binary signal, and the second item records time.

The sleep determination device SD has the processing function SDF shown in FIG. 4, and the data processing takes out data D1 corresponding to the time width TW for each fixed time TSF from the data recording device ST by means SD1, and means SD2 for each frequency component. Is calculated as the intensity WL of the low frequency component and the intensity WH of the high frequency component, and WH and WL are compared by means SD3. When WH is large, the “sleep state” of SD4 is determined, and when WL is large Each of the “wakefulness states” of SD5 is determined. These determination results are transmitted to the supervisor through the communication device SC shown in FIG. 1 for a while.

The behavior of the monitored person T relating to sleeping or getting up first enters the bedroom R from the outside, then lies on the bedding B at an arbitrary position, enters the sleeping state SL, and then moves to the awakening state WK. Moves inside the bedroom R to a basket or out of the room. When moved outside the room, both WH and WL become 0. Therefore, in the present invention, the “abnormal situation” is determined by further examining the time transition of SD3. That is, in the present invention, when WH = WL = 0 at the next time with SD4, it indicates that the body movement is 0 during sleep and is determined as an “abnormal situation”.

Here, by setting the time range TW for frequency analysis to TW = 60 to 90 minutes in the present invention, it is possible to more accurately capture around 90 minutes of the sleep cycle. Further, the frequency analysis constant time TSF is more accurate as it is executed more frequently. However, if it is too short, it is affected by noise. Therefore, in the present invention, in order to ignore noise of 5 minutes or less, TSF 5 to 30 minutes is set as an optimal time interval. In this embodiment, the sampling time TS is 1 minute.

The period calculation means SD2 of the sleep determination process SDF of the present invention analyzes the intensity for each period, and calculates the low frequency intensity WH and the high frequency intensity WL from the result. These WH and WL are evaluated by the following expressions. If expression 1 (WH> WL) is true, it is determined as “sleep state” and false is determined as “wake state”. This determination means SD3 determines the awakening output SD4 or the sleep output SD5.

Here, the intensity calculation means SD2 for each period is mounted as follows according to the present invention.

For ease of explanation, the signal D1 and time data in FIG. 3 are shown in the graph of FIG. 5 as time t and signal X (t). The sampling time TS is a time range sensed by the non-contact human sensor, and is 1 minute in this embodiment.

In the present invention, the calculation of the low frequency intensity WL and the high frequency intensity WH of the frequency component is implemented by the short-time discrete Fourier transform shown in FIG.

The expression in FIG. 6 represents time t with the sampling time TS as a basic unit. N is the number of basic units corresponding to the predetermined time TW, that is, TW = N × TS. k is a parameter representing a wave having a period Tk, and has a relationship of Tk = TW / k with the time range TW.

There are two reasons why the short-time discrete Fourier transform is used in the calculation of the intensity of the frequency component of the present invention. The first reason is that the wave component up to the period TW can be analyzed with high accuracy in the time range of TW. Moreover, it is because the change of the intensity | strength of the frequency component for every time TSF is detectable by shifting a time range only by TSF. The frequency component intensity calculation method of the present invention may be a weblet transform method having the same effect.

In the equation of FIG. 6, Y (k) is the complex amplitude of the wave k. The intensity of the wave k can be calculated from the absolute value of Y (k). FIG. 7 shows an example in which the intensity W (k) of the wave k is plotted for various k by short-time discrete Fourier transform. In FIG. 7, the horizontal axis is k, and the vertical axis is intensity W (k).

As an example specifically showing the effectiveness of the intensity calculation method in the present invention, the input data D1 is shown in FIG. 8, and the analysis result is shown in FIG. As shown. In FIG. 9, the time interval TW is 90 minutes. Also, W (0) having an infinite cycle is excluded. According to FIG. 9, k = 2, that is, the period Tk = TW / k = 90/2 = 45 minutes, that is, the wave intensity W (2) corresponding to the 45-minute period has a maximum value of 28.6. . From this, it can be seen that the binary data D1 can be extracted having a period of 45 minutes.

As can be seen in many of the literature on sleep, the sleep cycle is said to be about 90 minutes. Therefore, in the present invention, the intensity of the sleep cycle around Tk = 90 minutes (60 to 90 minutes) is used for the determination of the sleep state.

The intensity of the period may be excessively strong depending on the sensitivity of the sensor. Therefore, in the present invention, the intensity WL of the low cycle wave corresponding to the sleep cycle is standardized and evaluated by the wave intensity W (0) of the infinite cycle. That is, the value obtained by normalizing the sum of the wave intensities satisfying k> 3 by W (0) is set as WH, and when Equation 1 is true, the sleep state is determined.

If the sleep determination unit SD2 is periodically performed, the time variation of WH and WL as shown in FIG. 10 is obtained, and the sleep state can be examined in time series. In FIG. 10, the solid line is WL and the broken line is WH.

Before and after normal sleep, “wake” should be “sleep”, or “sleep” should be “wake”. In addition, since “no operation” is a situation where the non-contact human sensor such as a bedroom is outside the detection range or does not operate at all, in the present invention, if “awake” is “no operation”, “out of the bedroom” Can be determined. Further, since “sleep” to “no operation” means that a serious problem has occurred, it can be determined that it is an “emergency”.

At the end of this example, a graph of the recorded data of D1 obtained in our field experiment is shown in FIG. FIG. 12 shows a plot of WL (solid line) and WH (dashed line) calculated for α = 5. It is known that the time when the subject took sleep almost coincides with the portion where the intensity of the small high frequency wave seen in FIG. 12 is lower than the intensity of the low frequency wave.

This sleep monitoring device that monitors the safety of elderly people who are living alone or people with physical disabilities provides a more flexible and comfortable monitoring environment without infringing on the privacy of the monitoring target in a society where aging is expected. It is indispensable as a technology that can be used industrially.

Schematic of the sleep monitor device of the present invention. The block diagram of the sleep monitor apparatus of this invention. An example of binary signal D1 data of the apparatus ST1 of the present invention. The processing flow SDF figure of the sleep determination apparatus SD of this invention. The time history graph of the binary signal D1 of this invention. The short-time discrete Fourier transform formula of the present invention. The graph of the intensity | strength according to the period of this invention. The graph of the binary signal D1 of a 45 minute period. The graph of the intensity | strength according to the period of the binary signal D1 of a 45 minute period. Schematic diagram of the time history of WL and WH. The time history graph of the binary signal D1 obtained by experimental data. Time transition graph of WL and WH obtained from experimental data.

Explanation of symbols

T Monitored person S Non-contact human sensor SM Monitor device SC Communication device NS Noise source R Bedroom SC Communication device D1 Operation binary signal ST Recording device SD Sleep determination device

Claims (7)

In a system that monitors the sleep state of a monitored person using a non-contact human sensor and information processing function,
A motion detection unit that detects a human motion of a monitored person with a non-contact human sensor and converts it into a binary signal, a data recording unit that records the binary signal and the sensing time, and the data for a predetermined time regularly A sleep determination unit that separates the sleep state from the awake state by calculating and comparing the low-frequency component intensity and the high-frequency component intensity from the data of the recording unit, and a communication unit that communicates the result of the sleep determination to the monitor Sleep monitoring system characterized by being. As a sleep determination unit according to claim 1, when the WL exceeds WH when the low frequency component intensity WL and the intensity WH of the high frequency component are periodically calculated from the data of the data recording unit for a predetermined time and compared. A sleep watching system characterized in that at least one determination is made with “sleeping state” and when not exceeding “wakeful state”. As a sleep determination unit according to claim 1, when the WL exceeds WH when the low frequency component intensity WL and the intensity WH of the high frequency component are periodically calculated from the data of the data recording unit for a predetermined time and compared. “Sleeping state”, “wake state” when not exceeding, “urgent state” when WH changes from “sleeping state” to 0, and “absence state” when WH changes from “wake state” to 0 A sleep watching system characterized in that at least one determination is made. As the sleep determination unit according to claims 1 to 3, when the low-frequency component intensity WL and the high-frequency component intensity WH are periodically calculated and compared from the data of the data recording unit for a predetermined time, the periodic time is set to 5 A sleep watching system characterized in that it is set to 30 minutes. The sleep determination unit according to any one of claims 1 to 4, wherein when the low frequency component intensity WL and the high frequency component intensity WH are periodically calculated and compared from the data of the data recording unit for a predetermined time, the predetermined time is 60. A sleep watching system characterized by being set to ~ 90 minutes. When calculating and comparing the low frequency component strength WL and the strength WH of the high frequency component from the data of the data recording unit for a predetermined time period as the sleep determination unit according to claims 1 to 5, as the calculation method, A sleep monitoring system using a part of a short-time discrete Fourier transform or a part of a wavelet transform. As the sleep determination unit according to claim 1, when the low frequency component intensity WL and the high frequency component intensity WH are periodically calculated and compared from the data of the data recording unit for a predetermined time, a period is used as the comparison method. A sleep monitoring system characterized in that a value obtained by multiplying a wave intensity corresponding to 45 minutes to 90 minutes by a value of 1 to 20 is set as WH and a sum of wave intensity corresponding to a shorter time than 30 minutes is set as WL. .

Images