JP2021108048A - Fall detection terminal and program - Google Patents

Abstract

To determine fall of a user with high accuracy and high efficiency.SOLUTION: A fall detection terminal is carried by a user and detects fall of the user, and has a motion sensor, impact detection means, and fall determination means. The motion sensor detects the motion of the fall detection terminal and outputs motion data indicating the motion. The impact detection means detects an impact applied to the fall detection terminal based on the motion data. The fall determination means determines the presence or absence of fall of the user based on the motion data. Specifically, when a peak value of the motion data exceeds a predetermined threshold, the impact detection means detects the impact, and the fall determination means determines the presence or absence of the fall by inputting the motion data in a predetermined period corresponding to the time of detection of the impact to a determination model that is generated through machine learning having the motion data as input data and the presence or absence of the fall of the user as output data.SELECTED DRAWING: Figure 4

Description

The present invention relates to a fall detection terminal capable of detecting a person's fall and a program used for the fall detection terminal.

Conventionally, there is a system for detecting and reporting an accident such as a health abnormality or a fall of an elderly person or the like. The following Patent Document 1 describes a case where a user's fall is detected based on the motion data output by the motion sensor in a fall detection terminal carried by the user, and a predetermined cancellation signal is input after the fall is detected. , Cancels the fall detection, determines that an abnormality has occurred in the user if the fall detection is not canceled, and monitors the body movement detection for a predetermined first period after the fall detection is canceled. , It is described that it is determined that an abnormality has occurred in the user when no body movement is detected during the monitoring.

Here, the fall detection and the body movement detection are performed by a rule-based process of whether or not the output value of the motion sensor satisfies a predetermined condition.

Japanese Unexamined Patent Publication No. 2016-177354

However, in the rule-based processing described in Patent Document 1, it is said that the user has fallen even though he / she has not actually fallen due to the movement (falling) of the user, the mounting position of the motion sensor, the body shape of the user, and the like. In some cases, it was judged and misreported, or even though it actually fell, it was not judged that it had fallen and the report was lost.

An object of the present invention is to provide a fall detection terminal and a program capable of determining a user's fall with high accuracy and high efficiency.

In order to achieve the above object, the fall detection terminal according to one embodiment of the present invention is a fall detection terminal that is carried by a user and detects the user's fall, and includes a motion sensor, an impact detection means, and a fall determination means. Has. The motion sensor detects the motion of the fall detection terminal and outputs motion data indicating the motion. The impact detecting means detects an impact applied to the fall detection terminal based on the motion data. The fall determination means determines whether or not the user has fallen based on the motion data. Specifically, the impact detecting means detects the impact when the peak value of the motion data exceeds a predetermined threshold value, and the fall determining means uses the motion data as input data to prevent the user from falling. The presence or absence of the fall is determined by inputting the motion data for a predetermined period corresponding to the time when the impact is detected into the determination model generated by machine learning using the presence or absence as output data.

With this configuration, rule-based processing is used in the impact detection process in which the motion data does not change depending on the fall method, and a judgment model trained by machine learning is used in the fall determination process in which the motion data changes depending on the fall method. By properly using the judgment criteria and starting the fall judgment process triggered by the detection of the impact, the user's fall can be judged with high accuracy and high efficiency.

The motion sensor may be a 3-axis accelerometer. In this case, the impact detecting means may detect the impact based on the scalar value obtained from the acceleration data of each of the three axes. Further, in this case, the overturning determination means inputs the acceleration data of the three axes for a predetermined period corresponding to the detection time of the impact into the determination model generated by inputting the acceleration data of the three axes, respectively. Therefore, the presence or absence of the above-mentioned fall may be determined.

As a result, in the impact detection process in which it is not necessary to detect the acceleration as a vector and the magnitude of the acceleration can be detected, the impact can be prevented from increasing the unnecessary calculation amount by using the acceleration data of each of the three axes. For the determination of a fall that can be detected and accompanies a complicated movement, the fall can be determined with higher accuracy by using a determination model machine-learned with each acceleration data of the three axes.

The predetermined period is a period before and after the detection of the impact, and the period after the detection may be longer than the period before the detection of the impact in the period before and after the detection of the impact.

As a result, by shortening the period before impact detection, it is possible to prevent motion data in daily life from being input and causing false alarms or false alarms, and by lengthening the period after impact detection, Data is input until the acceleration after the impact settles down, and sufficient time can be secured to distinguish between a large movement of the user other than a fall and a fall.

A storage means for sequentially storing the motion data may be provided. In this case, when the impact is detected, the fall determination means may sequentially start inputting to the determination model from motion data stored before a predetermined time when the impact is detected.

As a result, when the impact due to a fall is detected, the determination process is started from the motion data slightly earlier than that, so that the fall determination process can be started without waiting until the motion data to be processed is accumulated.

A program according to another embodiment of the present invention is carried by a user and is attached to a fall detection terminal that detects the user's fall.
Based on the motion data output from the motion sensor that detects the motion of the fall detection terminal, the step of detecting the impact applied to the fall detection terminal and
The step of determining whether or not the user has fallen based on the motion data is executed.
Here, the step of detecting the impact detects the impact when the peak value of the motion data exceeds a predetermined threshold value, and the step of determining the presence or absence of the fall uses the motion data as input data. The presence or absence of the fall is determined by inputting the motion data for a predetermined period corresponding to the time when the impact is detected into the determination model generated by machine learning using the presence or absence of the user's fall as output data.

As described above, according to the present invention, it is possible to determine a user's fall with high accuracy and high efficiency. However, the effect does not limit the present invention.

It is a figure which shows the structure of the fall detection terminal which concerns on one Embodiment of this invention. It is a figure which showed the waveform example of the acceleration data used for the machine learning of the judgment model used by the fall detection terminal. It is a figure explaining the machine learning of the said determination model. It is a flowchart which showed the flow of the fall monitoring process by the said fall detection terminal. It is a figure which showed the waveform example of the scalar value of the acceleration used in the impact detection processing in the above-mentioned fall monitoring processing. It is a figure which showed the calculation method of the scalar value of the said acceleration. It is a figure which showed the calculation process of the peak value in the said impact detection process. It is a figure which showed the waveform example of the acceleration data used for the fall determination processing in the above-mentioned fall monitoring processing. It is a figure which showed the waveform example of the scalar value of the acceleration used for the body movement determination processing in the above-mentioned fall monitoring processing.

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

<Configuration of fall detection terminal>
FIG. 1 is a diagram showing a configuration of a fall detection terminal according to an embodiment of the present invention.

The fall detection terminal 100 according to the present embodiment is, for example, a wristband type (wristwatch type) wearable terminal, and is worn on the wrist or arm of a user such as an elderly person. In addition to the wristband type, for example, a pendant type that can be hung from the neck, a head mount type that can be attached to the head such as an ear, and a waist-mounted type such as a belt shape or a belt loop hanging shape can be taken.

As shown in the figure, the fall detection terminal 100 includes an acceleration sensor 11, a mounting sensor 12, a battery 13, a wireless communication unit 14, a notification unit 15, an operation display unit 16, a storage unit 17, and a control unit 18.

The acceleration sensor 11 is composed of, for example, a three-axis sensor, and detects the movement of the fall detection terminal 100 (the movement of the user wearing the fall detection terminal 100). The acceleration sensor 11 outputs acceleration data (motion data) detected in a predetermined sampling cycle. Instead of the acceleration sensor 11, another motion sensor such as an angular velocity sensor may be used.

The wearing sensor 12 is provided, for example, on the wristband portion of the fall detecting terminal 100, and detects the wearing state of the fall detecting terminal 100 on the human body (arm) of the user. As the detection method of the mounting sensor 12, for example, a capacitance type is adopted, but other types of sensors such as an inductive type, an ultrasonic type, a photoelectric type, and a magnetic type may be used. When the fall detection terminal 100 has a band shape and has a structure in which both ends thereof are connected and worn on the user's wrist, the connection between the ends is detected by a contact switch or an energized state, thereby mounting / non-wearing. It can also detect wearing.

The battery 13 supplies electric power to each part of the fall detection terminal 100. As the battery 13, for example, a rechargeable battery such as a lithium polymer battery, a lithium ion battery, or a nickel hydrogen battery is used.

The wireless communication unit 14 wirelessly communicates with the remote monitoring center C via a mobile communication network such as 3G or LTE (Long Term Evolution), and the user abnormality detected by the fall detection terminal 100 (control unit 18). To the monitoring center C. Further, the wireless communication unit 14 can also wirelessly communicate with a security terminal in the house connected to the monitoring center C by, for example, BLE (Bluetooth (registered trademark) Low Energy) or a specific low power radio.

A controller is stationed at the monitoring center C, and when an abnormality report is received from the fall detection terminal 100 (or via a security terminal in the house), necessary measures such as dispatching emergency personnel to the user's site are taken. Taken.

The notification unit 15 is configured as, for example, a vibration device, and notifies the user of an abnormality or an operation reception by a stimulus caused by vibration. In addition to vibration, notification may be made by voice.

The operation display unit 16 is configured as, for example, a touch panel display, and executes various display processes such as an abnormality notification together with or in place of the notification by the notification unit 15.

The storage unit 17 stores the acceleration data detected by the acceleration sensor 11 and its scalar data. The storage unit 17 may be configured as, for example, a ring buffer that temporarily stores a predetermined amount of the acceleration data or the like.

The control unit 18 comprehensively controls each unit of the fall detection terminal 100, and executes various calculations in response to inputs from the above sensors and operation units. The control unit 18 functions as a fall monitoring means 181, an impact detecting means 182, a fall determining means 183, and a body movement determining means 184. Each means may be configured as a dedicated hardware circuit or as software (program).

The fall monitoring means 181 monitors whether or not a fall abnormality has occurred in the user based on the determination results of the fall determination means 183 and the body movement determination means 184, and if a fall abnormality occurs, notifies the monitoring center C of the abnormality. do.

The impact detecting means 182 detects the impact applied to the user when the user falls by detecting the peak value of the acceleration (for example, 40 m / s2 or more) from the scalar value of the acceleration data detected by the acceleration sensor 11.

The fall determination means 183 uses each of the acceleration data before and after the peak value among the acceleration data of the three axes detected by the acceleration sensor 11 into a determination model (learned model) generated by machine learning by the learning means 200. By inputting, a fall event (whether or not the user has fallen) is determined.

The body motion determination means 184 is based on whether or not the scalar value of the acceleration data after the acceleration data input to the determination model among the acceleration data detected by the acceleration sensor 11 satisfies a predetermined condition. Determines the presence or absence of body movement of the user after a fall.

The learning means 200 is composed of, for example, a general-purpose computer such as a PC (Personal Computer) or a computer dedicated to learning processing, and the acceleration data of the three axes of the acceleration sensor 11 at the time of a fall is obtained from a person or a doll wearing the fall detection terminal 100. Is used as input data, for example, a determination model (learned model) used for determining a fall by the fall determination means 183 is generated in advance by machine learning such as deep learning such as a convolutional neural network (CNN). The generated determination model is installed in the fall determination means 183.

<Operation of fall detection terminal>]
Next, the monitoring operation of the fall detection terminal 100 configured as described above will be described.

[Learning process]
First, the learning process by the learning means 200 will be described. FIG. 2 is a diagram showing a waveform example of acceleration data used for machine learning of the determination model used by the fall detection terminal 100. Further, FIG. 3 is a diagram illustrating machine learning of the determination model.

As described above, the learning means 200 acquires acceleration data at the time of a fall from a person or a doll wearing the fall detection terminal 100, and learns in advance by machine learning using the acceleration data as input data.

As shown in FIG. 3, when a neural network is used as the machine learning method, as shown in FIG. 2, the learning means 200 has three-axis acceleration data (x-axis:) before and after the acceleration peak when the person or the doll falls. Fig. A, y-axis: Fig. B, z-axis: Fig. C) is cut out, and the cut out acceleration value is input to the input layer as input data. Dimension compression is performed.

Although not shown, for example, using a convolutional neural network having a Conv layer and a Pooling layer in the intermediate layer (hidden layer), feature point extraction by convolution calculation using a filter and dimension compression of the feature point data are performed to perform features. A method of extracting points is generally used.

As shown in the figure, the learning means 200 collectively inputs the acceleration values for a predetermined time when inputting the acceleration values of each axis to the input layer, and the next input is the time series of the previously input acceleration values. Enter the acceleration values for a predetermined time that are shifted by one in order.

For example, when the acceleration sensor 11 detects the acceleration value 50 times per second, the learning means 200 first inputs the acceleration values for three times (from 0.02 seconds to 0.06 seconds) as one input data set, and then the learning means 200. , The acceleration value to be input along the time series is shifted by one (from 0.04 to 0.08 seconds) and input as the next input data set.

Subsequently, the learning means 200 outputs from the output layer whether it is determined to be a fall or a non-fall (when the fall is 1 and the non-fall is 0, 1 or 0 is output from the output layer).

Then, the learning means 200 performs backpropagation based on the match / unmatch between the data input to the input layer and the data of the output layer, and updates the weighting coefficient between the neurons of the convolutional neural network and the like. The coefficient is updated, and when the error between the input and the output becomes smaller than a certain level, learning is terminated and the coefficient is installed in the fall detection terminal 100.

Further, the learning means 200 also inputs and learns the acceleration value of each axis when the user does not fall (in daily life or when a movement with an impact similar to a fall is performed).

The learning means 200 generates a determination model by repeating this for a large number of data sets (for example, 1000-1500 fall data and 1000-1500 non-fall data), and the generated determination model is used by the fall detection terminal 100. It is installed in the fall determination means 183. It should be noted that learning may be performed using only the fall data to generate a determination model.

[Fall monitoring process]
Next, a fall monitoring process by the control unit 18 (fall monitoring means 181) of the fall detection terminal 100 will be described. The fall monitoring means 181 determines whether or not the user has fallen based on the output data of the acceleration sensor 11. In the fall monitoring process, the range in which the acceleration data changes significantly depending on the user's fall method is determined using the determination model machine-learned by the learning means 200, and the acceleration data changes significantly depending on the user's fall method. The range not to be determined is determined by a rule base for confirming whether the acceleration data satisfies a predetermined condition.

In the following description, the control unit 18 of the fall detection terminal 100 or each means will be described as the main operating body, but this operation is also performed in cooperation with the program executed under the control of the control unit 18.

FIG. 4 is a flowchart showing the flow of the fall monitoring process by the fall monitoring means 181.

(Fall judgment processing)
First, a fall determination process for determining whether or not a user has fallen will be described. Since a fall involves a collision of the user's body with the ground or the like, the acceleration greatly changes due to the impact applied to the fall detection terminal 100. Therefore, the control unit 18 first determines whether or not an acceleration peak equal to or greater than the threshold value that triggers the fall determination process is detected (step 41 in FIG. 4).

Specifically, the control unit 18 (impact detecting means 182) obtains a scalar value from the acceleration data of the three axes measured by the acceleration sensor 11, and the obtained scalar value is equal to _{or higher than the threshold Th p} (for example, 40 m / s2). Accelerometer peak is detected.

FIG. 5 is a diagram showing a waveform example of a scalar value of acceleration data of three axes, and FIG. 6 is a diagram showing a method of calculating the scalar value.

As shown in FIG. 6, the scalar value of the vector of the acceleration component A (Ax, Ay, Az) at a certain time can be calculated by the square root of the sum of squares of each acceleration component of the x-axis, y-axis, and z-axis.

Here, the scalar value of the acceleration is a physical quantity having only the magnitude of the acceleration, and is a value capable of determining the absolute magnitude of the acceleration applied to the fall detection terminal 100.

When trying to determine the magnitude of acceleration with each component of the three axes of the acceleration sensor 11, the magnitude of the acceleration applied to the fall detection terminal 100 is dispersed in each of the three axes, and it is possible to measure the magnitude accurately. difficult. Further, since it is unknown in which direction and in what direction the user falls down, it is unknown in which of the three axes the peak value is detected. On the other hand, by scaling, the magnitudes of accelerations dispersed in three axes can be summarized, and by determining the absolute magnitude applied to the fall detection terminal 100, the peak value as shown in FIG. 5 can be determined. P can be detected.

In addition, when the acceleration of the three axes is used as it is, it is necessary to confirm the change of the acceleration of each of the three axes, but in the case of a scalar, it is sufficient to confirm the change of only the scalar value, so that the processing can be made more efficient. can.

Further, when the user falls, the peak value above the threshold value is always detected even if the method of falling is different (acceleration data does not change depending on the method of falling). Therefore, the impact detecting means 182 uses the peak value on a rule basis. Detection processing is performed.

In this way, by starting the fall determination with the peak detection as a trigger, the amount of calculation can be reduced as compared with the case where the possibility of the fall is always determined by the determination model based on the machine learning, and the battery 13 By extending the life of the user, it is possible to watch over the user for a long time.

FIG. 7 is a diagram showing a partially simplified waveform of FIG. 5 in order to explain a method of detecting a peak of acceleration when a user falls.

As shown in the figure, the impact detecting means 182 obtains a point Pa that falls below the _{threshold value Th p.} Similarly, the impact detecting means 182 obtains a point Pb rising from the _{threshold value Th p.} The point P (at the peak) where the acceleration value is maximized between the points Pa and the point Pb is detected as the acceleration peak, and the acceleration value at the peak is obtained as the peak value P.

When the acceleration peak is detected (Yes in step 41), the control unit 18 moves to the determination process of presence / absence of a fall using the determination model. FIG. 8 is a diagram showing a waveform example of acceleration data (x-axis) used in the fall determination process.

The fall determination means 183 determines the possibility of a fall by inputting acceleration data before and after the detected peak value P into the determination model (step 42).

Specifically, the fall determination means 183 inputs waveform data for a predetermined period as input data to the determination model among the waveform data of each of the three axes before and after the peak value P (first time point). Here, when setting the predetermined period (hereinafter, referred to as a fall determination period (first period)), the fall determination means 183 sets the period after the peak detection to be longer than the period before the peak. For example, as shown in FIG. 8, the waveform data from the time when the peak P is detected to 1 second before and the waveform data to be measured from now on for 3 seconds after the peak detection are input to the determination model as input data.

Here, if the data before the fall (before the peak value P) is input too much as input data (for example, 3 seconds before the fall), the data of the user's daily life before the fall is also taken into the judgment model, resulting in a false alarm / It causes a false alarm. Therefore, before the peak value P, it is preferable to be about 1 second before the free fall characteristic can be confirmed.

Also, by checking the waveforms of the three axes from after the peak value P until the acceleration settles down, the relationship between each axis and the acceleration value of each axis can be determined, for example, when the bat is shaken or when it hits an object. It is possible to capture the acceleration value of the movement when it falls, which is different from the movement accompanied by other impacts such as when it falls.

On the other hand, if the fall is judged by the scalar value as in the case of the impact detection, the direction of the acceleration cannot be determined. This will cause false alarms.

Further, since the waveform input after the fall determines the presence or absence of the fall from the waveform from the peak value P until the acceleration settles down, the period of the acceleration data after the peak input to the determination model is neither too long nor too short. No time (about 3 seconds) is preferable.

In addition, when the user falls, the detected acceleration data of the three axes also changes according to the fall method. Therefore, in the rule-based judgment using the threshold value, all falls are judged while suppressing false alarms / false alarms. It is difficult to set a possible threshold. Therefore, the fall determination means 183 determines the presence or absence of a fall using a machine-learned determination model.

Here, the acceleration data input to the determination model is detected by the acceleration sensor 11 and stored in the storage unit 17. As shown in the upper part of FIG. 3, the storage unit 17 stores acceleration data for each axis in chronological order.

For example, the fall determination means 183 collectively inputs a predetermined number of temporally continuous acceleration data for the fall determination period as input data for the acceleration values before and after the peak. For example, when the acceleration sensor 11 detects the acceleration value 50 times per second, first, as in the input method by the learning means 200, three times (from 0.02 seconds to 0.06 seconds) are used as one input data set. Input, and then input the acceleration value to be input in chronological order shifted by one (from 0.04 to 0.08 seconds) as the next input data set.

Further, the fall determination means 183 sequentially starts inputting from the acceleration value from the time of peak detection to 1 second before peak detection, which has already been measured and stored in the storage unit 17, and from the time of peak detection to 3 seconds after peak detection. The acceleration value is input as soon as the data sets that are measured, stored in the storage unit 17 and input are prepared. As a result, the fall determination process can be started without waiting until the acceleration data of the input target is accumulated in the storage unit 17.

The fall determination means 183 determines the user's fall by the determination model based on the input acceleration data, and outputs "possible fall" or "no possibility of fall" as output data (step 43). ..

When it is determined that there is no fall (No in step 43), the control unit 18 ends the fall determination process.

When it is determined that there is a fall (Yes in step 43), the control unit 18 moves to the body movement determination process.

(Body movement judgment processing)
Next, the body movement detection process for determining the fall abnormality after the fall is detected as described above will be described.

When it is determined by the fall determination means 183 that the user has a fall (possibility), the body movement determination means 184 determines the body movement determination period (for example, 10 seconds, hereinafter, the body movement determination period) based on the scalar value of the acceleration data. It is determined whether or not the user's body movement is detected during the second period). FIG. 9 is a diagram showing a waveform example of a scalar value of acceleration used in the body movement determination process.

Specifically, as shown in the figure, the body movement determination means 184 accelerates the body movement determination period starting after a predetermined time (for example, 2 seconds) has elapsed from the acceleration data input to the determination model in the fall determination process. The presence or absence of body movement is determined from the scalar value of the data (step 44). It should be noted that the presence or absence of body movement may be determined from the scalar value of the acceleration data immediately after the acceleration data input to the determination model without waiting for the elapse of a predetermined time. Further, the body movement determination period may be a period after the fall determination period, and the fall determination period and the body movement determination period may be continuous, and a part of the fall determination period and a part of the body movement determination period may be continuous. It may be duplicated.

Here, the presence or absence of body movement is determined from the scalar value instead of the three-axis data if the body movement determination process can detect a slight movement of the user after a fall rather than a large acceleration data at the time of impact. This is because it is not necessary to determine a complicated movement that differs depending on the direction as in the case of a fall, but simply to detect whether or not the user has moved.

Further, when determining that the user has lost consciousness and has fallen, the body movement determination means 184 is not a machine-learned determination model because it is determined that the user does not move (acceleration data does not change) after the fall. , Judge the presence or absence of body movement on a rule basis.

Specifically, the body movement determining means 184 determines that the user has body movement, for example, when the following conditions 1) to 4) are satisfied. The body movement determining means 184 may use any one of the following conditions, or may determine that there is body movement when any one of the following conditions is satisfied.
1) When a certain number of peak values (for example, 12 m / s ² ) are detected within the body movement judgment period 2) When the total of the above peak values exceeds the predetermined value within the body movement judgment period 3) Body When the total value obtained by accumulating the amount of change in acceleration during the motion judgment period exceeds a certain value 4) When the amount of change in acceleration per unit time exceeds a certain value within the body motion judgment period

If it is determined that the user is in motion by satisfying the above conditions after the elapse of a predetermined time after the fall determination (Yes in step 45), the user is not considered to be in a dangerous state, so that the control unit 18 (fall monitoring unit 181) ends without determining that it is a fall event.

On the other hand, if the user's body movement is not detected under the above conditions after the elapse of a predetermined time after the fall determination, the fall monitoring unit 181 determines the user's fall abnormality and notifies the monitoring center C. (Step 46).

[summary]
As described above, according to the present embodiment, in the fall determination process in which a large change in acceleration data is likely to occur depending on the method of fall, a determination model trained by machine learning is used, and a body in which a large change is unlikely to occur in acceleration data. Rule-based processing is used in the motion judgment processing, and by properly using the judgment criteria, false alarms and false alarms are suppressed, and the user's fall (the user loses consciousness and falls) is highly accurate and highly efficient. Can be determined. In addition, by reducing the load of calculation processing as compared with the case where the judgment model is used for all, it is possible to realize long-term monitoring.

[Modification example]
The present technology is not limited to the above-described embodiment, and can be variously modified without departing from the gist of the present technology.

In the above-described embodiment, the fall determination means 183 collectively inputs a predetermined number of time-continuous data among the acceleration data for the fall determination period before and after the peak as input data. However, the fall determination means may collectively input the acceleration data for the fall determination period into the determination model.

In the above-described embodiment, the fall determination means 183 uses the acceleration value detected by the acceleration sensor 11 as input data to the determination model. However, the input data may be other than the acceleration value. For example, the image data obtained by graphing the waveform data of each axis representing the acceleration data shown in FIG. 2 or the like in time series may be used as the input data. In this case, learning by the learning means 200 is also performed with the same image data.

In the above-described embodiment, the learning means 200 is provided outside the fall detection terminal 100, but the learning means 200 may be provided inside the fall detection terminal 100. In this case, the learning means 200 may execute the learning process with the same input data while executing the determination process by the fall determination means 183, and update the determination model each time.

In the above-described embodiment, the learning means 200 has learned the acceleration data before and after the impact detection including the impact detection time point as the input data. However, the period of the data to be input may be other than this. For example, the acceleration data before and after the impact detection other than the time of the impact detection may be learned as input data, or the acceleration data only before the impact detection or only after the impact detection may be learned as the input data. In this case, the input data input to the determination model in the fall monitoring process corresponds to the period of the learned input data.

In the above-described embodiment, the fall detection terminal 100 is configured to include various sensors, a notification unit, an operation display unit, a wireless communication unit, and the like as wearable terminals, but at least a part of each of these parts of the fall detection terminal 100. Is provided in another device (for example, a mobile terminal such as a smartphone carried by a user), and the same processing as in the above embodiment may be executed by the cooperation processing with the other device.

11 ... Accelerometer 12 ... Mounting sensor 12
14 ... Wireless communication unit 14
18 ... Control unit 181 ... Fall monitoring means 182 ... Impact detection means 183 ... Fall judgment means 184 ... Body movement judgment means 100 ... Fall detection terminal 200 ... Learning means C ... Monitoring center

Claims (5)

A fall detection terminal that is carried by a user and detects the user's fall.
A motion sensor that detects the motion of the fall detection terminal and outputs motion data indicating the motion,
An impact detecting means for detecting an impact applied to the fall detection terminal based on the motion data, and an impact detecting means.
It is provided with a fall determination means for determining whether or not the user has fallen based on the movement data.
The impact detecting means detects the impact when the peak value of the motion data exceeds a predetermined threshold value, and detects the impact.
The fall determination means inputs the motion data for a predetermined period corresponding to the impact detection time point into a determination model generated by machine learning using the motion data as input data and the presence or absence of the user's fall as output data. A fall detection terminal that determines the presence or absence of the fall. The fall detection terminal according to claim 1.
The motion sensor is a 3-axis accelerometer.
The impact detecting means detects the impact based on a scalar value obtained from each acceleration data of the three axes.
The tipping determination means inputs the acceleration data of the three axes for a predetermined period corresponding to the detection time of the impact into the determination model generated by inputting the acceleration data of the three axes, respectively. A fall detection terminal that determines the presence or absence of a fall. The fall detection terminal according to claim 1 or 2.
The predetermined period is a period before and after the detection of the impact.
A fall detection terminal in which the period after the impact is detected is longer than the period before the impact is detected in the period before and after the impact is detected. The fall detection terminal according to any one of claims 1 to 3.
It has a storage means for sequentially storing the motion data, and has a storage means.
The fall detection means is a fall detection terminal that, when the impact is detected, sequentially starts inputting to the determination model from motion data stored before a predetermined time from the time when the impact is detected. A fall detection terminal that is carried by a user and detects the user's fall.
Based on the motion data output from the motion sensor that detects the motion of the fall detection terminal, the step of detecting the impact applied to the fall detection terminal and
A step of determining whether or not the user has fallen based on the motion data,
Is a program that executes
The step of detecting the impact detects the impact when the peak value of the motion data exceeds a predetermined threshold value.
The step of determining the presence or absence of a fall is described in a determination model generated by machine learning using the motion data as input data and the presence or absence of a fall of the user as output data for a predetermined period corresponding to the time when the impact is detected. A program that determines the presence or absence of the fall by inputting motion data.

Images