JP6183839B2 - Action prediction system, action prediction device, action prediction method, action prediction program, and recording medium recording action prediction program - Google Patents

Description

  The present invention relates to a system, an apparatus, a method, a program for predicting the behavior of a subject, and a recording medium on which the program is recorded.

  At present, in an aging society, various systems have been devised or used for monitoring and monitoring in the care of subjects such as elderly people.

  For example, a system using a surveillance camera or microphone has been proposed in order to prevent subjects sleeping in a bed from waking up at night and causing drowning and falling accidents, and a sensor incorporating a piezoelectric element beside the bed. A system in which an alarm is sounded when a mat is installed and a subject puts his / her foot when getting out of bed is used in nursing homes.

  However, systems using surveillance cameras and microphones are not easy to install in toilets due to privacy issues, and the cameras are always directed at the subject even in places where such privacy issues do not occur. It gave a sense of anxiety and oppression.

  In addition, the system using the sensor mat as described above can play an effective role in the case where a fall accident has already occurred beside the bed even if the above-mentioned alarm sounds and the caregiver rushes. There wasn't. In addition, when the subject lies on the bed again after placing his feet on the mat, care is not necessarily required, so this system has many types of malfunctions for caregivers. It is evaluated as a thing.

  From such a situation, as shown in Non-Patent Documents 1 to 4, a system for grasping the sleeping state of the subject by monitoring the intensity change of the radio wave transmitted from the RFID (Radiofrequency Identification) tag (hereinafter, “ RFID tag system ") has been devised.

  On the other hand, Patent Document 1 discloses an apparatus, a system, and the like that predict an abnormality of a monitoring target by processing measured time-series data using an average value and a standard deviation.

JP 2010-2111440 A

Ryosaku Blacksmith, 2 others, “Proposal of a method for detecting a fall state in a closed space using an RFID tag system”, Transactions of Information Processing Society of Japan, Information Processing Society of Japan, March 2010, Vol. 51, No. 3, p. 1140 Ryosaku Blacksmith, 4 others, “Proposal of recording method of sleeping state using Radiofrequency Identification (RFID) tag system”, Tokyo Branch Academic Meeting Program / Abstract, Japan Society for Health Care Management, February 25, 2012, 12th , P.64 Ryosaku Blacksmith, 2 others, "Proposal of a method for analyzing patient's moving body in a closed space using Radiofrequency Identification (RFID) tag system", Journal of Medical Quality and Safety, Medical Quality and Safety Society, February 29, 2012 Sun, Vol. 7, No. 1, p.48-56 Ryosaku Blacksmith, 4 others, “Recording of bedtime with RFID tags: urinary intention, discomfort, insomnia, getting out of bed, fall estimation”, Journal of Medical Quality and Safety, Medical Quality and Safety Society, October 25, 2012, Volume 7, Supplement No., p.348

  In the RFID tag system described above, data is continuously acquired for a long time because of the property of monitoring the subject. As a result, at the stage before the person who caused the fall accident gets up from the bed at night. It became clear that the small operation which is not continued for several hours. This small motion that is not rising is almost the same as the normal rolling operation as the output value of the sensor, so the conventional method of detecting the rising motion cannot be distinguished from the normal rolling operation. there were.

  The present invention has been made to solve the above problems, and provides a system, an apparatus, a method, a program, and a recording medium on which the program can be recorded, which can predict a subject's behavior quickly and accurately with a simple configuration. The purpose is to do.

  In order to solve the above-mentioned problem, it is means for predicting the behavior of a subject, and the magnitude of the received signal or the magnitude of the intensity change of the received signal exceeds a predetermined first threshold value. Means for measuring the frequency and notifying that the measured frequency is equal to or higher than a predetermined second threshold value are provided.

  Further, as another means for solving the above-mentioned problem, it is a means for predicting the behavior of the subject, and the first of the intensity change according to the magnitude of the signal monitored in advance or the magnitude of the intensity change of the signal. And determining the number of times that the magnitude of the signal received at a predetermined first time or the magnitude of the intensity change of the received signal exceeds the first threshold, When the ratio of the number of times counted in the first time in the past is equal to or greater than a predetermined second threshold, the time for measuring the magnitude of the signal or the magnitude of the intensity change is Limited to a predetermined second time shorter than one time, the number of times the magnitude of the signal or intensity change measured at the second time exceeds the first threshold and the second time A third threshold with a predetermined ratio to the total number of measurements in time To provide a means for notifying that it has become a top.

  Further, as another means for solving the above-mentioned problem, it is a means for predicting the behavior of the subject, and the magnitude of the signal or the intensity change of the signal monitored within a predetermined time is set to a plurality of Classify into classes, count the frequency generated in the monitoring for each class, and the ratio with the minimum frequency for each class aggregated in the past time depends on the maximum value of the ratio Identify a class that exceeds the defined first threshold, set a value greater than the maximum value of the ratio for each class specified as the second threshold, and The ratio with the minimum value calculated for each class specified by aggregating the magnitude or magnitude of the signal strength change for each class is the second value in at least one of the specified classes. Threshold To provide a means for notifying that it has exceeded.

  Further, as another means for solving the above-mentioned problem, it is a means for predicting the behavior of the subject, and the magnitude of the signal or the intensity change of the signal monitored within a predetermined time is set to a plurality of Classify into classes, count the frequencies generated in the monitoring for each class, and determine the ratio with the minimum frequency for each class aggregated in the past time according to the maximum value of the ratio The class exceeding the specified first threshold is specified, the sum of the maximum values of the ratios for each specified class is set as the second threshold, and the magnitude of the signal monitored within the above time Or means for notifying that the sum of the ratio with the minimum value calculated for each class specified by counting the magnitude of the signal intensity change for each class exceeds the second threshold value. provide.

  Further, as another means for solving the above-mentioned problem, it is a means for predicting the behavior of the subject, and the magnitude of the signal or the intensity change of the signal monitored within a predetermined time is set to a plurality of Classify into classes, count the frequencies generated in the monitoring for each class, and the ratio with the minimum value of the frequency for each class that was aggregated in the past time depends on the maximum value of the ratio A class that exceeds a predetermined first threshold value is specified, and means for monitoring the ratio of the sum of the ratios for each specified class and the sum of the ratios for all the classes is provided.

  According to the present invention, it is possible to quickly and accurately extract a change to be noticed from changes in the intensity of a signal corresponding to the behavior of the subject. Can be predicted.

It is a figure which shows the structure of the action prediction system which concerns on embodiment of this invention. It is a flowchart for demonstrating 1st embodiment of the behavior prediction method implement | achieved by the behavior prediction system shown by FIG. FIG. 3 is a first diagram for explaining a method of determining a threshold value in step S4 shown in FIG. FIG. 4 is a second diagram for explaining a method for determining a threshold value in step S4 shown in FIG. It is a block diagram which shows the example of the action prediction apparatus which implement | achieves the action prediction method which concerns on Embodiment 1. FIG. It is a flowchart for demonstrating 2nd embodiment of the behavior prediction method implement | achieved with the behavior prediction system shown by FIG. It is a block diagram which shows the example of the action prediction apparatus which implement | achieves the action prediction method which concerns on Embodiment 2. FIG. It is a flowchart for demonstrating 3rd embodiment of the behavior prediction method implement | achieved with the behavior prediction system shown by FIG. It is a figure for demonstrating the case where a class value and a count value have an exponential relationship. It is a figure for demonstrating the automatic learning of the motion histogram in step S6 shown by FIG. It is a figure for demonstrating the setting of the threshold value in step S8 shown by FIG. It is a block diagram which shows the example of the action prediction apparatus which implement | achieves the action prediction method which concerns on Embodiment 3. FIG. It is a block diagram which shows the other example of the action prediction apparatus which implement | achieves the action prediction method which concerns on Embodiment 3. FIG.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In the drawings, the same reference numerals indicate the same or equivalent parts.
As mentioned above, at the stage before the person who had a fall accident got up from the bed at night, a small action that did not get up continued for several hours, and it was not necessarily continuous. However, it was revealed that the operation was performed many times in a certain time. This small operation that is not rising is almost the same as the normal turning operation as an output value of the sensor. Therefore, the conventional method for detecting the rising operation cannot be distinguished from the normal turning operation.

  In such a case, if a call is made when a small motion is detected at a stage before the rising motion is performed in order to prioritize the safety of the subject, the number of times that the caregiver is called will jump compared to the number of detections of the rising motion. (Increase in caregiver burden).

  In addition, in order to distinguish from a normal roll-over operation, confirming the subject by some means (silently, speaking) causes a side effect of causing a result of disturbing sleep.

  For this reason, even if it is an operation that is indistinguishable from a normal roll-over operation on a one-time basis, the occurrence of it in a discrete but many times within a certain time is a precursor to the subsequent accident. Therefore, there is a need to find out a method for mechanically extracting when to issue a report and a method for expressing the current sleeping state as an objective numerical value.

  Therefore, in the following, a behavior prediction system, a behavior prediction device, a behavior prediction method, a behavior prediction program, and a recording medium that records the program according to an embodiment of the present invention for solving such a problem will be described in detail.

  FIG. 1 is a diagram showing a configuration of an action prediction system according to an embodiment of the present invention. As shown in FIG. 1, the behavior prediction system according to the embodiment of the present invention is installed on the back side of a bed 1 placed in a room 10 and periodically, for example, weakly indicated by a broken line every second. A radio frequency identification (RFID) tag 2 that transmits a radio wave, a receiver 3 that is installed at a fixed point such as near the ceiling of the room 10 and has an antenna 4 that receives a radio wave transmitted from the RFID tag 2, and a receiver 3 The behavior prediction apparatus 20 connected with the wired cable 5 is provided. The receiver 3 and the behavior prediction device 20 may be connected by wireless communication.

  The behavior prediction apparatus 20 is a computer including a CPU 6, a memory 7, and a display unit 8, and the CPU 6, the memory 7, and the display unit 8 are connected to a common bus 9 connected to the wired cable 5.

  In the behavior prediction system, one RFID tag 2 is installed on the back side of the bed 1, but the RFID tag 2 is located at a position where the intensity of the radio wave received by the receiver 3 is clearly changed by the movement of the subject. For example, it may be installed on the wall of the bed.

  Moreover, the RFID tag 2 which comprises an action prediction system does not need to be one, and plural may be sufficient. In this case, each intensity of the radio wave transmitted from the plurality of RFID tags 2 is handled as a set of information (intensity data described later).

  Further, the behavior prediction system predicts the behavior of the subject by measuring the intensity change by utilizing the fact that the strength changes when the radio wave transmitted from the RFID tag 2 is disturbed by the subject. However, instead of such a change in the intensity of the radio wave, the output of a sensor such as a capacitive sensor, a Doppler radar, or a load sensor whose output changes according to the movement of the subject, or a moving body sensor such as an acceleration sensor attached to the subject. The output may be measured.

  The sensors such as capacitive sensors, Doppler radars, load sensors, etc., in addition to the types in which the magnitude of movement is not required unless the difference in output is taken, the signals in the behavior prediction system using the RFID tag 2 are also shown. Since there is a type that outputs a signal corresponding to the magnitude of intensity change, in the behavior prediction system using these sensors, by measuring the difference in output or the magnitude of the output from the sensor, The same function as the behavior prediction system using the RFID tag 2 can be realized.

  In these systems, one-dimensional data called a signal indicating the intensity change in accordance with the behavior of the subject is subjected to data processing. Therefore, regardless of the type of sensor, the subject can be used with the same algorithm regardless of the type of sensor. Can be predicted.

  In the behavior prediction system shown in FIG. 1, since the human body has a predetermined dielectric constant, whether or not a human is interposed between the RFID tag 2 and the antenna 4 or not depends on the mode. The intensity of the radio wave received by the antenna 4 changes.

  For example, as shown in FIG. 1, when a subject lies on his / her back on the bed 1 and the subject is on the path until the radio wave transmitted from the RFID tag 2 reaches the antenna 4, Since the radio wave is disturbed by the subject, the intensity of the radio wave received by the antenna 4 changes from the case where the subject does not intervene on the path.

Even when the subject is lying on the bed 1, the intensity of the radio wave received by the antenna 4 varies depending on whether the subject is sleeping in a supine posture or a lying posture. Is different. Specifically, when the subject is sleeping on his / her back in bed 1, the intensity of the radio wave received by the antenna 4 is higher than that when the subject is sleeping in a sideways posture. This is smaller than when the subject is sleeping in a sideways posture.
In the following, an embodiment according to a behavior prediction method realized by the behavior prediction system will be described.

[Embodiment 1]
FIG. 2 is a flowchart for explaining a first embodiment of a behavior prediction method realized by the behavior prediction system shown in FIG. Hereinafter, the behavior prediction method according to the first embodiment will be described in detail with reference to FIG.

  First, in step S1, the behavior prediction apparatus 20 measures the intensity of radio waves transmitted from the RFID tag 2 and received by the antenna 4 of the receiver 3 every second, and indicates the intensity obtained every second by this measurement. Data (hereinafter referred to as “intensity data”) is stored in the memory 7.

  Next, in step S2, the CPU 6 calculates the magnitude (change amount) of the intensity change of the radio wave in one second in real time by taking the difference between the continuous intensity data stored in the memory 7.

  In step S <b> 3, the CPU 6 is calculated in a state in which the subject has a normal life in the room 10, for example, in a state of performing daily activities such as lying on the bed 1, turning over and getting up. The amount of change is monitored in advance, and a frequency distribution diagram (hereinafter referred to as “reference histogram”) showing the relationship with the frequency of occurrence of the amount of change is created.

  It should be noted that when the subject has a normal life as described above, the amount of change tends to increase with a small value because the subject is almost asleep or unmanned. Although it is desirable to be zero, there may be cases where it does not become zero due to a measurement error due to environmental fluctuations such as the temperature and humidity of the room 10.

  In step S4, the CPU 6 determines from the reference histogram a threshold value for the amount of change that serves as a reference for detecting or recognizing the specific movement of the subject.

  FIG. 3 is a first diagram for explaining a method of determining the threshold value. Here, the graph shown by the dots in FIG. 3 shows an example of the reference histogram. The horizontal axis is the magnitude of the intensity change (the above change amount), and the vertical axis is the frequency (appearance) obtained in step S3. Number of times). As shown in FIG. 3, the reference histogram based on the actual measurement values is composed of two regions: a region with a small change amount that can be linearly approximated and a region with a large change amount that has only a small number of appearances.

  Then, when the region where the amount of change in the reference histogram is small has a linear shape like the graph GP1 shown in FIG. 3, the threshold value is linearly displayed on the graph GP1 as indicated by a one-dot chain line. The amount of change C1 at the intersection with the horizontal axis when extrapolated can be determined.

FIG. 4 is a second diagram for explaining a method of determining the threshold value. Here, the horizontal axis shown in FIG. 4 indicates the magnitude of the intensity change (the above change amount), and the vertical axis indicates the logarithm of the frequency (number of appearances). At this time, when a region with a small amount of change in the reference histogram has an exponential shape, 1 is added to the number of occurrences of each amount of change and a logarithm is used to illustrate the relationship with the amount of change. In the coordinate system shown in (1), a linear graph GP2 can be obtained.
Therefore, the change amount C2 at the intersection of the graph GP2 and the horizontal axis where the number of appearances is 0 (the value on the vertical axis is 1) can also be determined as the threshold value.

  Next, in step S5, the CPU 6 counts the frequency of occurrence (number of appearances) for each change amount for the change amount exceeding the threshold value determined in step S4 (hereinafter referred to as “motion histogram”). In real time.

  That is, the CPU 6 sets the change amount calculated by the method described in step S2 as a new aggregation target each time (every second), and at the same time, among the change amounts that are the aggregation target. The value that was targeted earlier is excluded from the new target for aggregation. In this way, the CPU 6 constantly generates the latest motion histogram at a predetermined time, for example, one hour, and stores it in the memory 7.

  Then, in step S6, the CPU 6 monitors the motion histogram, and when at least one of the frequencies for each class value of the histogram is equal to or higher than a predetermined threshold level as a level of caution, an abnormality occurs in the subject. The user of this system is notified by predicting that there is a possibility and displaying the fact on the display unit 8. In addition to the display by the display unit 8 as described above, this notification may be performed by a method of sounding a warning sound from a speaker (not shown) included in the behavior prediction apparatus 20.

  Here, the frequency for each class value of the motion histogram is monitored because the amount of change differs depending on the behavior of the subject. For example, the behavior such as turning over in the bed 1 by the subject showing the insomnia to be noticed is This is because the amount of change to be noted is not necessarily a large value because the movement is smaller than the movement up and down. Small movements are precursors of large movements (small movements occur earlier in time) and small movements last longer than large movements (small movements are observed until the end) This is because there is a characteristic that there is a large time margin for reporting to the Internet.

  In addition, as the level of caution, for example, the accident may occur empirically based on a motion histogram record measured in a case where the subject caused a fall accident due to insomnia and stored in the memory 7. A value that is considered to be high is set in advance. Note that, as the level of caution, it may be possible to set in advance a value that is at least three times the maximum value obtained for each class of the motion histogram obtained in a normal life state.

  According to the behavior prediction method according to the first embodiment of the present invention as described above, since the frequency of performing an action to be observed by monitoring the sleeping state of the subject can be measured in real time, for example, the action immediately before getting up It is possible to predict in advance signs of getting out of the bed without detecting the above.

  From this, for example, when the subject is an elderly person and needs care, the caregiver can rush to help with time.

Moreover, the behavior prediction method according to the first embodiment can be realized by, for example, the behavior prediction device shown in FIG.
The behavior prediction apparatus 30 illustrated in FIG. 5 includes an intensity change calculation unit 31, a measurement unit 32, a storage unit 33, and a notification unit 34.

Here, the intensity change calculation unit 31 calculates the intensity change of the received signal. The storage unit 33 also stores predetermined first and second threshold values.
Then, the measurement unit 32 compares the magnitude of the signal intensity calculated by the intensity change calculation unit 31 with the first threshold value stored in the storage unit 33, and the magnitude of the signal intensity is the first. Measure the frequency of exceeding the threshold.
At this time, the notification unit 34 outputs a notification signal when the frequency measured by the measurement unit 32 becomes equal to or higher than the second threshold value stored in the storage unit 33.

  The measuring unit 32 determines the first threshold value according to the magnitude of the intensity change of the signal monitored in advance, and stores the determined first threshold value in the storage unit 33. May be.

[Embodiment 2]
FIG. 6 is a flowchart for explaining a second embodiment of the behavior prediction method realized by the behavior prediction system shown in FIG. Hereinafter, the behavior prediction method according to the second embodiment will be described in detail with reference to FIG.

  Here, step S1 to step S4 shown in FIG. 6 are the same as step S1 to step S4 shown in FIG. 2 of the behavior prediction method according to the first embodiment, and the description thereof will be omitted. .

  In step S5 shown in FIG. 6, the CPU 6 monitors the amount of change in units of one day (24 hours), and the frequency of the amount of change exceeding the threshold value every day (hereinafter referred to as “movement frequency”). Sum up).

Next, in step S6, the CPU 6 calculates a ratio between the frequency totaled in step S5 and the minimum value of the frequency of motion totaled on each day of the past fixed period, for example, one week.
In step S7, when the ratio calculated in step S6 is equal to or higher than a predetermined threshold as the level of caution, the CPU 6 starts monitoring the subject from the previous 24 hours. Also, it is limited to a short predetermined range, for example, 9 hours from 21:00, which is the sleeping time of the subject, to 6:00 on the next day.

  Limiting the time for monitoring the subject in this way is generally that the subject's behavior is often limited to a predetermined time. Therefore, by limiting the monitoring time, the ratio of the motion frequency to be noticed to the total frequency is increased. This is because the detection sensitivity of the movement frequency can be increased.

Next, in step S8, the CPU 6 totals the frequency of movement in the range limited in step S7.
In step S9, the CPU 6 calculates the ratio of the movement frequencies totaled in step S8 to the total frequencies totaled for the above 9 hours.

  In this manner, in step S10, the CPU 6 determines that there is a possibility that the subject has changed when the ratio calculated in step S9 is greater than or equal to a predetermined threshold value as a danger level, such as insomnia. The user of this system is notified by predicting that it is in a state and displaying that fact on the display unit 8.

  In addition to the display by the display unit 8 as described above, this notification may be performed by a method of sounding a cautionary sound from a speaker (not shown) included in the behavior prediction apparatus 20. 1 is the same as the behavior prediction method according to 1.

  According to the behavior prediction method according to the second embodiment of the present invention as described above, when it is determined that there is a possibility that the subject has changed due to comparison with data measured in the past, Since the time for monitoring the subject is limited, the detection sensitivity of the motion frequency to which attention should be paid is relatively increased, so that the behavior of the subject can be predicted more quickly and accurately than before.

  Note that the behavior prediction method according to the second embodiment can be realized by, for example, the behavior prediction device 40 illustrated in FIG.

  7 includes an intensity change calculation unit 41, a calculation unit 42, a storage unit 43, a measurement time limiting unit 44, and a notification unit 45.

Here, the intensity change calculation unit 41 calculates the intensity change of the received signal. The storage unit 43 stores first to third threshold values in advance.
Then, the calculation unit 42 uses the first threshold value stored in the storage unit 43 as the magnitude of the intensity change calculated by the intensity change calculation unit 41 for the signal received at the predetermined first time. And the ratio of the number of times counted in the first time in the past to the minimum value is calculated. In addition, the calculating part 42 shall memorize | store the minimum value of the frequency | count counted in said 1st time in the past.

  In addition, the measurement time limiting unit 44 determines the magnitude of the intensity change when the ratio calculated by the calculation unit 42 is equal to or greater than the second threshold stored in the storage unit 43. The time to measure is limited to a predetermined second time shorter than the first time.

  Then, the value of the ratio between the number of times the intensity change calculated by the intensity change calculation unit 41 in the second time exceeds the first threshold and the total number of measurements in the second time is calculated by the calculation unit 42. The calculated value is supplied to the notification unit 45. At this time, the notification unit 45 outputs a notification signal when the value supplied from the calculation unit 42 becomes equal to or greater than the third threshold value stored in the storage unit 43.

[Embodiment 3]
FIG. 8 is a flowchart for explaining a third embodiment of the behavior prediction method realized by the behavior prediction system shown in FIG. Below, the said action prediction method which concerns on Embodiment 3 is demonstrated in detail using FIG.

  Here, steps S1 to S3 shown in FIG. 8 are the same as steps S1 to S3 shown in FIG. 2 of the behavior prediction method according to the first embodiment, respectively, and thus description thereof is omitted. .

  In step S4 shown in FIG. 8, the CPU 6 monitors the change amount in units of a predetermined first time, for example, overnight (9 hours), and changes the frequency of the change amount every night. Aggregate by quantity.

  Specifically, the magnitude of the intensity change of the received signal is classified into a plurality of classes, and the frequency is counted for each class value. And the minimum value of the count value for every class value is recorded (learning of the minimum value).

  Next, in step S5, the CPU 6 adds the value obtained by adding 1 to the count value for each class value and the value obtained by adding 1 to the minimum value for each class value for every evening used for learning the minimum value. Normalization is performed by calculating the divided count ratio (creation of a normalized histogram).

  Here, when the relationship between the class value and the count value is exponential, when the count value of the relatively low class value decreases and the count value of the high range value increases by the decrease, The count ratio for each class value varies greatly in a high range of class values.

  Specifically, as shown in FIG. 9, the frequencies derived from the environment are mostly concentrated in the lower class values, and the frequencies derived from human movement are distributed to the higher class values than those derived from the environment. To do.

  At this time, referring to the normalized histogram of each night, as shown in FIG. 10, the normalized frequency (count ratio) derived from the above-described human movement is a mountain shape.

  Therefore, data of a class value range (hereinafter referred to as “first range”) larger than a value automatically calculated based on the peak value of the histogram is represented by the motion histogram according to the first embodiment. In order to play an alternative role when it is difficult to determine the first threshold value used in the calculation, in step S6, the CPU 6 specifies the first range (automatic learning of the range of the motion histogram).

Here, the “automatically calculated value” is, for example, 50% (Rt / 2) of the peak value Rt, as shown in FIG. Alternatively, the peak value exp (-1) may be used. Note that the range of the “first range” affects the detection sensitivity. In general, the closer the “automatically calculated value” is to a peak value, the lower the detection sensitivity. On the other hand, it has the property that noise becomes easier to pick up as the value is farther from the peak value.

  In step S7, the CPU 6 records the maximum value of the count ratio in the first range of the data for each night for each class value (learning of the maximum value group).

  Next, in step S8, the CPU 6 determines a value larger than the maximum value of the count ratio recorded in step S7, for example, a value twice the maximum value for each class value, such as a value indicated by a black circle in FIG. Set as the threshold value. The greater the multiplier to the maximum value, the lower the sensitivity, and the smaller the multiplier, the greater the noise.

  Through the above learning process, in step S9, the CPU 6 again calculates the count ratio for each class in the first range every nine hours, and the calculated at least one count ratio is set in step S8. When the value is exceeded, it is determined that an abnormal situation is approaching the subject. At this time, the CPU 6 notifies the user of this system by displaying the fact on the display unit 8.

  Here, in addition to the display by the display unit 8 as described above, this notification may be based on a method of sounding a warning sound from a speaker (not shown) included in the behavior prediction device 20. This is the same as the behavior prediction method according to aspect 1.

  Furthermore, if the first time is shortened according to the condition of the subject by focusing on the data in the first range, the ratio of the motion frequency to be focused on the total frequency is increased, and the detection sensitivity of the motion frequency is increased. Can be increased.

  According to the behavior prediction method according to the third embodiment of the present invention as described above, since the range of the motion histogram can be automatically learned, the method is the same as the behavior prediction method according to the second embodiment. The behavior of the subject can be predicted more quickly and accurately than in the past.

Note that the behavior prediction method according to the third embodiment can be realized by, for example, the behavior prediction device 50 illustrated in FIG.
The behavior prediction apparatus 50 illustrated in FIG. 12 includes an intensity change calculation unit 51, a storage unit 52, a focus range identification unit 53, and a notification unit 54.

Here, the intensity change calculator 51 calculates the intensity change of the received signal. The storage unit 52 classifies the magnitude of the intensity change of the signal monitored within a predetermined time into a plurality of classes, and stores the aggregated minimum value of the frequency for each class.
Then, the range-of-interest specifying unit 53 counts the frequency obtained in the monitoring within the time for each class, calculates the ratio with the minimum value stored in the storage unit 52, and sets the maximum value of the ratio. A class that exceeds the first threshold value determined accordingly is specified.

  At this time, the notification unit 54 sets a value larger than the maximum value of the ratio for each class as a second threshold for each class specified by the target range specifying unit 53, and is monitored within the time period. When the ratio calculated by summing the magnitude of the signal intensity change for each class exceeds the second threshold value in at least one of the classes specified by the target range specifying unit 53, Output a notification signal.

  The threshold value set in step S8 may be a scalar value obtained by adding a plurality of threshold values obtained in the same step. In this case, the count ratio within the first range of the normalized histogram is set. When the added scalar value exceeds the threshold value, it is possible to obtain the same effect as the behavior prediction method according to the third embodiment by determining that the subject is approaching an abnormality.

The behavior prediction method according to the present embodiment can also be realized by the behavior prediction device 50 shown in FIG.
That is, the notification unit 54 shown in FIG. 12 holds the sum of the maximum values of the above ratios for each class specified by the range-of-interest specifying unit 53 as the second threshold value or stores it in the storage unit 52. The sum of the ratios with the minimum values calculated for each class specified by the target range specifying unit 53 by adding up the magnitude of the intensity change of the signal monitored within the above time for each class is the second sum. A notification signal may be output when the threshold value is exceeded.

  In addition, a value (ratio) obtained by dividing the total value of the scalar values in the first range by the total value of the scalar values in the entire range can be referred to as an index representing the degree of sleep. Shows awakening. By monitoring this index in time series, it is possible to grasp whether the subject tends to fall asleep or to awaken.

Note that the behavior prediction method according to the present embodiment can be realized by the behavior prediction device 60 illustrated in FIG. 13, for example.
The behavior prediction apparatus 60 shown in FIG. 13 includes an intensity change calculation unit 61, a storage unit 62, a focus range identification unit 63, and a monitoring unit 64.

  Here, the intensity change calculation unit 61 calculates the intensity change of the received signal. The storage unit 62 classifies the magnitude of the intensity change of the signal monitored within a predetermined time into a plurality of classes, and stores the aggregated minimum value of the frequency for each class.

Then, the range-of-interest specifying unit 63 aggregates the frequencies obtained in the monitoring within the time for each class, calculates a ratio with the minimum value stored in the storage unit 62, and sets the maximum value of the ratio. A class that exceeds the first threshold value determined accordingly is specified.
At this time, the monitoring unit 64 monitors the ratio between the sum of the ratios for each class specified by the target range specifying unit 63 and the sum of the ratios for all the classes.

  By the way, in each of the behavior prediction methods according to the first to third embodiments, a program describing the procedure is recorded on a computer-readable recording medium such as an optical disk, and the recording medium is a behavior prediction apparatus that is a computer. It can be easily realized by attaching it to 20 drives (not shown) and causing the CPU 6 included in the behavior prediction apparatus 20 to execute the program.

  As described above, according to the behavior prediction system, the behavior prediction device, the behavior prediction method, the behavior prediction program, and the recording medium that records the program according to the embodiment of the present invention, it is transmitted from the RFID tag 2 and received by the receiver 3. Therefore, it is possible to quickly and accurately extract the intensity change to be noticed due to the behavior of the subject, so that the behavior of the subject can be predicted earlier and more accurately than before with a simple configuration.

2 RFID (Radiofrequency Identification) tag 3 Receiver 6 CPU
7 Memory 8 Display unit 20, 30, 40, 50, 60 Behavior prediction device 31, 41, 51, 61 Intensity change calculation unit 32 Measurement unit 34, 45, 54 Notification unit 42 Calculation unit 44 Measurement time limitation unit 53, 63 Range identification unit 64 Monitoring unit

Claims (19)

A behavior prediction system for predicting a subject's behavior,
Signal receiving means for receiving a signal that changes according to the behavior of the subject;
The number of times that the magnitude of the signal received by the signal receiving means or the magnitude of the intensity change of the signal exceeds a predetermined first threshold in the past and the first time in the past. A time for measuring the magnitude of the signal or the magnitude of the intensity change when the ratio with the minimum value of the number of times counted in one time is equal to or greater than a predetermined second threshold value; It is limited to a predetermined second time shorter than the first time, and the magnitude of the signal received by the signal receiving means at the second time or the magnitude of the intensity change of the signal is Action prediction provided with action prediction means for notifying that the ratio of the number of times exceeding the first threshold and the total number of measurements in the second time is equal to or greater than a predetermined third threshold system. A behavior prediction system for predicting a subject's behavior,
Signal receiving means for receiving a signal that changes according to the behavior of the subject;
The magnitude of the signal monitored within the predetermined time using the signal receiving means or the magnitude of the intensity change of the signal is classified into a plurality of classes, and the frequency generated in the monitoring is aggregated for each class. Then, in the past, the class that has a ratio with the minimum value of the frequency for each class aggregated in the time in the past exceeds the first threshold value determined according to the maximum value of the ratio is specified. For each specified class, a value larger than the maximum value of the ratio is set as the second threshold value, and the magnitude of the signal received by the signal receiving means or the magnitude of the intensity change of the signal is set. By summing up for each of the classes, the ratio with the minimum value calculated for each of the identified classes exceeds the second threshold value in at least one of the identified classes. The The behavior prediction system provided with the behavior prediction means which notifies that. A behavior prediction system for predicting a subject's behavior,
Signal receiving means for receiving a signal that changes according to the behavior of the subject;
The magnitude of the signal monitored within the predetermined time using the signal receiving means or the magnitude of the intensity change of the signal is classified into a plurality of classes, and the frequency generated in the monitoring is aggregated for each class. Then, in the past, the class that has a ratio with the minimum value of the frequency for each class aggregated in the time in the past exceeds the first threshold value determined according to the maximum value of the ratio is specified. The sum of the maximum values of the ratios for each of the specified classes is set as a second threshold value, and the magnitude of the signal received by the signal receiving means or the magnitude of the intensity change of the signal is set as the second threshold. Action provided with action predicting means for notifying that the sum of the ratio with the minimum value calculated for each of the specified classes exceeds the second threshold by counting for each class Prediction system. A behavior prediction system for predicting a subject's behavior,
Signal receiving means for receiving a signal that changes according to the behavior of the subject;
The magnitude of the signal monitored within the predetermined time using the signal receiving means or the magnitude of the intensity change of the signal is classified into a plurality of classes, and the frequency generated in the monitoring is aggregated for each class. Then, in the past, the class that has a ratio with the minimum value of the frequency for each class aggregated in the time in the past exceeds the first threshold value determined according to the maximum value of the ratio is specified. An action prediction system comprising: an action prediction unit that monitors, using the signal receiving unit, a ratio of the sum of the ratios for each specified class and the sum of the ratios for all the classes. It further comprises signal transmission means for transmitting a radio signal,
It said signal receiving means, behavior prediction system according to any of claims 1 4 for receiving the radio wave signal changes according to the behavior of the subject. A behavior prediction device for predicting a subject's behavior,
Counts the number of times the magnitude of the signal received at a predetermined first time or the magnitude of the intensity change of the signal exceeds a predetermined first threshold, and counts at the first time in the past Calculating means for calculating a ratio with the minimum value of the number of times performed,
When the ratio calculated by the calculation means is equal to or greater than a predetermined second threshold value, the time for measuring the magnitude of the signal or the magnitude of the intensity change is set to be longer than the first time. Measuring time limiting means for limiting to a short predetermined second time;
The number of times the magnitude of the signal or intensity change measured at the second time set by the measurement time limiting means exceeds the first threshold value and all measurements at the second time A behavior prediction apparatus comprising: a notification unit that notifies that a ratio with the number of times is equal to or greater than a predetermined third threshold value. The behavior prediction device is a device that predicts a subject's behavior using disturbance of a radio signal by the subject,
The received signal, behavior prediction apparatus according to claim 6, wherein the radio signal received at the fixed point. A behavior prediction device for predicting a subject's behavior,
The magnitude of the signal monitored within the predetermined time or the magnitude of the intensity change of the signal is classified into a plurality of classes, and the frequency generated in the monitoring is totaled for each class, and the past time A range of interest specifying means for specifying the class in which a ratio with the minimum value of the frequency for each class aggregated within exceeds a first threshold value determined according to the maximum value of the ratio;
For each class specified by the range-of-interest specifying means, a value larger than the maximum value of the ratio for each class is set as a second threshold value, and the magnitude of the signal monitored within the time or The ratio calculated by aggregating the magnitude of the intensity change of the signal for each class is the second threshold value in at least one of the classes specified by the range-of-interest specifying means. An action prediction apparatus comprising notification means for notifying that the number has been exceeded. A behavior prediction device for predicting a subject's behavior,
The magnitude of the signal monitored within the predetermined time or the magnitude of the intensity change of the signal is classified into a plurality of classes, and the frequency generated in the monitoring is totaled for each class, and the past time A range of interest specifying means for specifying the class in which a ratio with the minimum value of the frequency for each class aggregated within exceeds a first threshold value determined according to the maximum value of the ratio;
The sum of the maximum values of the ratios for each class specified by the range-of-interest specifying means is set as a second threshold value, and the magnitude of the signal or the change in the intensity of the signal monitored within the time is set. By summing up the size for each class, it is notified that the sum of the ratio with the minimum value calculated for each class specified by the target range specifying means has exceeded the second threshold value. An action prediction apparatus comprising a notification means. A behavior prediction device for predicting a subject's behavior,
The magnitude of the signal monitored within the predetermined time or the magnitude of the intensity change of the signal is classified into a plurality of classes, and the frequency generated in the monitoring is totaled for each class, and the past time A range of interest specifying means for specifying the class in which a ratio with the minimum value of the frequency for each class aggregated within exceeds a first threshold value determined according to the maximum value of the ratio;
A behavior prediction apparatus comprising: means for monitoring a ratio between the sum of the ratios for each class specified by the range-of-interest specifying means and the sum of the ratios for all the classes. The behavior prediction device is a device that predicts a subject's behavior using disturbance of a radio signal by the subject,
The monitored signals are behavior prediction apparatus according to claim 8 which is the radio signal that is monitored in a fixed point 10. A behavior prediction program that makes a computer predict a subject's behavior,
In the computer,
A first procedure for determining a first threshold for the signal magnitude or the intensity change in response to a pre-monitored signal magnitude or magnitude of the signal intensity change;
A second procedure for counting the number of times the magnitude of the signal received at a first predetermined time or the magnitude of the change in intensity of the signal exceeds the first threshold;
When the ratio between the number of times counted in the second procedure and the minimum value of the number of times counted in the first time in the past is equal to or greater than a predetermined second threshold value, A third procedure for limiting the time for measuring the magnitude of the signal or the magnitude of the intensity change to a predetermined second time shorter than the first time;
A fourth procedure for counting the number of times that the magnitude of the signal or the magnitude of the intensity change measured at the second time exceeds the first threshold;
And a fifth procedure for notifying that the ratio of the number of times counted in the fourth procedure and the total number of measurements in the second time is equal to or greater than a predetermined third threshold value. Behavior prediction program. A behavior prediction program that makes a computer predict a subject's behavior,
In the computer,
A first procedure for classifying the magnitude of the signal monitored within a predetermined time or the magnitude of the intensity change of the signal into a plurality of classes, and counting the frequencies generated in the monitoring for each class;
The ratio for each class between the minimum value of the frequency for each class aggregated in the past time and the frequency for each class aggregated in the first procedure depends on the maximum value of the ratio. A second procedure for identifying the class that exceeds a first threshold defined in
In the class specified in the second procedure, a third procedure for storing the maximum value of the ratio for each class;
A fourth procedure for setting a value greater than the maximum value as the second threshold value for each of the classes specified in the second procedure;
The magnitude of the signal monitored within the time or the magnitude of the change in the intensity of the signal is aggregated for each class, and the ratio of the minimum value to the minimum value is the second threshold value in the class. A behavior prediction program that executes a fifth procedure for notifying that it has been exceeded. A behavior prediction program that makes a computer predict a subject's behavior,
In the computer,
A first procedure for classifying the magnitude of the signal monitored within a predetermined time or the magnitude of the intensity change of the signal into a plurality of classes, and counting the frequencies generated in the monitoring for each class;
The ratio for each class between the minimum value of the frequency for each class aggregated in the past time and the frequency for each class aggregated in the first procedure depends on the maximum value of the ratio. A second procedure for identifying the class that exceeds a first threshold defined in
In the class specified in the second procedure, a third procedure for storing the maximum value of the ratio for each class;
A fourth procedure for setting the sum of the maximum values stored in the third procedure as a second threshold;
The magnitude of the signal monitored during the time or the magnitude of the intensity change of the signal is aggregated for each class, and the sum of the ratio with the minimum value calculated for each class is the second value. A behavior prediction program for executing a fifth procedure for determining that there is a possibility that an abnormality has occurred in the subject when the threshold value is exceeded. A behavior prediction program that makes a computer predict a subject's behavior,
In the computer,
A first procedure for classifying the magnitude of the signal monitored within a predetermined time or the magnitude of the intensity change of the signal into a plurality of classes, and counting the frequencies generated in the monitoring for each class;
The ratio for each class between the minimum value of the frequency for each class aggregated in the past time and the frequency for each class aggregated in the first procedure depends on the maximum value of the ratio. A second procedure for identifying the class that exceeds a first threshold defined in
A behavior prediction program for executing a third procedure for monitoring a ratio between the sum of the ratios in the class specified in the second procedure and the sum of the ratios in all the classes. A computer-readable recording medium recording a behavior prediction program for predicting the behavior of a subject with respect to a computer,
In the computer,
A first procedure for determining a first threshold for the signal magnitude or the intensity change in response to a pre-monitored signal magnitude or magnitude of the signal intensity change;
A second procedure for counting the number of times the magnitude of the signal received at a first predetermined time or the magnitude of the change in intensity of the signal exceeds the first threshold;
When the ratio between the number of times counted in the second procedure and the minimum value of the number of times counted in the first time in the past is equal to or greater than a predetermined second threshold value, A third procedure for limiting the time for measuring the magnitude of the signal or the magnitude of the intensity change to a predetermined second time shorter than the first time;
A fourth procedure for counting the number of times that the magnitude of the signal or the magnitude of the intensity change measured at the second time exceeds the first threshold;
And a fifth procedure for notifying that the ratio of the number of times counted in the fourth procedure and the total number of measurements in the second time is equal to or greater than a predetermined third threshold value. A computer-readable recording medium on which an action prediction program is recorded. A computer-readable recording medium recording a behavior prediction program for predicting the behavior of a subject with respect to a computer,
In the computer,
A first procedure for classifying the magnitude of the signal monitored within a predetermined time or the magnitude of the intensity change of the signal into a plurality of classes, and counting the frequencies generated in the monitoring for each class;
The ratio for each class between the minimum value of the frequency for each class aggregated in the past time and the frequency for each class aggregated in the first procedure depends on the maximum value of the ratio. A second procedure for identifying the class that exceeds a first threshold defined in
In the class specified in the second procedure, a third procedure for storing the maximum value of the ratio for each class;
A fourth procedure for setting, as the second threshold value, a value greater than the maximum value for each class specified in the second procedure;
The magnitude of the signal monitored during the time or the magnitude of the change in the intensity of the signal is aggregated for each class, and the ratio of the minimum value to the minimum value is the second threshold value in the class. A computer-readable recording medium recording an action prediction program for executing a fifth procedure for notifying that the number has been exceeded. A computer-readable recording medium recording a behavior prediction program for predicting the behavior of a subject with respect to a computer,
In the computer,
A first procedure for classifying the magnitude of the signal monitored within a predetermined time or the magnitude of the intensity change of the signal into a plurality of classes, and counting the frequencies generated in the monitoring for each class;
The ratio for each class between the minimum value of the frequency for each class aggregated in the past time and the frequency for each class aggregated in the first procedure depends on the maximum value of the ratio. A second procedure for identifying the class that exceeds a first threshold defined in
In the class specified in the second procedure, a third procedure for storing the maximum value of the ratio for each class;
A fourth procedure for setting the sum of the maximum values stored in the third procedure as a second threshold;
The magnitude of the signal monitored during the time or the magnitude of the intensity change of the signal is aggregated for each class, and the sum of the ratio with the minimum value calculated for each class is the second value. A computer-readable recording medium recording a behavior prediction program for executing a fifth procedure for determining that a subject may have changed when a threshold value is exceeded. A computer-readable recording medium recording a behavior prediction program for predicting the behavior of a subject with respect to a computer,
In the computer,
A first procedure for classifying the magnitude of the signal monitored within a predetermined time or the magnitude of the intensity change of the signal into a plurality of classes, and counting the frequencies generated in the monitoring for each class;
The ratio for each class between the minimum value of the frequency for each class aggregated in the past time and the frequency for each class aggregated in the first procedure depends on the maximum value of the ratio. A second procedure for identifying the class that exceeds a first threshold defined in
A computer-readable recording of a behavior prediction program for executing a third procedure for monitoring a ratio of the sum of the ratios in the class specified in the second procedure and the sum of the ratios in all the classes. recoding media.

Images