JP5037423B2 - Bed leaving prediction apparatus and bed leaving prediction method - Google Patents

Description

  The present invention relates to a bed leaving prediction apparatus and method for determining whether or not a person on a bed has started a bed leaving operation.

Conventionally, mainly hospitals are equipped with a bed detection device that automatically detects a patient's bed leaving the bed in order to reduce the labor of patrol monitoring. The bed leaving detection device includes a pressure sensor installed on the floor surface around the bed and detects the patient's bed leaving by the pressure change that occurs when the patient lowers the foot on the sensor. There is one that attaches a load sensor to the sensor and detects the presence or absence of the patient's bed by changing the output of the sensor. Moreover, as a technique similar to bed detection, a roll-over management system that detects a change in the posture of a patient has been proposed (see Patent Document 1). In this system, measurement sensors are attached to the shoulder and waist of a patient, and radio waves and ultrasonic waves transmitted from the measurement sensor are received by a receiving antenna and an ultrasonic microphone installed on a bed, respectively. A change in the posture of the patient is detected based on the time difference that occurs during the reception. Furthermore, this system limits the radio wave communication range between the measurement sensor and the receiving antenna to the space above the bed, and determines that the patient has left the bed when the radio wave communication is interrupted. I am doing so.
JP 2006-325683 A

  In the above-mentioned bed detection device using the pressure sensor and load sensor, the patient's bed is detected only when the patient has lowered his / her foot from the bed. Therefore, when the nurse arrives, the patient has already left the bed. Time has passed since then. Therefore, there is a problem in terms of early detection of the patient's bed and ensuring sufficient patient safety.

  In the system disclosed in Patent Literature 1, the patient's bed is detected for the first time in a state where the patient no longer exists in the space above the bed. For this reason, a patient's bed cannot be discovered at an early stage like the above-mentioned pressure-sensitive sensor.

  In addition, measurement sensors attached to the shoulders and waist of patients do not reach the ultrasonic microphone when they are hidden in the futon. It was.

  Furthermore, none of the bed leaving determination methods as described above correspond to the automatic change of the bed height. For this reason, it is necessary to change the bed height setting every time the patient or nurse changes the bed height. If the bed height setting was not changed, it would cause false alarms.

  In view of the above circumstances, an object of the present invention is to detect the start of a bed leaving operation of a human body on a bed early and accurately.

  An apparatus according to the present invention, which was created to solve the above-described problems, is a transmitter that continuously transmits an ultrasonic signal to a management area including a bed while being attached to a stationary system, and is attached to the stationary system. A receiver that receives the ultrasonic signal reflected in the management area, and is acquired at a predetermined time by the receiver corresponding to the ultrasonic signal transmitted from the transmitter. The 3D position of the human head or other object on the bed is calculated momentarily from the reflected wave data, and this 3D position is classified into multiple human head groups or other object groups. Whether or not the human body on the bed has started to move out of the bed based on the detection result of detecting the time change from the lying position of the three-dimensional position of the human body from the calculation result of the position calculation means and the position calculation means Determination means for determining whether or not A notification means for issuing a notification signal when the start of the person's bed leaving movement is detected by the means, and the determination means removes a group of heads whose movement cannot be tracked for a predetermined time or more from the subsequent processing. It is characterized by having.

  With the above configuration, since a temporal change from the lying position of the three-dimensional position of the human body on the bed is detected, it is possible to detect the start of a bed leaving operation such as a rising operation on the bed, for example. Therefore, before the human body drops his / her foot from the bed, it is possible to predict the bed leaving of the human body and detect it at an early stage. In addition, the head of the human body is usually outside the futon, and the ultrasonic signal reflected by the head of the human body is not disturbed by the futon. Therefore, it is possible to accurately detect the start of the patient's bed leaving movement by detecting the temporal change of the three-dimensional position of the human body from the lying position. In addition, since a transmitter and a receiver are attached to a stationary system, it is possible to avoid a situation in which a burden is imposed on a patient as in the case where a sensor is directly attached to a human body. Further, by providing the erasing means, it is possible to remove the group of ultrasonic reflected signals that cannot be tracked for a predetermined time or longer from the subsequent processing. That is, depending on the patient's posture, the reflected wave of the ultrasonic wave is small and the peak value may not be captured. In that case, the patient cannot be captured due to missing three-dimensional position data for a certain measurement. Once the capture is interrupted, the group remains, which hinders tracking of the patient's head. Therefore, a delete process is performed in which the delete means deletes a group (missing frame) that has been interrupted for a predetermined time or longer. Thereby, noise can be suppressed.

  The position calculating means can include moving average processing means for attenuating unnecessary signals from the reflected wave data using moving average processing. The moving average is a method of smoothing the series data, and means a process of attenuating a signal unnecessary for subsequent processing such as disturbance ultrasonic noise and ultrasonic waves from the adjacent three-dimensional ultrasonic sensor unit.

  An initialization processing means for automatically detecting the bed height and setting an initial state can be provided. That is, even if the patient or nurse changes the bed height, the initialization processing means automatically detects the bed height and reflects it in the set value when determining whether or not the bed leaving operation has started. Can do.

The set value required to determine whether or not the human body on the bed has started getting out of bed can be changed according to the received angle of the received ultrasonic wave. That is, for a receiving angle where the reflected wave of the ultrasonic wave is weak, the head can be detected reliably by lowering the set value (threshold value) of the head detection, and the misreporting can be reduced. . In this case, the set value can be (cos θ) ⁴ with respect to the receiving angle θ.

  An alarm area is set at least on the outer side of the long side of the bed, and the determination means can determine whether to leave the bed when a human head stays in the alarm area for a predetermined time or more. By setting the alarm area outside the bed, the alarm area can be set in a wide range, and the time during which the person's head exists in the alarm area can be extended. In other words, it is possible to prevent a false alarm that occurs when the head of a person who moves quickly passes through the alarm area in a short time, and can reliably determine getting out of bed.

  An inner alarm area that is smaller than the outer alarm area is set above the bed, and a non-alarm area is provided between the inner alarm area and the outer alarm area on the outer periphery of the inner alarm area. be able to. Thereby, even if a person other than the patient (such as a nurse or an attendant) approaches the bed, false alarms can be reduced in the non-alarm area.

  It is preferable that the wave receiver is an array sensor in which a plurality of wave receiving elements are arranged. Accordingly, it is possible to receive an ultrasonic signal reflected from the management area only by electronic scanning without mechanically scanning the receiver by rotating the receiver.

  The method according to the present invention, which was created to solve the above problems, continuously transmits an ultrasonic signal from a transmitter attached to a stationary system to a management area including a bed and is attached to the stationary system. A reflected wave data acquisition step of receiving the ultrasonic signal reflected in the management area by a receiver and acquiring reflected wave data corresponding to the ultrasonic signal transmitted from the transmitter every predetermined time And, from the reflected wave data acquired in the reflected wave data acquisition step, the head of the human body or other object on the bed is calculated from time to time, and the three-dimensional positions of the heads of a plurality of human bodies are calculated. A position calculation process for classifying into a group or a group of other objects, and a temporal change from the lying position of the three-dimensional position of the human body is detected from the calculation result of the position calculation process, and on the bed based on the detection result Human body separated A determination step for determining whether or not an operation has started, and a notification step for generating a notification signal when the start of a human body leaving motion is detected by the determination step, wherein the position calculation step is a moving average process And a moving average processing step for attenuating unnecessary signals, and the determination step performs an erasure process for excluding a group of heads that cannot track movement for a predetermined time or more from the subsequent processing. .

  An initialization process step of automatically detecting the bed height and setting an initial state can be provided.

  The set value required to determine whether or not the human body on the bed has started getting out of bed can be changed according to the received angle of the received ultrasonic wave.

  As described above, according to the present invention, since the temporal change from the lying position of the three-dimensional position of the human body on the bed is detected, for example, the start of a bed leaving operation such as a rising operation on the bed is detected. be able to. Therefore, before the human body drops his / her foot from the bed, it is possible to predict the bed leaving of the human body and detect it at an early stage. In addition, the head of the human body is usually outside the futon, and the ultrasonic signal reflected by the head of the human body is not disturbed by the futon. Therefore, by detecting a temporal change from the lying position of the three-dimensional position of the human body, it is possible to accurately detect the start of the bed leaving operation of the human body. In addition, groups that cannot track movement for a predetermined time or longer are excluded from subsequent processing, unnecessary signals are attenuated from reflected wave data using moving average processing, and bed height is automatically measured and set. The false alarm can be reduced by reflecting the value. Furthermore, the misreporting can be reduced by changing the set value necessary for the determination of getting out of bed according to the received angle of the received ultrasonic wave, or setting the alarm area on both outer sides of the bed.

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

  FIG. 1 is a block diagram schematically illustrating a bed leaving detection apparatus according to an embodiment of the present invention, and FIG. 2 is a perspective view schematically illustrating an attachment mode thereof. As shown in the figure, the bed leaving prediction apparatus 1 includes a transmitter 2, a receiver 3, a position calculation unit 4, a determination unit 5, a notification unit 6, and an initialization processing unit 22. .

  The transmitter 2 is configured to continuously transmit ultrasonic signals to the management area 8 including the bed 7 at a predetermined interval while being attached to the ceiling. The receiver 3 is configured to receive an ultrasonic signal reflected in the management area 8 while being mounted on the ceiling. The transmitter 2 and the receiver 3 constitute a sensor unit 9.

  In this embodiment, the transmitter 2 is configured to diffuse the ultrasonic signal into the management area 8 and transmit it. On the other hand, the receiver 3 is composed of an array sensor in which a plurality of receiving elements are arranged. For example, the following configuration can be given as an example of the array sensor. That is, as the array sensor 3, as shown in FIGS. 3A and 3B, a plurality of receiving elements 3a are arranged on a single plane (substrate 3b), for example, in a lattice shape or a cross shape. Alternatively, various arrangements can be adopted by arranging as shown in FIGS. 3 (c) and 3 (d). Then, by adding the signal obtained by each receiving element 3a after being delayed by the time corresponding to the incident angle of the ultrasonic signal reflected in the management area 8 and the position of each receiving element 3a, Examples thereof include one configured to acquire three-dimensional reflected wave data by electronic scanning. Furthermore, in this case, it is preferable to set the center-to-center distance X between adjacent receiving elements 3 a to be equal to or less than a half wavelength of the ultrasonic signal transmitted from the transmitter 2. This effectively reduces the situation in which a so-called ghost component that causes an error occurs in the three-dimensional reflected wave data obtained by delay-adding the signals obtained by the plurality of receiving elements 3a. be able to.

  The initialization processing means 22 automatically detects the bed height and performs algorithm initialization processing. That is, even if the patient or nurse changes the bed height, the initialization processing means can automatically detect the bed height.

  The position calculation means 4, from time to time, determines the three-dimensional position of the patient 10 on the bed 7 from the reflected wave data acquired every predetermined time by the wave receiver 3 corresponding to the ultrasonic signal transmitted from the wave transmitter 2. Arithmetic and configured to classify the three-dimensional position into a group of multiple patient heads or other object groups.

  The position calculation means 4 is provided with moving average processing means 20. The moving average processing means 20 performs processing for attenuating unnecessary signals such as disturbance ultrasonic noise and ultrasonic waves from the adjacent three-dimensional ultrasonic sensor unit in the acquired reflected wave data.

  The determination means 5 detects a temporal change from the lying position of the three-dimensional position of the patient 10 from the calculation result of the position calculation means 4, and based on the detection result, whether or not the human body on the bed has started getting out of bed. Configured to determine whether or not.

  The determination means 5 is provided with an erasure means 21. The erasure unit 21 removes the group of heads that cannot track movement for a certain period of time or more from the subsequent processing for the groups classified by the position calculation unit 4.

  The notification unit 6 is configured to issue a notification signal when the determination unit 5 detects the start of the bed leaving operation of the patient 10.

  Next, a procedure for predicting bed leaving by the bed prediction device 1 configured as described above will be described.

  FIG. 4 is a flowchart showing the procedure for detecting getting out of bed. As shown in the figure, this bed detection procedure includes an initialization processing step S1 for automatically detecting the bed height to initialize the algorithm, and an ultrasonic signal from the transmitter 2 to the management area 8 including the bed 7. Reflected wave data corresponding to the ultrasonic signal transmitted from the transmitter 2 by receiving the ultrasonic signal reflected in the management area 8 by the receiver 3 while continuously transmitting at a predetermined interval. A reflected wave data acquisition step S2 for acquiring a predetermined time every time, a position calculation step S3 for momentarily calculating the three-dimensional position of the patient 10 on the bed 7 from the reflected wave data acquired in the reflected wave data acquisition step S2, A temporal change from the recumbent position of the three-dimensional position of the patient 10 is detected from the calculation result of the position calculation step S3, and it is determined whether or not the patient 10 on the bed 7 has started getting out of bed based on the detection result. Determination step S4 to perform, A constant step S4 and a notification step S5, issuing a notification signal if the start of the lifting operation of the patient 10 is detected. In this embodiment, the initialization processing unit 22 executes the initialization processing step S1. A reflected wave data acquisition step S <b> 2 is executed by the receiver 3, and a position calculation step S <b> 3 is executed by the position calculation means 4. Further, the determination unit 5 executes the determination step S4. In addition, the notification unit 6 executes the notification step S5.

First, detection of the room size and bed size, setting of various threshold values in the position calculation step S3 and determination step S4, and measurement of reflected wave data in the management area 8 when the patient is absent or at bedtime are performed. In this embodiment, 25 pieces of reflected wave data are obtained every 12 cm, but the number of reflected wave data and the division interval are not limited to this. Further, the initialization processing means 22 can change the subsequent threshold according to the received angle θ of the received ultrasonic wave. For example, the threshold value is set to be reduced to (cos θ) ⁴ for the reception angle θ where the reflected wave of the ultrasonic wave is weak. As a result, the head can be detected with certainty, and misreporting can be reduced.

  Then, an initialization process for automatically detecting the bed height and setting an initial state is performed. The initialization processing step S1 is performed by a method as shown in FIG. That is, a peak value is detected for the reflected wave data in the management area 8 when the patient is absent or at bedtime, which has been measured in advance (step S11). Thereafter, the long distance mask value is set as the detection target distance (step S12). The long-distance mask value is a value set so that a distance where the patient's head should not be clearly present, such as under the bed, is excluded from the head detection process. Then, it is determined whether or not the peak value exceeds the bed threshold (threshold for recognizing that a bed exists) (step S13). If it exceeds, the distance close to the sensor by one distance (one from the reflected wave data) The reflected wave data close to the minute sensor (12 cm in this embodiment) is set as the detection target distance (step S14), and the determination in step S13 is performed again. On the other hand, if the peak value does not exceed the bed threshold, the bed height is obtained by subtracting the patient's body width from the peak value (step S15). In this way, the bed height can be automatically detected by examining the distance at which the reflected wave data disappears from the floor toward the sensor unit 9. Then, using the detected bed height as a long-distance mask, the head detection range is excluded (step S16).

  In the reflected wave data acquisition step S <b> 2, each time an ultrasonic signal is transmitted from the transmitter 2, three-dimensional reflected wave data is acquired from the ultrasonic signal received by the receiver 3. The reflected wave data is composed of plane data obtained by dividing the management area 8 into a plurality of planes in the height direction, and the signal intensity distribution of the reflected wave is recorded in each plane data. In this case as well, 25 pieces of reflected wave data are obtained every 12 cm, but the number of reflected wave data and the division interval are not limited thereto.

  The position calculation step S3 is roughly divided into a difference processing step S10, a moving average processing step S20, a peak calculation step S30, and a grouping step S40.

  In the difference processing step S10, the reflected wave data in the management area 8 when the patient is absent or at bedtime, which is measured in advance, is subtracted from the reflected wave data acquired in the reflected wave data acquisition step S1. Thereby, the stationary object included in the reflected wave data is deleted in the data, and it becomes possible to clearly distinguish the stationary object and the moving object (patient 10 or the like). For example, even if there are bed rails 23 on both sides of the patient 10, the influence of ultrasonic reflection from the bed rails 23 can be eliminated by performing the above processing.

The reflected wave data after the differential processing step also receives disturbance ultrasonic noise and ultrasonic waves from the adjacent three-dimensional ultrasonic sensor unit, which are signals unnecessary for the subsequent processing. Therefore, in the moving average processing step S20, as shown in the following equation 1, the reflected wave data is smoothed by performing the p-point simple moving average processing, and disturbance ultrasonic noise, an ultrasonic sensor installed adjacently, etc. Signals unnecessary for subsequent processing, such as unnecessary signals such as ultrasonic waves from, can be attenuated. Note that y [n] indicates data at the sampling position n after smoothing, and x [n] indicates data at the sampling position n before smoothing.

  In the plane data constituting the reflected wave data capturing the head of the patient 10, when the signal intensity is taken on the vertical axis, the head of the patient 10 appears as a convex portion as shown in FIG. 6, for example. Therefore, in the peak calculation step S30, the peak value and the position of the signal intensity are detected for each of the plurality of plane data constituting the reflected wave data after the moving average processing step S20. Then, a signal whose signal intensity exceeds a predetermined peak determination threshold value is stored as a three-dimensional position, and the patient's head or other object is present there. In this way, since the three-dimensional position of the head of the patient 10 can be defined by points, the three-dimensional position of the head of the patient 10 can be defined by the entire portion where the signal intensity is a convex part. Thus, it is possible to simplify the arithmetic processing. The stored three-dimensional position is stored as head position data in a state in which the stored three-dimensional positions are arranged in order from the one closest to the receiver 3 (sensor unit 9).

  In the present embodiment, the grouping step S40 is executed in order to determine whether or not the three-dimensional position calculated in the peak calculation step S30 captures the head of the same patient 10. This is due to the following reason. In general, when an ultrasonic wave propagating in the air hits an object such as the head of the patient 10, it is reflected by the object and returns without going straight. However, it is difficult to transmit an ultrasonic signal to be transmitted as an ideal pulse wave having no amplitude decay time (a time from an amplitude decay start time (maximum amplitude time) to an amplitude decay end time), In practice, as shown in FIG. 7, an ultrasonic signal having a predetermined amplitude decay time ΔT is often transmitted. For this reason, the ultrasonic signal reflected by the same object may be detected with a width in time from the relationship with the amplitude attenuation time ΔT of the transmitted ultrasonic wave. In this case, among a plurality of plane data constituting the reflected wave data, a plurality of peak values due to the same object are detected in several continuous plane data. Therefore, also from the viewpoint of simplifying the arithmetic processing, it is necessary to determine which peak value among a plurality of peak values is obtained by capturing the same patient 10.

  Therefore, in the grouping step S40, when a plurality of three-dimensional positions are calculated from one reflected wave data in the peak calculation step S30, these are temporarily set as temporary three-dimensional positions, and the three-dimensional positions between them are calculated. The three-dimensional position where the magnitude of the displacement vector is equal to or smaller than the grouping threshold value (predetermined value) is selected. At this time, it is determined that the selected three-dimensional position captures the head of the same patient 10. Then, the three-dimensional position (representative position) of the patient 10 is set again based on the selected three-dimensional position.

  This grouping process S40 is performed in the procedure shown in FIG. As shown in the figure, in the grouping step S40, first, among the head position data stored in the peak calculation step S30, the head three-dimensional position closest to the sensor unit 9 (provisional three-dimensional position) is used as a reference. In addition to setting the position, a new group having the three-dimensional position as a representative position is registered and the new group is set as a target group (step S401). Thereafter, it is determined whether or not the next three-dimensional position (provisional three-dimensional position) exists in the head position data (step S42).

  As a result, when the next three-dimensional position exists, the three-dimensional position is set as the target position (step S403). Then, a displacement vector from the reference position to the target position is calculated (step S404), and it is determined whether or not the magnitude of the displacement vector is equal to or smaller than the grouping threshold (step S405).

  As a result, if it is equal to or smaller than the grouping threshold, the target position is set as a new reference position, and additionally registered in the target group (step S406), and the process returns to step S402.

  On the other hand, if it is determined in step S405 that the grouping threshold has been exceeded, a new group with the target position as the new reference position and the next three-dimensional position as the representative position is registered and the new group Is set as a new target group (step S407), and the process returns to step S402. Thereafter, the above steps are repeated until it is determined in step S402 that the next three-dimensional position does not exist, and all the three-dimensional positions stored in the peak calculation step S30 are classified into a predetermined group, and the group A representative position is set for each.

  Here, the grouping threshold is set based on the amplitude decay time ΔT of the ultrasonic signal to be transmitted. This is because it is possible to predict to what extent a signal obtained by capturing the head of the same patient 10 in advance is detected from the amplitude decay time ΔT.

  In the above-described grouping step S40, the head position data in which the three-dimensional positions are arranged in order from the sensor unit 9 is processed in order from the top three-dimensional position. The closest head position is set as the representative position. Therefore, the first reflection of the object (the head of the patient 10) in time can be set as the representative position of each group. That is, even when an ultrasonic signal having a predetermined amplitude decay time is transmitted from the sensor unit 9, the same result as that obtained when an ideal pulse wave without an amplitude decay time is transmitted can be obtained.

  The processing result of the grouping step S40 processed in this way is conceptually shown in FIG. In the illustrated example, six three-dimensional positions a1, a2, a3, b1, b2, and b3 are detected in the peak calculation step S30. As a result of the grouping step S40, the group A (for example, the patient 10) including the three-dimensional positions a1 to a3 is detected. And a group B (for example, a group indicating the patient 10) consisting of the three-dimensional positions b1 to b3. In this case, the representative positions of the groups A and B are set to the head positions a1 and b1 that are closest to the sensor unit 9, respectively. In addition, the representative position of the group is not limited to the head position closest to the sensor unit 9, but for example, after all three-dimensional positions included in the head position data are classified into a predetermined group, each group It is also possible to calculate the centroid position of the three-dimensional position belonging to and set the centroid position as the representative position.

  The determination step S4 is roughly divided into a movement determination step S50, an area determination step S60, and a bed leaving operation determination step S70.

  In the movement determination step S50, a displacement vector between the representative position of the group detected from one already obtained reflected wave data and the representative position of the group detected from other newly obtained reflected wave data is calculated. Based on the magnitude of the displacement vector, a temporal change of the three-dimensional position of the patient 10 from the lying position is detected. That is, the above-described grouping step S40 is processing within the same reflected wave data, but this movement determination step S50 compares two different reflected wave data (for example, previous reflected wave data and current reflected wave data). To process.

  This movement determination process S50 is performed in the procedure shown in FIG. As shown in the figure, in the movement determination step S50, first, it is determined whether or not the group detected in the grouping step S40 exists (step S501). As a result, if a group exists, the representative position of the group is acquired and set as the target position (step S502). Then, it is determined whether or not there is a tracked group closest to the target position (step S503). If there is a tracked group, the representative position of the tracked group closest to the target position is acquired and set as a reference position (step S504). Thereafter, it is determined whether or not the magnitude of the displacement vector is equal to or smaller than a movement determination threshold value (step S505). As a result, if it is equal to or smaller than the movement determination threshold value, a new representative position of the tracked group with the target position as the determination reference position is set (step S506), and the process returns to step S501. That is, in this case, it is determined that the representative position (three-dimensional position) has moved from the reference position to the target position. On the other hand, if the movement determination threshold is exceeded, the target position is set as a new representative position of the newly tracked group (step S507), and the process returns to step S501. The movement determination threshold is set based on the movement speed of the patient 10. Specifically, the movement determination threshold is based on the maximum estimated movement amount that the patient 10 can actually move during the time from when the previous reflected wave data is acquired until the current reflected wave data is acquired. Is set.

  If it is determined in step S503 that no tracked group exists, the target position is set as a new representative position of the newly tracked group (step S507), and the process returns to step S501.

  Then, the above steps are repeated until it is determined in step S501 that the next group does not exist, and the temporal change in the representative position of each group is tracked.

  In the present embodiment, ultrasonic measurement is performed at regular intervals (100 msec). Depending on the patient's posture, the reflected wave of the ultrasonic wave may be small and the peak value may not be captured. In that case, the patient cannot be captured due to missing three-dimensional position data for a certain measurement. Once the capture is interrupted, the group remains and obstructs the tracking of the patient's head. Therefore, an erasure process is performed to erase a group (exclusion frame) in which the capture is interrupted for a predetermined time or more.

  The erasure process is performed as follows. If it is determined in step S501 whether or not the group detected in the grouping step S40 exists, it is determined whether or not a tracked group already exists in the previous reflected wave data. (Step S508). If it exists, it is determined whether or not the period (number of missing frames) in which the three-dimensional position data of the head is lost and the capture is interrupted exceeds a threshold (step S509). Is deleted (step S510), and the process returns to step S508. On the other hand, if the number of missing frames does not exceed the threshold value in step S509, the process returns to step S508 without deleting the tracked group.

  In step S508, the above steps are repeated until it is determined that no tracked group already exists in the previous reflected wave data.

  The processing result of the movement determination step S50 processed in this way is conceptually shown in FIG. This example shows a state in which the representative positions A1 and B1 are detected in the grouping step S40 for the previous reflected wave data, and the representative positions A2 and B2 are detected in the grouping step S40 for the current reflected wave data. Yes. Then, in the movement determination step S50, the representative position A1 detected in the previous reflected wave data is moved to the representative position A2 detected in the current reflected wave data, and the representative position B1 detected in the previous reflected wave data. Shows a state in which it is determined that the head has moved to the representative position B2 detected from the current reflected wave data.

  As shown in FIG. 12, when the representative position C2 obtained from the current reflected wave data is within the movement determination threshold with respect to both the representative positions A1 and B1 obtained from the previous reflected wave data. It is determined that the representative position (B1) near the displacement vector has moved.

  Further, the tracking group in which the capture is interrupted is deleted by performing an erasure process as in step S510. That is, as shown in FIG. 13, if D (tracking group in which capture is interrupted) closer to B2 exists within the threshold range of B1, it is determined that B1 has moved to D. In this case, the actual frame moves to B3 in the next frame. However, since D to B3 are out of the threshold range, they cannot be captured. Therefore, such a situation can be avoided by performing the erasure process as in step S510.

  In the area determination step S60, the management area 8 is defined by dividing it into a plurality of areas in advance, and it is detected in which area of the management area 8 the representative position of each group that has undergone the movement determination step S50 is located. Then, it is determined whether or not the group has captured the head of the patient 10. Here, as shown in FIGS. 14 and 15, the management area 8 is a lying area 11 where the head of the patient 10 lying on the bed 7 is located, and a sitting position where the head of the patient 10 sitting on the bed 7 is located. 12 and an inner alarm area consisting of a standing area 13 where the head of the patient 10 standing on the bed 7 is located, and an outer alarm area consisting of a floor area 14 formed so as to surround the bed 7. Defined in a split state. The recumbent area 11 is formed only on the pillow side of the bed 7 so as not to mistakenly recognize the foot of the patient 10 as a head. An inner alarm area may be set in a range smaller than the outer alarm area, and a non-alarm area 16 disposed between the inner alarm area and the outer alarm area may be provided on the outer peripheral side of the inner alarm area. The non-alarm area 16 is a gap between the inner alarm area and the outer alarm area, and is an area to be determined in the area determination step S60 described later (the recumbent area 11, the sitting area 12, the standing area 13, the leaving area 14). ) Does not belong. Thereby, even if a person other than the patient (such as a nurse or an attendant) approaches the bed, the false alarm can be reduced in the non-alarm area 16 without entering the inner alarm area. The range of the non-alarm area 16 can be arbitrarily set according to the installation environment or the like.

  This area determination step S60 is performed according to the procedure shown in FIG. As shown in the figure, in the area determination step S60, it is first determined whether or not the patient 10 has been captured (step S601), and if it has not been captured, the group already tracked in the previous reflected wave data. It is determined whether or not exists (step S602). If it does not exist, the process proceeds to the bed leaving movement determination step S70, and if it exists, it is determined whether or not the patient posture is equal to or greater than a threshold value (step S603). This is determined to be a person if the number of three-dimensional position information collected in the same group in the grouping step is equal to or greater than a threshold value. In the case of reflection from a person such as a patient, the number of 3D position information is large due to reflection from not only the head but also the shoulders and back, while the number of 3D position information is in the case of a stationary object such as a reading light. Take advantage of small things. If it is less than the threshold value, the process returns to step S602, and if it is greater than or equal to the threshold value, the representative position of the next group is acquired and set as the target position (step S604). Next, it is detected which area of the management area 8 the target position corresponds to (step S605). Then, it is determined whether or not the corresponding area is the recumbent area 11 of the patient 10 (step S606). As a result, when the corresponding area is equal to the recumbent area 11, the group set as the target position is regarded as the patient 10 (step S607), and the process proceeds to the bed leaving movement determination step S70. If the corresponding area is not equal to the recumbent area 11 in step S606, the process returns to step S602 and is repeated until the corresponding area becomes equal to the recumbent area 11 of the patient 10 in step S606.

  If the patient 10 has been captured in step S601, the representative position of the tracked group regarded as a patient is acquired and set as the target position (step S608). Next, it is detected which area of the management area 8 the target position corresponds to (step S609). It is determined whether or not the corresponding area has been found (step S610). If it is not found, the determination value (value used for area determination) in the previous tracked group is adopted as it is (step S613), and the bed leaving movement determination is performed. Proceed to step S70. If the corresponding area is found in step S610, it is determined whether it is the same as the determination value in the previous tracked group (step S611), and if it is the same, the duration time is updated (step S612). Then, the process proceeds to the bed leaving operation determination step. If it is different from the determination value in the previous tracked group in step S611, the determination value is updated to set the number of durations to 1 (step S614), and the process proceeds to the bed leaving operation determination step.

  In the leaving action determination step S70, after the group indicating the patient 10 is detected in the area determination step S60, it is determined whether or not the group indicating the patient 10 stays in the alarm area for a predetermined time or more. To do. The alarm area can be set separately for each of the areas 11, 12, 13, and 14. In this embodiment, the alarm area is set on the outer side of the long side of the bed. Since the patient 10 may stay in the recumbent area 11 for a predetermined time by turning over, the temporal change in the three-dimensional position of the patient 10 detected in the movement determination step S50 is changed from the recumbent area 11 to the sitting area 12. After reaching the alarm area and changing so as to stay for a predetermined time or more, it is determined that the patient 10 has started to get out of bed. Thereby, it can prevent misidentifying the start of the bed leaving operation of the patient 10.

  In determining whether to get out of bed, the patient's posture can be grasped, and a nurse call can be issued only when the patient wakes up (wake-up mode). Moreover, it can be set so that a nurse call is issued after the patient sits on the end of the bed and leaves the bed (end sitting mode).

  In the notification step S4, a notification signal is issued when the start of the bed leaving operation of the patient 10 is determined in the above bed movement determination step S60, and the start of the bed removal operation of the patient 10 is notified to the nurse or the like of the nurse center. Specifically, for example, the notification unit 6 is connected to a nurse call signal system, and when a notification signal is output from the notification unit 6, the nurse call is issued. In addition, the notification signal may be transmitted as a radio wave, and the notification signal may be received by a predetermined terminal installed in a specific place or a predetermined terminal (such as a mobile phone) possessed by a nurse or the like. Such a method of transmitting a notification signal using radio is effective as a means for notifying the start of a human body leaving operation in a general home where there is no nurse call.

  As described above, according to the bed leaving prediction apparatus 1 according to the present embodiment, since the temporal change from the lying position of the three-dimensional position of the patient 10 on the bed 7 is detected, the bed leaving on the bed 7 is detected. The start of operation can be detected. Therefore, before the patient 10 drops his / her foot from the bed 7, it is possible to predict the bed leaving of the patient 10 and detect it early. In addition, the head of the patient 10 is usually outside the futon, and the ultrasonic signal reflected by the head of the patient 10 is not inhibited by the duvet. Therefore, the start of the bed leaving movement of the patient 10 can be accurately detected by detecting the temporal change of the three-dimensional position of the patient 10 from the lying position. In addition, groups that cannot track movement for a predetermined time or longer are excluded from subsequent processing, unnecessary signals are attenuated from reflected wave data using moving average processing, and bed height is automatically measured and set. The false alarm can be reduced by reflecting the value. Furthermore, the misreporting can be reduced by changing the set value necessary for the determination of getting out of bed according to the received angle of the received ultrasonic wave, or setting the alarm area on both outer sides of the bed.

  As another embodiment, as shown in FIG. 17, there is a bed leaving prediction apparatus in which a sensor unit 9 ′ is attached to a wall 17 via an intermediate member 18. This bed leaving prediction apparatus can be configured by adding coordinate conversion means for performing coordinate conversion to the structure of the bed leaving prediction apparatus in which the sensor unit 9 is installed on the ceiling 19. The sensor unit 9 ′ may be directly attached to the wall 17 by being embedded in the wall 17. Attachment of the sensor unit 9 ′ to the wall 17 may be fixed to the wall 17, and the position and angle with respect to the wall 17 may be adjustable. Also in this case, an inner alarm area is set in a range smaller than the range constituted by the outer alarm area and the wall 17, and on the outer peripheral side of the inner alarm area, between the inner alarm area and the outer alarm area, and A non-alarm area 16 may be provided that is disposed between the inner alarm area and the wall 17. The non-alarm area 16 is a gap between the inner alarm area and the outer alarm area, and a gap between the inner alarm area and the wall 17, and is an area to be determined in the area determination step S60 (the recumbent area 11, the sitting area 12). , Standing area 13 and getting-off area 14). Thereby, even if a person other than the patient (such as a nurse or an attendant) approaches the bed, the false alarm can be reduced in the non-alarm area 16 without entering the inner alarm area. The range of the non-alarm area 16 can be arbitrarily set according to the installation environment or the like.

  The above coordinate conversion will be described in detail. For example, assume that the coordinate axes for the wall-mounted sensor unit 9 'shown in FIG. 17 are shown in FIG. The Z ′ axis is a direction perpendicular to the surface of the wall-mounted sensor unit 9 ′ that transmits and receives ultrasonic waves, and is directed to the lower right of the figure. The Y ′ axis is a direction along the plane of the ultrasonic wave transmission and reception in the wall-mounted sensor unit 9 ′ and the direction along the plane of the drawing, and is the upper right direction in the figure. The X ′ axis is the direction from the back of the paper surface to the front surface in a direction perpendicular to the paper surface.

  For example, the coordinate axes for the ceiling-mounted sensor unit 9 shown in FIG. 14 are shown in FIG. The Z-axis is a direction perpendicular to the surface of the ceiling-mounted sensor unit 9 that transmits and receives ultrasonic waves, and is directed downward in the figure. The Y-axis is the direction along the plane of the sensor unit 9 that transmits and receives ultrasonic waves and the left side of the figure. The X axis is the direction from the front to the back of the paper in a direction perpendicular to the paper.

  Consider a case where a wall-mounted sensor unit 9 ′ having a coordinate axis shown in FIG. 18A is installed on the wall 17 as shown in FIG. 17. The setting ranges of the areas 11 to 15 for the area bed 7 in the bed leaving prediction apparatus using the wall-mounted sensor unit 9 'are the same as those in the above embodiment. It is assumed that the ceiling-mounted sensor unit 9 having the coordinate axes shown in FIG. 18B is installed on the ceiling 16 in the room shown in FIG. The setting ranges of the areas 11 to 14 for the area bed 7 in the bed leaving prediction apparatus using the ceiling-mounted sensor unit 9 are also the same as those in the above embodiment.

  A mathematical coordinate conversion formula is created so that the coordinate value on the coordinate axis of the wall-mounted sensor unit 9 ′ for the arbitrary position in the room becomes the coordinate on the coordinate axis of the ceiling-mounted sensor unit 9. If this coordinate conversion formula is used, the algorithm in the bed prediction apparatus using the ceiling-mounted sensor unit 9 can be used in the bed prediction apparatus using the wall-mounted sensor unit 9 '. That is, by adding a coordinate conversion means for performing coordinate conversion using the coordinate conversion formula and a coordinate conversion step to the bed leaving prediction device and the bed leaving prediction method using the ceiling installation type sensor unit 9, respectively, the wall installation type sensor unit 9 The bed leaving prediction device and bed leaving prediction method using 'can be easily configured.

  Since the wall-mounted sensor unit 9 ′ is easier to change the installation work and the installation location than the ceiling-mounted sensor unit 9, the bed leaving prediction apparatus and the bed leaving prediction method using the wall-mounted sensor unit 9 ′. Improves convenience and acceptability.

  Further, the wall-mounted sensor unit 9 ′ may be attached to a base such as a tripod instead of the wall 17 and the base may be movably disposed on the floor. In this case, since it is easier to change the installation work and the installation location, the bed leaving prediction apparatus and bed leaving prediction method using the wall-mounted sensor unit 9 ′ further improve convenience and acceptability. The wall-mounted sensor unit 9 ′ may be directly attached to the table, or may be attached to the table via an intermediate member. Further, the mounting of the wall-mounted sensor unit 9 ′ on the table may be fixed to the table, or the position and angle with respect to the table may be adjustable. Further, since the base on which the wall-mounted sensor unit 9 ′ is mounted is an extendable base, a base that can be bent, or the like, the wall-mounted sensor unit 9 ′ can be adjusted in position and angle with respect to the floor. Good.

  In addition, this invention is not limited to said embodiment at all, In the range which does not deviate from the summary of this invention, it can implement with a various form. For example, in the above embodiment, the case of predicting the bed leaving motion of the patient 10 has been described. However, the present invention can be similarly applied to a person other than the patient 10 in a nursing home or a general home.

  In the above-described embodiment, the case where only the head position of the patient 10 is tracked and detected has been described. Thus, the start of the bed leaving operation of the patient 10 may be determined. Further, in the embodiment, the recumbent area 11 is formed only on the pillow side of the bed 7 so as not to mistakenly recognize the foot of the patient 10 as the head, but can be extended to the end of the bed on the foot side. .

  In the determination of S606, not only the recumbent area but also the recumbent area or the sitting area may be determined. This is a measure to prevent the patient himself / herself from being regarded as a stationary object and difficult to detect in the recumbent area when the patient sleeping management area 8 is measured in the initialization process.

  Furthermore, in the above-described embodiment, a case where the management area 8 including one bed 7 is managed by one sensor unit 9 (one set of transmitter 2 and receiver 3) has been described. You may make it manage using the unit 9. FIG. In this case, it is preferable that an ultrasonic signal transmitted from each sensor unit 9 does not interfere with an ultrasonic signal transmitted from another sensor unit 9. As a countermeasure, it is possible to transmit ultrasonic waves sequentially from each sensor unit 9 with a predetermined time difference so that the ultrasonic waves transmitted from each sensor unit 9 do not interfere. As another countermeasure, for example, the wavelength of the ultrasonic signal transmitted from each sensor unit 9 is changed, and the received ultrasonic signal is subjected to filtering processing for each wavelength component of the ultrasonic signal transmitted from each sensor unit 9. It is possible to measure separately. Further, the management area 8 including the plurality of beds 7 may be managed by one sensor unit 9. In this case, in order to expand the management area 8, it is preferable to use an omnidirectional transmitter as the transmitter 2.

  In addition, a transmitter for transmitting an ultrasonic signal is separately attached on the bed 7 (for example, the four corners of the bed 7), and the ultrasonic signal is transmitted toward the sensor unit 9 at a predetermined timing, whereby the determination unit 5 The position of the bed 7 may be recognized.

  Furthermore, the prediction of the bed leaving operation is not limited to manual operation, and may be started automatically. In the case of automatically starting bed prediction, for example, an infrared sensor is separately attached to a stationary system such as a ceiling. This infrared sensor detects infrared rays of a specific wavelength emitted from the human body such as the patient 10 from the bed 7. When infrared rays are detected over a predetermined time, a signal is output from the infrared sensor to the determination means 5 to automatically start getting out of bed. Furthermore, the bed leaving detection may be temporarily terminated when the notification means 6 issues a notification signal.

It is a block diagram which shows the bed leaving prediction apparatus which concerns on one Embodiment of this invention. It is a perspective view which shows typically the attachment aspect of the bed leaving prediction apparatus which concerns on this embodiment. (A) is a top view which shows typically an example of the receiver used for the bed leaving prediction apparatus which concerns on this embodiment, (b) is a top view which shows typically another example of a receiver. It is. It is a flowchart which shows the bed prediction procedure by the bed prediction apparatus which concerns on this embodiment. It is a flowchart which performs automatic detection of bed height. It is a graph which shows an example of reflected wave data. It is a graph which shows typically an example of the waveform of the ultrasonic signal to transmit. It is a flowchart which shows the grouping process included in the bed leaving prediction procedure of FIG. It is a figure which shows notionally the process result of a grouping process. It is a flowchart which shows the movement determination process included in the bed leaving prediction procedure of FIG. It is a figure which shows notionally the process result of a movement determination process. It is a figure which shows notionally the process result of a movement determination process. It is a figure which shows notionally the process result of a movement determination process. It is a side view which shows the division | segmentation aspect of the management area seen from the side of the bed. It is a side view which shows the division | segmentation aspect of the management area seen from the front of the bed. It is a flowchart which shows the area determination process included in the bed leaving prediction procedure of FIG. It is a figure which shows the bed leaving prediction apparatus which concerns on another embodiment of this invention. FIG. 6A is a diagram illustrating an example of coordinate axes of the bed leaving prediction apparatus. FIG. 6A illustrates a case where a wall-mounted sensor unit is used, and FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Bed leaving prediction apparatus 2 Transmitter 3 Receiver 4 Position calculating means 5 Determination means 6 Notification means 7 Bed 8 Management area 9 Sensor unit 10 Patient 20 Moving average processing means 21 Erasing means 22 Initialization processing means S1 Acquisition of reflected wave data Step S2 Position calculation step S3 Determination step S4 Notification step

Claims (13)

A transmitter that continuously transmits an ultrasonic signal to a management area including a bed while attached to a stationary system;
A receiver for receiving the ultrasonic signal reflected in the management area in a state attached to a stationary system;
The three-dimensional position of the human head on the bed or other objects is sometimes determined from the reflected wave data acquired at predetermined intervals by the receiver in response to the ultrasonic signal transmitted from the transmitter. Position calculation means for calculating every moment and classifying this three-dimensional position into a group of a plurality of human heads or a group of other objects;
A determination unit that detects a temporal change from the lying position of the three-dimensional position of the human body from the calculation result of the position calculation unit, and determines whether or not the human body on the bed has started leaving the bed based on the detection result When,
A notification unit that issues a notification signal when the determination unit detects the start of a human bed leaving operation;
The bed leaving prediction apparatus characterized in that the determination means includes a erasure means for removing a group of heads that cannot track movement for a predetermined time or longer from the subsequent processing.   2. The bed leaving prediction apparatus according to claim 1, wherein the position calculating means includes moving average processing means for attenuating unnecessary signals from reflected wave data using moving average processing.   The bed leaving prediction apparatus according to claim 1 or 2, further comprising an initialization processing means for automatically detecting a bed height and setting an initial state.   The set value required to determine whether or not the human body on the bed has started the bed leaving operation is changed according to the received angle of the received ultrasonic wave. The bed leaving prediction apparatus according to item 1. The bed leaving prediction apparatus according to claim 4, wherein the set value is (cos θ) ⁴ with respect to the receiving angle θ.   An outside alarm area is set at least on the outer side of the long side of the bed, and the determination means determines whether to leave the bed when a human head stays in the alarm area for a predetermined time or more. The bed leaving prediction apparatus according to any one of claims 1 to 5.   An inner alarm area that is smaller than the outer alarm area is set above the bed, and a non-alarm area is provided between the inner alarm area and the outer alarm area on the outer periphery of the inner alarm area. The bed leaving prediction apparatus according to claim 6.   The bed receiving prediction apparatus according to any one of claims 1 to 7, wherein the wave receiver is an array sensor in which a plurality of wave receiving elements are arranged. An ultrasonic signal is continuously transmitted from a transmitter attached to a stationary system to a management area including a bed, and the ultrasonic signal reflected in the management area is received by a receiver attached to the stationary system. Then, a reflected wave data acquisition step of acquiring reflected wave data corresponding to the ultrasonic signal transmitted from the transmitter every predetermined time;
From the reflected wave data acquired in the reflected wave data acquisition step, a three-dimensional position of a human head or other object on the bed is calculated from moment to moment, and the three-dimensional position is calculated as a group of a plurality of human heads or A position calculation step for classifying into other object groups;
A determination step of detecting a temporal change from the lying position of the three-dimensional position of the human body from the calculation result of the position calculation step, and determining whether or not the human body on the bed has started a bed leaving operation based on the detection result When,
A notification step of issuing a notification signal when the determination step detects the start of a human bed leaving operation;
The determination step is characterized by performing a erasure process in which a group of heads whose movement cannot be tracked for a predetermined time or more is excluded from the subsequent processes.   10. The bed leaving prediction method according to claim 9, wherein the position calculating step includes a moving average processing step of attenuating unnecessary signals using moving average processing. The bed leaving prediction method according to claim 9 or 10, further comprising an initialization process step of automatically detecting a bed height and setting an initial state.   The set value required to determine whether or not the human body on the bed has started to move out of the bed is changed according to the received angle of the received ultrasonic wave. The method for predicting getting out of bed. The bed leaving prediction method according to claim 12, wherein the set value is (cos θ) 4 with respect to the receiving angle θ.

Images