JP7267104B2 - system - Google Patents

Description

The present embodiment relates to a system and the like.

Systems that provide various types of support to users using bed apparatuses have been proposed.

For example, a camera is installed to photograph the fence attached to the electric bed from the outside, and the camera captures images of the bed while it is in motion and a reference image that records the positions of the edges of the vertical pipes of the fence and the sides of the fence in advance. Using this method, the peripheral image of the edge and the side image of the fence are respectively extracted, and based on the data in which the edge is not detected in the position where the edge should be captured and the data in which the edge of the obstacle is detected in the side image, image processing is performed as an error, and an error image is obtained. An invention has been disclosed in which the operation of a bed is stopped because there is a risk that an obstacle will be caught if the value of the error concentration portion in .

The monitoring bed includes a load-measuring pin sensor, a signal processing unit that processes the output of the load-measuring pin sensor to obtain a biological signal, and display means for displaying the processing result of the signal processing unit. a monitoring system, wherein the signal processing unit includes a biological amplifier that amplifies a biological signal, and biological offset follow-up processing that dynamically controls the offset and gain of the output of the biological amplifier according to the biological signal amplitude to be measured. An invention is disclosed that provides a monitoring system using a monitoring bed equipped with a load measuring pin sensor that can be easily used in an existing bed by including a part (see, for example, Patent Document 2).

JP 2008-167931 A JP 2018-134372 A

This embodiment is to provide a system or the like that can prevent a user from falling or falling.

The system of this embodiment includes at least one or more first information among the posture of the user, information on the motion of the user, biometric information of the user, and disease information of the user, and information on the environment around the user. a prediction unit that predicts the risk of falling of the user based on the information obtained by the acquisition unit; and a notification unit that provides notification according to the prediction result of the prediction unit.

According to this embodiment, it is possible to provide a system or the like that can prevent a user from overturning or falling.

It is a figure showing the whole system in a 1st embodiment. It is a figure for demonstrating the installation condition of the camera in 1st Embodiment. 3 is a diagram for explaining a functional configuration in the first embodiment; FIG. It is a figure for demonstrating the surrounding environment acquisition process in 1st Embodiment. FIG. 4 is a diagram for explaining posture estimation processing in the first embodiment; FIG. 7 is a diagram for explaining motion acquisition processing in the first embodiment; FIG. 7 is a diagram for explaining overturn prediction processing in the first embodiment; It is a figure for demonstrating the correspondence execution process in 1st Embodiment. It is a figure for demonstrating the 1st operation example in 1st Embodiment. It is a figure for demonstrating the 2nd operation example in 1st Embodiment. FIG. 11 is a diagram for explaining a third operation example in the first embodiment; It is a figure for demonstrating the 4th operation example in 1st Embodiment. It is a figure for demonstrating the 5th operation example in 1st Embodiment. It is a figure for demonstrating the 6th operation example in 1st Embodiment. It is a figure for demonstrating the 7th operation example in 1st Embodiment. It is a figure which shows the whole system in 2nd Embodiment. It is a figure for demonstrating the functional structure in 2nd Embodiment.

One form for implementing the system of the present disclosure will be described below with reference to the drawings. It should be noted that the content of the present disclosure merely shows an example of the form for implementation, and is not limited to the numerical values and configurations disclosed, and is within the equivalent range that can be conceived by those skilled in the art. is also included.

[1. First Embodiment]
Various techniques have been proposed to prevent users from falling or falling. For example, the system of the comparative example includes a sensor located at the edge of the bed apparatus, and the sensor detects and notifies that the user has taken the edge sitting position. As a result, the system of the comparative example can recognize that the user may leave the bed. In addition, the system of the comparative example includes a sensor installed on the floor that comes into contact when the user puts his/her foot down from the bed apparatus, and specifically determines whether the user has used the bed apparatus (for example, whether the user has whether or not the patient has been in bed) to medical staff, staff, caregivers, etc.

Here, if the timing of notifying medical personnel, staff, caregivers, etc. is late, there is a risk that the user will fall from the bed apparatus. In addition, after the user leaves the bed system, there is a risk that the user will fall due to the environment around the bed system (for example, the position of the wheelchair, the position of the shoes, the arrangement of the side table, etc.). Sometimes.

However, if the system of the comparative example issues a notification at a timing earlier than the timing at which the user leaves the bed apparatus, excessive (unnecessary) notification is given to medical staff, staff, caregivers, etc. There is a problem that it leads to work burden. In addition, the degree of risk of overturning/falling differs depending on the physical condition of the user and the surrounding environment. Therefore, if the system of the comparative example sets the notification conditions and then notifies the staff according to the uniform criteria according to the set conditions, the staff will be able to respond appropriately according to the user's characteristics and the surrounding environment. can become incapable of taking

Therefore, according to the first system of the present embodiment, it is possible to predict the risk of the user falling or falling, and to provide a system capable of notifying the user of the risk at an appropriate timing.

It should be noted that the user in this specification refers to a person who uses the bed equipment (mattress), and is not limited to those who receive treatment for illness (for example, patients), but also to those who receive nursing care at facilities and those who are in bed equipment. A healthy person can be a user as long as he or she does (for example, a person lying on his or her back).

In this specification, the term "staff" refers to medical personnel, facility staff, and people who assist users, including those at home.

Further, in this specification, an obstacle means something that causes a user to fall or fall. For example, objects such as a table, poles, and wheelchairs placed near the bed device 5, user's footwear (shoes and sandals), curtains, and other objects included in the living room are expressed as obstacles.

Further, if the bed device 5 itself causes the user to fall or fall, it can become an obstacle. For example, if the direction of the casters of the bed device 5 is in an unlocked state, the bed device 5 can become an obstacle.

In this specification, image data is data based on images captured by the camera device 20 and includes still images, moving images, and all other images captured by the camera device 20 .

[1.1 Overall description]
[1.1.1 Description of the entire system]
FIG. 1(a) is an overall view for explaining the outline of the system 1 in this embodiment. The system 1 includes a control device 10 , a server device 50 , a terminal device 52 and a mobile terminal device 54 . The control device 10, the server device 50, the terminal device 52, and the mobile terminal device 54 are communicably connected via the network NW. The living room R is a room in a hospital, a room in a facility, a home, or the like, in which the bed device 5 used by the user P is arranged.

A bed device 5 used by a user P is installed in the living room R. The bed apparatus 5 is provided with one or a plurality of movable back bottoms, waist bottoms, knee bottoms, leg bottoms, etc. (hereinafter collectively referred to as "bottom parts"). It can be performed. Further, the bed device 5 has an elevating mechanism, and the height of the bottom portion (floor height) of the bed device 5 can be changed. In this specification, the floor height of the bed device means the distance from the floor on which the bed device 5 serving as a reference is placed to the bottom portion. The floor height of the bed apparatus may be the distance from the floor to the upper frame, or the distance from the floor to the top of the mattress, instead of the distance from the floor to the bottom. Further, the floor height of the bed apparatus may be determined by using the lower frame instead of the floor as a reference position.

Furthermore, the bed apparatus 5 can perform an operation (tilt operation) of tilting the entire bed apparatus 5 . The bed apparatus 5 may be a bed apparatus in which none of the parts are movable.

Further, the bed apparatus 5 can be provided with a headboard on the head side and a footboard on the foot side. Further, the bed device 5 has a camera device 20, which photographs the user P, the surrounding environment, the state of the living room R, the staff, and the like. A method for installing the camera device 20 will be described later.

Here, the "surrounding environment" refers to the range around where the bed apparatus 5 is installed (for example, the range included within a predetermined distance from the position where the bed apparatus 5 is installed, the range including the living room R, It refers to the environment within which the camera device 20 can shoot). For example, the surrounding environment includes things arranged around the bed device 5 (for example, bed headboard, IV pole, etc.), movable things such as shoes and wheelchairs, brightness of the room, multiple beds, etc. In the case of a room, the location of other users, staff, etc. is included.

Also, in this specification, the term “periphery” refers to a range that can be photographed by the camera device 20 . For example, if the camera device 20 can photograph the doorway of the living room in which the bed device 5 is installed and the entire range of the living room, the doorway of the living room and the entire range of the living room are also included in the "periphery".

A sensor device 30 is placed on the bed device 5 . The sensor device 30 is a sensor that detects body movements of the user on the bed device 5 . By using the sensor device 30, the control device 10 can acquire biological information values such as the user's heartbeat and respiratory rate, and can detect the user's position, center of gravity, posture, and the like.

The sensor device 30 may be placed, for example, on a mattress on the bed device 5 or between the bottom part of the bed device 5 and the mattress. Moreover, the sensor device 30 may be placed on the bed device 5 at a position where the body movement of the user can be detected. The position where the sensor device 30 is placed is preferably a position corresponding to the user's back when the user lies down.

For example, as shown in FIG. 1B, the sensor device 30 is placed on or under the mattress 7 placed on the bottom portion of the bed device 5 . At this time, the sensor device 30 is placed at a position separated by a distance M2 from the head-side end so as to be positioned on the user's P back (in the vicinity of the chest). For example, M2 is 40 cm.

Further, the sensor device 30 may be realized by providing a load sensor under the bed device 5 (between the floor and the bed device 5), or by providing a strain gauge on the motor supporting the frame or the bed so that the user can Body motion may be detected.

In living room R, fixtures are placed as various items. For example, in FIG. 1(a), fixtures such as a stand 72, a side table 73, and a wheelchair 74 are placed. In addition to the fixtures, user P's shoes 75 are placed beside the bed device 5 . In this manner, various fixtures and objects are placed in the living room R. As for these objects, the system may grasp the positions of fixtures and objects with the camera device 20, as will be described later. Also, by building an IoT unit (including a communication module) into each fixture such as the stand 72, the side table 73, and the wheelchair 74, the system may grasp the position of the fixture.

Furthermore, a window 76 is provided in the hospital room. A curtain is attached to the window 76 . The system may grasp whether the curtain is open or closed by using the camera device 20 or the IoT unit. In addition, the system may use the camera device 20 or the IoT unit to grasp whether the curtains installed in the multi-floor room are open or closed.

The control device 10 is a device that controls the entire system. The control device 10 of this embodiment is installed in a tablet device, and each device may be controlled from another device. For example, the control device 10 may display biological information values such as heart rate and respiration value acquired by the sensor device 30 on a monitor.

Also, the control device 10 may be further connected to other devices. For example, controller 10 may be connected to a bed controller that controls bed apparatus 5 . The control device 10 can acquire the state of the bed apparatus 5 and control the bed apparatus 5 (for example, control of raising the back/lowering the back).

Note that the control device 10 may be configured integrally with the camera device 20 or may be provided in a bed control device or the like different from the camera device 20 .

In addition, the control device 10 may be realized by a device located in another place instead of the living room R, or may be provided by installing an application in a terminal device (for example, a smartphone) of a user, staff, or the like. may Also, the functions implemented by the control device 10 may be implemented by the server device.

The control device 10 is configured to be connectable to the network NW. The control device 10 connects to the access point 12 via a wireless LAN, for example, and connects to the network NW. The access point 12 is a wireless LAN base station device, and is capable of wireless communication such as IEEE802.11a/b/g/n. Note that the control device 10 may communicate using a wired LAN, short-range wireless communication such as Bluetooth (registered trademark), or other communication lines such as LTE/5G instead of the wireless LAN.

The server device 50 is a server that manages necessary information in the system. The server device 50 manages various types of information such as information on user's disease, information on medication, hospitalization history, and information on fall history. The terminal device 52 and the mobile terminal device 54 are devices used by the staff. For example, the control device 10 may notify the terminal device 52 or the mobile terminal device 54 when there is a risk of falling.

Note that the server device 50 may receive and manage the data acquired by the control device 10 . For example, the server device 50 may receive and manage image data captured by the camera device 20 . The server device 50 is not limited to a server device that manages this system, but includes a server device that manages a medical system/hospital system, an electronic chart server device, and a management server device customized by a hospital or facility.

[1.1.2 Camera installation pattern]
FIG. 2 is a diagram for explaining a camera installation pattern. The five patterns are described below. Note that the camera device 20 may be built in or externally attached to the frame of the bed device 5 or the bottom of the bed device 5 in advance. Further, the camera device 20 may be built in or externally attached to the bed device 5 later.

(first pattern)
As shown in FIG. 2( a ), the camera device 20 is installed around the bed device 5 . Specifically, the camera device 20 can be mounted on the head side, the foot side, and the longitudinal side (user's right side, user's left side) of the bed device 5, a board, a fence/aid bar attached to the bed, and others. It is installed in a dedicated mounting device using a fence hole or a board. Thereby, the camera device 20 can photograph the outer periphery of the bed device 5, the user, the staff, and the like. For example, camera device 20 may preferably be capable of photographing in the range of 120-230 degrees.

In FIG. 2A, the camera device 20 is installed at the bottom and the center of the board. It may be installed at each of the four locations, or two or more, or a plurality of units, may be installed on one side. Camera devices 20 having different angles of view may be installed in combination, or may be installed anywhere. A lens of the camera device 20 is installed so as to face the outside of the bed device 5 .

(second pattern)
As shown in FIG. 2B, the camera device 20 is installed on a board. Specifically, the camera device 20 is installed in the center of the footboard (or the headboard) on the foot side of the bed device 5 so that the entire user of the bed device 5 can be photographed. Camera device 20 may preferably be capable of photographing in the range of 60 degrees to 230 degrees. In addition, two or more camera devices 20 may be installed, or camera devices 20 having different angles of view may be installed in combination. In addition, the camera device 20 can be installed anywhere, for example, on the right side or left side of the footboard (headboard), as long as the top of the bed device including the user can be photographed. good too. The lens of the camera device 20 is installed so as to face the inside of the bed device 5. - 特許庁

(Third pattern)
As shown in FIG. 2(c), the camera devices 20 are installed on the boards on both sides. Specifically, the camera device 20 is installed in the center of the footboard and headboard of the bed device 5 so that the entire user of the bed device 5 can be photographed. Other explanations are the same as those of the second pattern, and detailed explanations are omitted.

(4th pattern)
The camera device 20 may be installed in a state in which the first to third patterns described above are combined. FIG. 2(d) shows a state in which the camera device 20 is placed at the edge of the headboard and/or footboard. For example, the camera device 20 is installed near the end of the footboard or near the grip provided above the footboard. In this way, the camera device 20 is arranged at a place where the user or the surroundings of the bed device 5 can be photographed.

In this way, the camera device 20 is provided so as to correspond to one bed device 5 rather than a camera device already in the room. As a result, the camera device 20 does not need to be calibrated for the bed device 5, the surrounding environment, and the like, thereby improving convenience for the user. In addition, even if the user of the bed device 5 changes or the living room in which the bed device 5 is placed changes, the bed device 5 and the camera device 20 are compatible, so the data saved in the system can be changed. is not required, which improves user convenience.

(Fifth pattern)
In the above-described pattern, the camera device 20 is installed in the bed device 5 as an example, but the camera device 20 already installed in the living room R can also be used, for example. Specifically, if a monitoring camera or the like has already been installed, the image thereof may be used. In this case, the user P on the bed device uses the camera device 20 at a position within the angle of view.

[1.2 Description of functional configuration and processing]
Next, in the system 1 according to the present embodiment, a functional configuration centering on the control device 10 will be described with reference to FIG.

[1.2.1 Control unit/storage unit]
The control device 10 includes a control section 100 , a storage section 120 , a display section 130 , an operation section 135 , a notification section 140 and a communication section 150 . Also, the communication unit 150 can be connected to the camera device 20 , the sensor device 30 , the bed control device 40 , the server device 50 and the peripheral device 60 .

The control unit 100 controls the entire control device 10 . The control unit 100 is one or a plurality of arithmetic units (for example, a CPU (Central Processing Unit)) that implements various functions by reading and executing various programs stored in the storage unit 120 .

The control unit 100 reads out and executes a program stored in the storage unit 120 to obtain a peripheral environment acquisition unit 102, a posture acquisition unit 104, a motion acquisition unit 106, a biological information acquisition unit 108, and disease information. It functions as an acquisition unit 110 , a prediction unit 112 , and a correspondence execution unit 114 .

The storage unit 120 stores various programs and data necessary for the operation of the control device 10 . The storage unit 120 is configured by, for example, a semiconductor memory such as an SSD (Solid State Drive) or an HDD (Hard Disk Drive).

The storage unit 120 also includes a surrounding environment storage area 1202, a user information storage area 1204 for storing user information, and an image data storage area 1240, and includes an explanatory variable table 1220 and a prediction dictionary DB 1230. I remember.

(1) Surrounding Environment Acquisition Unit (First Acquisition Unit)
Based on the image data captured by the camera device 20, the surrounding environment acquisition unit 102 acquires the surrounding environment of the bed device 5, such as obstacles and surrounding brightness, from equipment included in the image data. The surrounding environment acquisition unit 102 outputs and stores the feature amount based on the acquired surrounding environment to the surrounding environment storage area 1202 .

The surrounding environment acquisition process executed by the surrounding environment acquisition unit 102 will be described with reference to FIG. 4(a). First, the surrounding environment acquisition unit 102 acquires image data from the camera device 20 (step S1002). Specifically, when the control device 10 periodically receives image data from the camera device 20 , the control section 100 stores the received image data in the image data storage area 1240 . In this case, the surrounding environment acquisition unit 102 reads and acquires the image data from the image data storage area 1240 in step S1002. Without being limited to this, for example, when the control device 10 controls the camera device 20 to receive image data, the surrounding environment acquisition unit 102 may directly acquire image data from the camera device 20 in step S1002.

The surrounding environment acquisition unit 102 recognizes the state of the bed apparatus 5 based on the image data (step S1004). Here, the state of the bed apparatus 5 means the state of the movable portion when at least a portion of the bed apparatus 5 is movable. At this time, it is preferable that the camera devices 20 are installed in the second pattern or the third pattern in FIG.

The surrounding environment acquisition unit 102 analyzes the image data and determines the height of the bottom portion of the bed apparatus 5 (floor height), the back lift angle, the knee lift angle, the inclination angle (tilt angle), and the interlocking of the back bottom and the knee bottom. It recognizes items such as whether or not the bed apparatus 5 operates as the state of the bed apparatus 5 . For example, the surrounding environment acquisition unit 102 outputs the floor height of the bed apparatus of "20 cm", the back-raising angle of "20 degrees", and the knee-raising angle of "10 degrees" as the first feature amounts.

The surrounding environment acquisition unit 102 also analyzes the image data and recognizes obstacles (step S1006). When the surrounding environment acquisition unit 102 recognizes an obstacle as the surrounding environment, it recognizes, as an obstacle, equipment or objects included in the image data that may cause a fall. Then, the surrounding environment acquisition unit 102 recognizes the type, position, orientation, presence or absence of the obstacle on the path, size, shape, presence/absence of motion, distance from the bed apparatus 5, etc., as necessary. The surrounding environment acquisition unit 102 outputs the first feature amount based on the information about the recognized obstacle.

For example, the surrounding environment acquisition unit 102 outputs that the type of obstacle "wheelchair" has a size of "56 cm×100 cm" and a position of "30 cm on the right side of the bed apparatus 5" as feature amounts.

The surrounding environment acquiring unit 102 may also recognize the orientation of the casters of the bed apparatus 5 and the locked state as the state of the bed apparatus and obstacles. The surrounding environment acquisition unit 102 outputs the recognition result as a feature amount.

Furthermore, the surrounding environment acquisition unit 102 recognizes a mitigation factor that reduces the risk of the user falling down (step S1008). The mitigation factor is a factor that reduces the risk of the user falling down (the probability of getting out of bed or falling down, and the probability of aggravation). Mitigating factors include the presence of staff and equipment that reduces the risk of falling, such as side fences, canes, and shock-absorbing mats. The surrounding environment acquisition unit 102 outputs, as a feature quantity, what the staff and the like have recognized, and what equipment such as side fences, walking sticks, and shock absorbing mats have been recognized to reduce the risk of falling.

In the above description, the surrounding environment acquisition unit 102 acquires the surrounding environment by recognizing the image data of the camera device 20, but may acquire the surrounding environment from other sources than the image data. For example, the surrounding environment acquiring unit 102 may directly acquire the back-raising angle, knee-raising angle, floor height, etc. of the bed device 5 from the bed control device 40 and output them as feature amounts. Further, the surrounding environment acquisition unit 102 may acquire the type, position, size, etc. of the obstacle by communicating with the peripheral device 60 provided on the obstacle.

Further, the surrounding environment acquisition unit 102 may selectively execute steps S1004 to S1008 in FIG. 4(a). For example, only one of the three steps may be performed, or two may be performed. The surrounding environment acquisition unit 102 may be executed in any combination.

FIG. 4B is a diagram showing an example of the first feature amount related to the surrounding environment stored in the surrounding environment storage area 1202. As shown in FIG. For example, the surrounding environment storage area 1202 stores a feature amount of the state of the bed apparatus 5, a feature amount based on obstacles, and a feature amount based on mitigation factors.

The surrounding environment storage area 1202 stores, as feature values related to the bed apparatus 5, floor height, back-raising angle, knee-raising angle, leg-raising angle, inclination angle (tilt angle), and whether or not the back bottom and the knee bottom are interlocked. and so on can be stored. When recognizing the state of the bed apparatus 5 , the surrounding environment acquiring unit 102 may also recognize the state of an accessory (for example, an air mattress) used with the bed apparatus 5 . In addition, the surrounding environment storage area 1202 may also store the operation state of the bed apparatus 5 as a feature amount. For example, the surrounding environment storage area 1202 stores a flag "1" as a feature amount when the back bottom and knee bottom of the bed device 5 are linked, or stores a rolling motion in the bed device 5. In this case, flag "1" may be stored as a feature amount.

Also, the surrounding environment storage area 1202 stores the type of obstacle, the position of the obstacle, etc. as the feature quantity related to the obstacle. The surrounding environment storage area 1202 may store the position of an obstacle as a feature amount from the bed device 5 or the distance from the camera device 20, or may store relative position coordinates (for example, the living room R as a virtual space). XYZ coordinates) may be stored. In addition, the surrounding environment storage area 1202 may store a feature amount corresponding to the types of obstacles such as "wheelchair" and "floor stand", or may store the size of the obstacle itself (e.g., width x cm, length y cm, height z cm) may be stored as the first feature amount.

Further, the surrounding environment storage area 1202 may store the first feature amount in the user information storage area 1204 in association with the user's ID and time.

(2) Posture Acquisition Unit (Second Acquisition Unit)
The posture acquisition unit 104 acquires the posture of the user from the image data captured by the camera device 20, and outputs a feature amount. Here, the posture of the user means the way the user holds the body, and is estimated based on the image data. In addition, the posture acquisition unit 104 acquires, for example, the position of the user (whether the user is out of bed) or the location (whether the user is sitting on the edge), in addition to the posture of the user. good too. As will be described later, "acquiring the posture" means specifying the posture based on the feature points of the user, but the concept of estimating the posture of the user based on some feature points is also used. includes.

Further, the posture acquisition unit 104 may store the acquired posture of the user in the user information storage area 1204 in association with the user together with the acquisition time.

Then, posture acquisition section 104 outputs a feature amount based on the acquired posture. The feature amount may be a feature amount for each posture, such as "1" for the sitting position, "2" for the supine position, and "0" for other positions. "10 degrees to the right"), the position of the head (e.g. "+5 degrees" to the vertical direction), the way the user holds the body (e.g. "the angle at which the right hand is raised", etc.) good. Further, the posture acquisition unit 104 may set all postures and postures of the body as attributes, and may output “1” as a feature amount if applicable.

The posture acquisition unit 104 stores the acquired second feature amount based on the posture of the user in the user information storage area 1204 in association with each user.

Posture acquisition processing executed by the posture acquisition unit 104 will be described with reference to FIG. First, the posture acquisition unit 104 acquires image data (step S1102). The posture acquisition unit 104 acquires image data using the same method as that used by the surrounding environment acquisition unit 102 to acquire image data.

The posture acquisition unit 104 recognizes the skeletal parts of the user as feature points from the acquired image data (step S1104). For example, the posture acquisition unit 104 recognizes the positions of the user's shoulders, face, hands, and feet, and also recognizes the positions of joints such as knee joints and elbow joints, through pattern image recognition. .

FIG. 5(b) is a diagram schematically showing the skeletal parts of the user. For example, the posture acquisition unit 104 recognizes the photographed skeletal parts of the user by pattern image analysis, and recognizes the feature points of the user. The posture acquisition unit 104 can recognize the skeleton of the user by detecting lines connecting the recognized feature points of the human body.

The posture acquisition unit 104 uses the feature points to determine the size of the user, the direction of the user's body (including the direction of the body, the direction of the face, and the direction of the line of sight), and the way the user holds the body. posture is obtained (step S1106). The posture of the user may be acquired by, for example, performing coordinate regression analysis on the feature points, or by using the result of machine learning. Posture acquisition section 104 then outputs a second feature amount based on the acquired posture.

Although the posture acquisition unit 104 has been described as acquiring the user's posture by recognizing feature points from captured image data, for example, by using a neural network, the user's posture can be obtained directly from the image data. may be obtained.

Also, other methods of acquiring the posture of the user may be used. For example, the user state detection method described in Japanese Patent Application Laid-Open No. 2008-206869 (title of the invention: user state detection system in bed, filing date: November 13, 2007), Japanese Patent Application Laid-Open No. 2009-118980 (Title of invention: bed device, filing date: February 27, 2007) can be used. This patent application is incorporated by reference in its entirety.

It has been described that the posture acquisition unit 104 described above outputs the feature amount based on the user's posture based on the image data captured by the camera device 20 . However, in addition to this, the posture acquisition unit 104 may acquire the posture of the user by using the sensor device 30, for example, and output the feature amount.

Also, the posture acquisition unit 104 may acquire the facial expression from the user's face by analyzing the image data. When the facial expression is acquired, the posture acquisition unit 104 outputs it as a second feature amount.

(3) Motion Acquisition Unit (Third Acquisition Unit)
The motion acquisition unit 106 acquires a user's motion based on information (sensor information) acquired by the camera device 20 or the sensor device 30, and outputs a third feature amount based on the user's motion. Here, the motion of the user means the movement of the user's body. For example, the motion of the user includes the amount of movement of the user, the movement of the user's hands and feet, the position of the center of gravity, the sway, the gait, the behavior, and turning over. In addition, the motion of the user includes not only the motion of standing up, but also the speed of standing up, the speed of walking, the direction of direction change, and the time required for direction change. The motion acquisition unit 106 may store the acquired motions of the user in the user information storage area 1204 in chronological order for each user.

A motion acquisition process executed by the motion acquisition unit 106 will be described with reference to FIG. First, the motion acquisition unit 106 acquires sensor information from the camera device 20 or the sensor device 30 (steps S1102 and S1104). The sensor information output by the sensor device 30 is, for example, vibration data. The motion acquisition unit 106 can acquire the user's motion by analyzing the vibration data or the image data (step S1106).

Here, the information that the motion acquisition unit 106 acquires from the camera device 20 is image data. The motion acquisition unit 106 estimates the user's skeleton from the image data in time series, and acquires the user's motion such as the position of the center of gravity, sway, gait, and behavior. Information that the motion acquisition unit 106 acquires from the sensor device 30 in order to acquire the user's motion includes, for example, the amount of change in the load on the bed, center-of-gravity data, and air pressure vibration data (user's vital data). (including body motion data, heartbeat data, and respiration data). That is, the motion acquisition unit 106 can acquire the user's motion based on the information acquired from the sensor device 30 .

The motion acquisition unit 106 then outputs the feature amount based on the acquired motion as the third feature amount, and stores it in the user information storage area 1204 . Here, what is stored as the third feature amount based on the motion is, for example, as follows.

- When it is acquired that the user's body movement is large, "1" is stored as the feature amount.
- When it is acquired that the user (or part of the user's body) has moved a predetermined distance or more, "1" is stored as a feature amount.
- Output "1" as a feature amount when it is acquired that the load change is greater than or equal to the threshold. Also, the numerical value of the load change is stored as a feature amount.
- The number of rollovers is stored as a feature amount.

In addition, especially when there is no change in the user's action, the action acquisition unit 106 may store the fact that there is no change in the user information storage area 1204 as a feature amount. That is, the motion acquisition unit 106 periodically (for example, every 1 second, 5 seconds, 1 minute, 5 minutes, etc.) acquires the motion of the user, and uses the third feature value based on the motion as the user information. It may be stored in storage area 1204 .

(4) Biometric information acquisition unit (fourth acquisition unit)
The biometric information acquisition unit 108 acquires biometric information based on the information acquired by the sensor device 30 . For example, the biological information acquisition unit 108 executes the biological information acquisition process to calculate a respiratory waveform and a heartbeat waveform based on the body motion received from the sensor device 30 . Then, the biological information acquiring unit 108 acquires the respiratory rate from the calculated respiratory waveform and the heart rate from the calculated heartbeat waveform as biological information values. The biometric information acquisition unit 108 calculates and outputs a fourth feature amount based on the biometric information value.

In addition, as a method for acquiring biological information such as these respiratory rate and heart rate, for example, Japanese Patent Application Laid-Open No. 2016-30177 (title of the invention: respiratory disorder determination device, respiratory disorder determination method and program, filing date: Heisei 26 July 30, 2003) can be used. Moreover, you may utilize well-known technique other than that.

In addition to the respiratory rate and heart rate, the biological information acquisition unit 108 can also acquire biological information values such as body temperature, blood pressure, and percutaneous arterial oxygen saturation (SpO2). In addition, the biological information acquisition unit 108 may acquire biological information (for example, pulse value and respiratory rate) that can be continuously acquired by the sensor device 30, blood pressure measured by a sphygmomanometer, and blood pressure measured by a thermometer. Any biological information that can be obtained discretely (spotwise) by an external device or the like, such as body temperature, may be obtained.

The biometric information acquisition unit 108 may store the biometric information value as it is as the fourth feature amount in the user information storage area 1204 for each user. Further, the biometric information acquisition unit 108 may acquire that the biometric information value is at the normal level, the caution level, or the warning level, for example, and output the fourth feature amount according to the level.

Also, the biometric information acquisition unit 108 may acquire information about the user that can be acquired based on the biometric information. For example, the biological information acquisition unit 108 may acquire the sleep state/awakening state of the user, and further acquire REM sleep and non-REM sleep as sleep states. When the user is in a sleeping state, the biological information acquisition unit 108 may output "1" as the sleep feature amount, and may output "0" as the sleep feature amount when the user is in the wakeful state. .

As a method for determining the sleep state of these users, for example, Japanese Patent Application Laid-Open No. 2010-264193 (title of the invention: sleep state determination device, program and sleep state determination system, application date: May 18, 2009 ), the method described in JP-A-2014-229118 (title of the invention: sleep state determination device, sleep state determination method and program, filing date: November 11, 2014) can be used. This patent application is incorporated by reference in its entirety. Alternatively, other known methods may be used to obtain the sleep/wake state of the user.

Also, the biometric information acquisition unit 108 may acquire the biometric information of the user from the camera device 20 . As biometric information that the biometric information acquisition unit 108 can acquire, for example, it is possible to acquire the user's facial expression, line of sight, and body movement (for example, restless leg syndrome during sleep). In addition, the biological information acquiring unit 108 can also detect body movement and acquire respiration and heartbeat by performing image analysis on image data. Also, the biometric information acquisition unit 108 can acquire body temperature by using a thermal infrared camera device as the camera device 20 .

(5) Disease information acquisition unit (fifth acquisition unit)
The disease information acquisition unit 110 acquires information (disease information) about a user's disease. For example, when the disease information acquisition unit 110 executes the disease information acquisition process, it connects to the server device 50 and accesses the electronic medical record DB 502 . Then, the disease information acquisition unit 110 acquires information about the user's disease. Here, the disease information broadly includes information related to the user's disease, and includes not only simple medical conditions, but also information on medication, information on treatment, information on surgery, information on hospitalization history, and limb paralysis. , precautions for meals, whether or not to use assistive devices, and precautions for users.

The disease information acquisition unit 110 stores the fifth feature amount based on the disease information in the user information storage area 1204 . The fifth feature amount may be, for example, a feature amount based on a history of illness or history of hospitalization. For example, the disease information acquisition unit 110 may output the number of days of hospitalization as the fifth feature amount. In addition, the disease information acquisition unit 110 may output “1” as the fifth feature value if the disease corresponds to a specific disease, or may output the fifth feature value based on the user's blood pressure value or blood sugar level. A feature amount may be calculated and output.

(6) Predicting Unit The predicting unit 112 predicts the risk of falling of the user and the time when there is a risk of falling. The prediction unit 112 outputs the risk leading to overturning/falling as a probability or as a level, based on the feature amount described above. Also, the prediction unit 112 predicts the time and duration when there is a risk of falling or falling based on the feature amount.

Further, the prediction unit 112 may further calculate an explanatory variable from the feature amount described above, and predict the risk of falling down based on the explanatory variable. That is, the prediction unit 112 selects one or a plurality of feature amounts from among the plurality of feature amounts, and calculates explanatory variables. Then, the prediction unit 112 uses the first to fifth explanatory variables calculated from the first to fifth feature amounts described above to predict whether or not the person will fall. A detailed description of the first to fifth explanatory variables will be given later.

The prediction unit 112 may output the risk or probability of falling/falling as a numerical value or as a level. Further, the prediction unit 112 may select a necessary one from the first to fifth feature amounts or the first to fifth explanatory variables and use it for the process of predicting a fall. may be used after weighting. Also, the prediction unit 112 may use a combination of feature amounts and explanatory variables. That is, the prediction unit 112 outputs the risk leading to overturning/falling as a prediction result from the feature amount and/or explanatory variables.

A prediction process executed by the prediction unit 112 will be described with reference to FIG. First, the prediction unit 112 executes the surrounding environment acquisition process in FIG. 4A to acquire a first feature amount (step S1202). Then, the prediction unit 112 calculates a first explanatory variable based on the first feature amount (step S1204). Here, the first explanatory variable is a variable that indicates how likely the user to be evaluated falls over based on the surrounding environment of the bed apparatus 5 . The first explanatory variable may directly include a feature amount according to need from among the feature amounts output from the surrounding environment acquisition process, or may be a matrix. Also, the first explanatory variable may be a numerical value obtained by summing (or integrating) points scored for each feature amount.

For example, the surrounding environment acquisition unit 102 obtains the floor height of the bed device 5, the bed posture, the presence or absence of an obstacle, the shape of the obstacle, the presence or absence of movement of the obstacle, the size of the obstacle, and whether the obstacle is on the route. Feature values are calculated based on attributes such as presence, distance of obstacles, whether the light is sufficiently bright, and whether the sound is quiet. The prediction unit 112 selects one or a plurality of feature quantities, multiplies them by predetermined weighting coefficients, sums them, and sets them as first explanatory variables (step S1204). The prediction unit 112 stores the first explanatory variable in the explanatory variable table 1220 .

Next, the prediction unit 112 executes the posture acquisition process of FIG. 5A to acquire a second feature amount (step S1206). The prediction unit 112 calculates a second explanatory variable based on the second feature amount (step S1208). Here, the second explanatory variable is a variable that indicates how likely the user to be evaluated falls over based on the user's posture. The second explanatory variable may include the feature amount output from the posture acquisition process as it is, or may be a matrix. Also, the second explanatory variable may be a numerical value obtained by summing (or integrating) points scored for each feature amount.

For example, the posture acquisition unit 104 calculates feature amounts based on attributes such as the size of the user, the orientation of the user's body, and the way the user holds the body. The prediction unit 112 selects one or a plurality of feature quantities, multiplies them by predetermined weighting coefficients, sums them, and uses them as second explanatory variables (step S1208). The prediction unit 112 stores the second explanatory variable in the explanatory variable table 1220. FIG.

Next, the prediction unit 112 executes the action acquisition process of FIG. 6 to acquire a third feature amount (step S1210). The prediction unit 112 calculates a third explanatory variable based on the third feature amount (step S1212). Here, the third explanatory variable is a variable that indicates how likely the user to be evaluated falls over based on the user's motion. The third explanatory variable may directly include the feature amount output from the motion acquisition process as required, or may be a matrix. Also, the third explanatory variable may be a numerical value obtained by summing (or integrating) points scored for each feature amount.

For example, the motion acquisition unit 106 determines whether the user's body movement was large, whether the user (or part of the user's body) moved a predetermined distance or more, or whether there was a change in the user's load. Based on this, the feature amount is calculated. The prediction unit 112 selects one or a plurality of feature amounts, multiplies them by predetermined weighting coefficients, sums them, and sets them as third explanatory variables (step S1212). The prediction unit 112 stores the third explanatory variable in the explanatory variable table 1220. FIG.

Next, the prediction unit 112 executes biometric information calculation processing to acquire a fourth feature amount (step S1214). The prediction unit 112 calculates a fourth explanatory variable based on the fourth feature amount (step S1216). Here, the fourth explanatory variable is a variable that indicates how likely the user to be evaluated falls over based on the user's biological information. The fourth explanatory variable may contain the feature amount output from the biometric information calculation process as it is, or may be a matrix. The fourth explanatory variable is a numerical value obtained by summing (or multiplying) the points scored for each feature amount.

For example, the biometric information acquisition unit 108 calculates feature amounts based on attributes such as whether the user's heart rate is stable, whether the user's breathing rate is stable, and whether the user has sufficient sleep. do. The prediction unit 112 selects one or a plurality of feature quantities, multiplies them by predetermined weighting coefficients, sums them up, and uses them as fourth explanatory variables (step S1216). The prediction unit 112 stores the fourth explanatory variable in the explanatory variable table 1220. FIG.

Next, the prediction unit 112 executes disease information acquisition processing to acquire a fifth feature amount (step S1218). The prediction unit 112 calculates a fifth explanatory variable based on the fifth feature amount (step S1220). Here, the fifth explanatory variable is a variable that indicates how likely the user to be evaluated falls over based on the user's disease information. The fifth explanatory variable may contain the feature amount output from the disease information acquisition process as it is, or may be a matrix. Also, the fifth explanatory variable is a numerical value obtained by summing (or integrating) the points scored for each feature amount.

For example, the disease information acquisition unit 110 determines whether the user has paralysis of the limbs, frequent urination, is taking medicine that causes dizziness, has a history of falling, or has rheumatism or the like that interferes with the use of the limbs. A feature amount is calculated based on attributes such as whether or not there is delirium. The prediction unit 112 selects one or a plurality of feature quantities, multiplies them by predetermined weighting coefficients, sums them up, and uses them as fifth explanatory variables. The prediction unit 112 stores the fifth explanatory variable in the explanatory variable table 1220. FIG.

The prediction unit 112 uses the first to fifth explanatory variables stored in the explanatory variable table 1220 to execute the process of predicting the fall risk (step S1222).

Specifically, using the first to fifth explanatory variables, the prediction unit 112 performs regression analysis, extracts feature points in units of pixels and obtains approximate expressions, Predict the probability of danger leading to , and the time in the future. In addition, hereinafter, when simply expressed as "probability", it means the probability of danger leading to a fall.

In addition, the prediction unit 112 may predict the probability and the future time leading to a fall by weighting the first to fifth explanatory variables. Also, the prediction unit 112 may weight the feature amount in each explanatory variable.

The prediction unit 112 may also predict the probability of falling and the future time using the prediction dictionary DB 1230, which is a prediction model. The prediction dictionary DB 1230 is dictionary data of learned models generated by any machine learning method.

The prediction unit 112 inputs the first to fifth explanatory variables to the AI program. The AI program uses the prediction dictionary DB 1230 to output the probability of danger leading to a fall and the future time.

In some cases, the prediction unit 112 outputs a level as the fall risk instead of the probability. For example, the prediction unit 112 may output that the risk is high when the probability is equal to or greater than a threshold (eg, 50% or more), and that the risk is low when the probability is less than the threshold (eg, less than 50%). Further, the prediction unit 112 may output a plurality of levels such as "high", "slightly high", "slightly low", and "low" by setting a plurality of thresholds.

The threshold for the prediction unit 112 to determine the risk may be commonly set by the staff or the like, or may be set for each user by the staff or the like. In addition, the control unit 100 may change it as appropriate according to the state of the user.

In this specification, the prediction unit 112 may output the risk of falling over as a probability or as a level. Therefore, even if the prediction unit 112 explains that the risk of falling over will be output in terms of probability, outputting it in terms of levels is included. In addition, even when the prediction unit 112 explains that the risk of overturning is to be output as a level, outputting it as a probability is included.

In addition, the prediction unit 112 can output the time and the duration when the risk of falling down becomes high. For example, the prediction unit 112 can output that the risk of tumbling and falling will be high at 6:00 in the morning, or that the risk of tumbling and falling will be high five minutes from the current time.

(7) Response Execution Unit The response execution unit 114 executes an instruction according to the output of the prediction unit 112 to each device, for example. For example, the response execution unit 114 instructs the notification unit 140, the terminal device 52, and the mobile terminal device 54 to make appropriate notifications as necessary, or instructs the display unit 130 to output advice. or

The operation of the handling execution process executed by the handling execution unit 114 will be described with reference to FIG. First, the prediction unit 112 executes notification processing when the probability of danger leading to a fall (hereinafter simply referred to as "probability") is equal to or greater than a predetermined threshold (step S1302; Yes→step S1304).

For example, when the probability is equal to or greater than a predetermined threshold value (preferably 40% to 60%), the response execution unit 114 determines that the risk of falling is high, and instructs the notification unit 140 to make a notification. The notification unit 140 receives the instruction and gives notification.

In addition, the response execution unit 114 may instruct, via the communication unit 150, that other devices (the terminal device 52 and the mobile terminal device 54) should notify. Further, the response execution unit 114 may instruct a device set by the staff or the like to notify, or may change the device for giving the instruction according to the probability or the situation of the user or the staff.

For example, when the probability is 40% or higher, the response execution unit 114 instructs the notification unit 140 only to perform notification. However, when the probability is 50% or higher, the response execution unit 114 may instruct not only the notification unit 140 but also the terminal device 52 and the mobile terminal device 54 to notify. Further, when each device performs notification, the response execution unit 114 may instruct only the portable terminal device 54 of a staff member who is free or who is in a short distance to perform notification.

Further, the notification unit 140 may output an alarm, or may notify by light emission or vibration as a notification method. Also, the notification unit 140 may display a warning on the display unit 130 . In addition, the notification unit 140 may also announce, for example, layout precautions, or may notify the user of a warning by voice.

Subsequently, the response execution unit 114 executes device control processing (step S1308) when it is necessary to issue a control instruction to each device (step S1306). The response execution unit 114 may instruct, for example, the bed control device 40 to automatically adjust the floor height and the back bottom angle of the bed by executing the device control process. In addition, the response execution unit 114 also instructs, for example, to automatically turn on the foot light to the toilet or the exit, instructs to automatically switch the operation of the sensor setting button to disabled, You may give an instruction to automatically change the setting of .

Subsequently, if the advice output is possible, the response execution unit 114 executes advice output processing using the fall/fall DB (step S1310; Yes→step S1312). By executing the advice output process, the response execution unit 114 automatically selects equipment according to the fall risk level, or presents a room layout plan to a predetermined or nearby staff member, for example. good too.

Next, when executing the learning process, the correspondence executing unit 114 may execute the learning process of the prediction dictionary DB 1230 . For example, the response execution unit 114 instructs the display unit 130 to display that there is a high risk of falling. At this time, if a staff member or the like comes to help, but there is no possibility of the user falling down, the staff member or the like inputs through the operation unit 135 that there is no possibility. Conversely, when the notification is given by the instruction of the countermeasure execution unit 114, the staff or the like inputs through the operation unit 135 that there was an actual risk of falling.

Information input by these staff members is stored in the prediction dictionary DB 1230 . Further, based on the information input by the staff or the like, the response execution unit 114 performs machine learning to learn the prediction dictionary DB 1230 .

[1.2.2 Other configurations]
The display unit 130 displays various information. For example, the display unit 130 is configured with a liquid crystal display or an organic EL display. Also, the display unit 130 may be another display device connected by HDMI (registered trademark) or D-SUB.

The operation unit 135 inputs various operations from a user, a staff member, or the like. For example, input devices such as a remote control and a nurse call are also included. A touch panel formed integrally with the display unit 130 also functions as the operation unit 135 .

The notification unit 140 performs notification. For example, it may be a speaker capable of outputting an alarm sound, or a device that notifies with light or vibration. Further, the notification unit 140 may output the content to be notified by voice output, or display the content of the warning on the display unit 130 for notification.

The communication unit 150 communicates with other devices. For example, it includes a LAN interface that connects to a network, a USB that connects to other devices, and a short-range wireless communication unit. The communication unit 150 may be a communication device connectable to a mobile phone communication network such as LTE/5G.

The camera device 20 is a photographing device for photographing surrounding conditions. The camera device 20 can shoot still images and moving images.

The sensor device 30 can acquire the user's biological information and the user's position and posture by detecting the user's body movement and the like. Moreover, the sensor device 30 may acquire biological information by detecting a load using a load cell provided in the bed device 5 .

The bed control device 40 controls the bed device 5 . For example, the bed control device 40 realizes back raising/back lowering and leg raising/leg lowering actions by controlling the back bottom, waist bottom, knees and bombs, and leg bottoms. The bed control device 40 can also change the floor height of the bed device 5 .

The server device 50 stores an electronic medical record DB 502 that stores information about diseases of users. The server device 50 is a server that manages an in-hospital system, and may be, for example, an electronic chart server or an ordering system server. In addition to basic information such as the user's name, date of birth, blood type, etc., the electronic medical chart DB 502 of the server device 50 stores disease history, examination history, surgery information, medication information, precautions, hospital visit history, and fall history. Various types of information necessary for such treatment are stored. It should be noted that these pieces of information are examples, and the server device 50 does not necessarily have to store and manage them. For example, information on surgery may be managed by a surgery management server (not shown), and information on medication may be managed by a medication server (not shown).

A peripheral device 60 is provided in another device. For example, wheelchairs, treatment tables, walking sticks, and other obstacles can be equipped with IoT modules. These obstacles can output the positions of the obstacles and the like to the control device 10 as the peripheral device 60 .

[1.3 Operation example]
An operation example using the system 1 will be described below. Note that the feature amount is calculated by the control unit 100 executing each function. Then, the prediction unit 112 outputs the risk of falling down based on the feature quantity and/or the explanatory variables stored in the explanatory variable table 1220 . The response execution unit 114 issues an instruction such as notification based on the output risk (for example, probability) leading to a fall.

The operation example described below explains the notification pattern according to what kind of information is acquired and how it is used as a feature amount. That is, the following operation example explains a highly effective combination among a plurality of feature quantities and explanatory variables.

[1.3.1 First operation example]
A first operation example will be described with reference to FIG.

A first operation example will be described with reference to FIG. That is, the prediction unit 112 acquires the first feature amount related to the obstacle output from the surrounding environment acquisition unit 102 (step A104). At this time, the feature amount of the obstacle is calculated by the surrounding environment acquisition unit 102 based on the image data captured by the camera device 20 . The camera device 20 acquired as image data is either image data capturing the surrounding environment of the bed device 5 in FIG. 20 must be image data of the surrounding environment.

The first feature value is calculated from attributes such as the type of obstacle (wheelchair, IV pole), size, shape, color, location, presence or absence on the route, presence or absence of movement, and the like. ing.

The prediction unit 112 determines whether the distance from the camera device 20 or the bed device 5 to the obstacle is less than a threshold value (step A106) or whether the obstacle is located within the specified range (step A108).

Here, when the distance to the obstacle is less than the threshold, the prediction unit 112 outputs that the probability of falling is high (step A106; Yes). Moreover, the prediction unit 112 similarly outputs that the probability is high when there is an obstacle within the specified range (step A106; No → step A108; Yes). That is, the prediction unit 112 predicts that there is a high risk of the user falling.

When the prediction unit 112 outputs that the probability of falling is high, the response execution unit 114 instructs the notification unit 140 to make a notification (step A110). In addition, the response execution unit 114 may issue an instruction to announce precautions regarding the layout using the display unit 130 in addition to issuing the notification from the notification unit 140 . Note that the prediction unit 112 may perform learning on the prediction dictionary DB 1230 based on the feature amount and explanatory variables of the environment when there is an obstacle and the user does not fall.

Specifically, for example, the prediction unit 112 instructs to make a notification when the distance to the obstacle is less than "1 meter" as the first feature value. Further, the prediction unit 112 issues an instruction to notify when there is an obstacle in the direction in which the patient leaves the bed apparatus 5 .

Also, the prediction unit 112 may calculate the first explanatory variable by combining one or more of these feature amounts. Then, the prediction unit 112 can output a probability based on the first explanatory variable.

In addition, the obstacle may include movable parts such as casters of the bed apparatus 5 that are not in proper positions. In addition, the first feature amount includes the arrangement position and orientation of the bed device 5 (for example, is the bed device 5 arranged in an appropriate position, or is the bed device 5 oriented in an appropriate direction?). Attribute-based features may also be included. The prediction unit 112 outputs whether the risk of falling is high from the feature amount or the first explanatory variable based on the feature amount.

In addition, in the first operation example, the prediction unit 112 can predict the risk of overturning based on a plurality of feature amounts and explanatory variables. For example, FIG. 9(b) is a diagram showing an example of the correspondence between feature amounts, explanatory variables, and results output by the prediction unit 112. As shown in FIG.

For example, the first case will be described. If there is an obstacle, the prediction unit 112 predicts that there is a high risk of overturning based on the first feature quantity indicating that there is an obstacle. Moreover, when the brightness of the living room is dark, the prediction unit 112 predicts that the risk of falling down is high based on the first feature value indicating the brightness of the living room. As a result, the prediction unit 112 predicts that there is a high risk of falling, and outputs the result.

Similarly, as in the second case, when the brightness is low, the risk of falling is high, so even if there is an obstacle with a low risk, the prediction unit 112 predicts that the risk of falling is high. ,Output. Similarly, in the third case, the prediction unit 112 predicts that the risk of falling over is high and outputs the result.

In the fourth case, the prediction unit 112 predicts that the risk of overturning is high, based on the first feature amount that there is an obstacle. However, since the user is in a sleeping state, the prediction unit predicts that the risk of falling is low based on the third feature amount. In this case, the prediction unit 112 predicts and outputs that the overall risk of overturning is low. It should be noted that it is also possible to provide advice for reducing the risk of falling, such as removing obstacles.

In addition, since the staff is around the bed apparatus 5 as a mitigating factor, the prediction unit 112 predicts that the risk of falling over is low. In this case, even if there is an obstacle, the prediction unit 112 predicts that the risk of overturning is low.

[1.3.2 Second operation example]
A second operation example will be described with reference to FIG.

That is, the surrounding environment acquiring unit 102 acquires the floor height of the bed apparatus 5 as a feature amount (step B104). Further, the control unit 100 acquires the height of the user from the user information storage area 1204 or the electronic medical record DB 502, and specifies the ideal floor height of the bed apparatus (step B106). The ideal floor height of the bed apparatus may be calculated from a table pre-stored in the storage unit 120 by the control unit 100, or calculated from the user's height. For example, when the height of the user is "150 cm", the ideal floor height of the bed apparatus 5 is specified as "60 to 80 cm". The height of the user may be directly input by the user or staff.

Also, for example, the control unit 100 acquires the length of the user's lower limbs from the user's skeleton. Then, the control unit 100 may specify, as the ideal floor height, the height at which the user's feet touch the ground when the knee angle is 90° with respect to the floor when the user takes the edge-sitting posture. The prediction unit 112 outputs the floor height of the bed apparatus as a feature amount.

Then, the prediction unit 112 determines if the floor height of the bed apparatus 5 specified in step B106 and the floor height of the ideal bed apparatus deviate by a predetermined floor height or more (if they are apart by a predetermined threshold value or more). (Step B108; Yes), it is predicted that the probability of falling is high.

Then, the response execution unit 114 executes notification processing (step B110) and instructs the bed control device 40 about the floor height of the bed device 5 . The bed control device 40 automatically adjusts the floor height of the bed device 5 to the ideal floor height based on the instruction from the response execution unit 114 (step B112).

In addition, in the second operation example, the prediction unit 112 can predict the risk of overturning based on a plurality of feature amounts and explanatory variables. For example, FIG. 10(b) is a diagram showing an example of the correspondence between feature amounts, explanatory variables, and results output by the prediction unit 112. As shown in FIG.

For example, the first case will be described. If the risk is high because the deviation between the ideal floor height and the floor height of the bed apparatus is large, the prediction unit 112 predicts that the risk of falling is high. Moreover, when the brightness of the living room is dark, the prediction unit 112 predicts that the risk of falling down is high based on the first feature value indicating the brightness of the living room. As a result, the prediction unit 112 predicts that there is a high risk of falling, and outputs the result.

Similarly, as in the second case, when the brightness is dark, the risk of falling over is high. The unit 112 predicts that there is a high risk of overturning and outputs the result. Similarly, in the third case, since the user has delirium, the prediction unit 112 predicts and outputs that the risk of falling is high.

In the fourth case, there is a gap between the ideal floor height of the bed apparatus and the current floor height of the bed apparatus, so the prediction unit 112 predicts that the risk of falling over is high. However, since the user is in a sleeping state, the prediction unit 112 predicts that the risk of falling over is low based on the third feature amount. In this case, the prediction unit 112 predicts and outputs that the overall risk of overturning is low. It should be noted that advice such as changing the floor height to reduce the risk of falling may be notified.

In this way, the prediction unit 112 can predict the fall risk based on a plurality of feature quantities and explanatory variables. For example, the prediction unit 112 predicts that the risk of overturning is high in the operation example of FIG. A low risk of falls may be expected.

[1.3.3 Third operation example]
A third operation example will be described with reference to FIG.

The prediction unit 112 identifies the positions of the user's head, shoulders, feet, hands, knees, and center of gravity based on the second feature amount output from the posture acquisition unit 104 (step C106).

Note that the posture of the user and the center of gravity may be acquired by the posture acquisition unit 104 . The prediction unit 112 uses feature amounts based on the user's posture and the center of gravity acquired by the posture acquisition unit 104 . It should be noted that it is not necessary to specify all of these skeletons, but the more points that are specified, the more accurate the posture estimated by posture acquisition section 104 is. The prediction unit 112 may use these second feature amounts as they are, or may select one or a plurality of feature amounts to calculate the second explanatory variable.

The prediction unit 112 uses the second feature amount or the second explanatory variable specified in step C106 to predict the probability of falling (whether the risk is high) (step C108). Specifically, based on the absolute position and relative position of each part, the prediction unit 112 predicts whether the head is not on the base of support, the body is bent, the body is sitting on the edge, the sitting position is standing, or the body is standing. It is determined whether the person is in a posture that is not suitable for walking, etc. Then, when it is determined that the user's posture poses a high risk of overturning, the response execution unit 114 executes notification processing (step C110). Furthermore, the risk of falling from each position may be calculated by machine learning without relying on the above criteria.

Specifically, the prediction unit 112 identifies, for example, that the user's center of gravity is not on the base of support and that the user is not in the correct posture, or that the user is in the correct posture due to medication. By identifying that the patient is sitting on the far right side even though the movement of the right side of the body is poor due to the treatment, it is identified that the patient is in a high-risk posture.

In addition, in the third operation example, the prediction unit 112 can predict the risk of overturning based on a plurality of feature amounts and explanatory variables. For example, FIG. 11(b) is a diagram showing an example of the correspondence between feature amounts, explanatory variables, and results output by the prediction unit 112. As shown in FIG.

For example, the first case will be described. Based on the second feature amount or the second explanatory variable, the prediction unit 112 predicts that the risk of falling is high when the risk based on the posture is high. Moreover, the prediction unit 112 predicts that the risk of falling is high when the risk based on the floor height of the bed apparatus is high, based on the first feature amount or the first explanatory variable. As a result, the prediction unit 112 predicts that there is a high risk of falling, and outputs the result.

As in the second case, the prediction unit 112 predicts that the posture-based fall risk is medium based on the second feature amount or the second explanatory variable. Moreover, the prediction unit 112 predicts that the risk of falling is high when the risk based on medication is high, based on the fifth feature amount or the fifth explanatory variable. As a result, the prediction unit 112 predicts that there is a high risk of falling, and outputs the result.

As in the third case, the prediction unit 112 predicts that the risk of falling based on the posture is high, based on the second feature amount or the second explanatory variable. Also, the prediction unit 112 acquires the presence of staff, which is a mitigating factor, based on the first feature amount or the first explanatory variable, and predicts that the risk of falling is low. As a result, the prediction unit 112 predicts and outputs that the risk of falling is low.

As in the fourth case, the prediction unit 112 predicts that the risk of falling based on posture is high, based on the second feature amount or the second explanatory variable. Based on the first feature amount or the first explanatory variable, the prediction unit 112 also predicts that there is equipment that is a mitigating factor (for example, a shock absorbing mat, a cane), and that the risk of falling is low. As a result, the prediction unit 112 predicts that the risk of overturning is medium and outputs the result.

In this way, the prediction unit 112 can predict the fall risk based on a plurality of feature quantities and explanatory variables. For example, the prediction unit 112 may predict that the risk of overturning is low according to the type of facility that is the mitigation factor.

[1.3.4 Fourth Operation Example]
A fourth operation example will be described with reference to FIG.

That is, the prediction unit 112 acquires the user's head, shoulders, feet, hands, knees, and center of gravity from the posture acquisition unit 104 (step D106). That is, the second feature amount acquired from the posture acquisition unit 104 is used, or the second explanatory variable calculated from the second feature amount is used.

The prediction unit 112 predicts the risk of falling from the trajectory of the user's center of gravity specified based on the third feature amount or the third explanatory variable (step D108). Specifically, the prediction unit 112 determines that the trajectory of the center of gravity of the user is swaying when the trajectory of the user's center of gravity fluctuates greatly with respect to the direction in which the user travels, and increases the risk. For example, when the user shifts from a sitting position to a standing position, the position of the center of gravity of the upper half of the body usually rises, but the prediction unit 112 predicts that the center of gravity of the user will lower or rise from the lower half of the body. In some cases, identify abnormal behavior and increase the risk of falls. A fall risk may be calculated from each trajectory by machine learning without relying on the above criteria. Here, when the prediction unit 112 predicts that the risk of overturning is high, the response execution unit 114 executes notification processing (step D110).

Note that the prediction unit 112 may combine a plurality of feature amounts and explanatory variables to predict the fall risk. For example, the prediction unit 112 may predict the fall risk by combining the fifth feature value/fifth explanatory variable related to medication or disease, or predict the fall risk to be low if there is a mitigating factor. You may

[1.3.5 Fifth Operation Example]
A fifth operation example will be described with reference to FIG.

That is, the prediction unit 112 acquires the shoulder position of the user based on the third feature amount output from the motion acquisition unit 106 or the third explanatory variable calculated from the third feature amount (step E106). .

Based on the third feature amount or the third explanatory variable, the prediction unit 112 determines that the risk of falling is high when the position of the user's shoulder deviates from the direction of travel by a predetermined threshold or more from the direction of travel. (Step E108; Yes). In this case, the response execution unit 114 executes notification processing (step E110). Here, when the position of the user's shoulder deviates from the direction of travel by a predetermined threshold or more from the direction of travel, it means that the user's original movement deviates from the direction of travel, and even if the movement is back and forth. , left and right movement.

[1.3.6 Sixth Operation Example]
A sixth operation example will be described with reference to FIG.

The prediction unit 112 acquires the facial expression of the user based on the second feature amount from the posture acquisition unit 104 or the second explanatory variable calculated from the second feature amount. Also, the prediction unit 112 acquires the motion of the user based on the third feature amount or the third explanatory variable calculated from the third feature amount from the motion acquisition unit 106 (step F104).

Subsequently, the prediction unit 112 identifies a change area from the motion of the user (step F106). The changed area of the user means, for example, a change in facial expression or a movement of the body. Also, the prediction unit 112 acquires a change in body temperature when the body temperature can be acquired from the third feature amount (third explanatory variable).

The prediction unit 112 determines the risk of falling from the movement and facial expression of the user (step F110). Here, when the prediction unit 112 determines that the user is in poor physical condition from the movement of the user's facial expression, the transition, etc., the response execution unit 114 changes the setting of the bed leaving sensor via the communication unit 150, for example. is instructed (step F112). Specifically, the prediction unit 112 issues an instruction to change the setting of the bed exit sensor to a setting that facilitates one-step notification (step F114). In addition, the prediction part 112 may perform an alerting|reporting process collectively at this time.

If the prediction unit 112 determines that the user's body temperature has exceeded the threshold (Step F114; Yes), the response execution unit 114 executes notification processing. (Step F116).

In this case, the response execution unit 114 may instruct the terminal device 52 or the mobile terminal device 54 to perform notification, for example, in addition to executing the instruction to perform notification by the notification unit 140 . In addition, the response execution unit 114 may instruct the portable terminal device 54 owned by a pre-registered nurse or medical staff, or the nurse or medical staff acquired from the electronic medical record DB 502 to perform notification. .

[1.3.7 Seventh Operation Example]
A seventh operation example will be described with reference to FIG.

The prediction unit 112 acquires the user's disease information from the disease information acquisition unit 110 as a fifth feature amount or a fifth explanatory variable calculated from the fifth feature amount (step G102). In addition, the prediction unit 112 may also acquire a first feature amount (first explanatory variable) from the surrounding environment acquisition unit 102 . The prediction unit 112 acquires information about hospitalization history and medical condition history from the fifth feature amount (fifth explanatory variable).

The prediction unit 112 predicts the level of the risk of falling based on the user's disease information, such as hospitalization history and medical condition history (step G104). For example, the prediction unit 112 predicts that the risk of falling is high if there is a history of falling accidents, or that the risk of falling is high if the person is taking drugs that tend to lead to dizziness due to a medical condition. Here, the prediction unit 112 uses an AI program, for example, and refers to the prediction dictionary DB 1230 to predict the risk level of falling.

If the prediction unit 112 determines that the risk of falling over is high (step G104; Yes), the response execution unit 114 selects equipment that reduces the risk of falling over (for example, a shock absorbing mat, a cane), The staff or the like is instructed to present (step G108). Equipment that reduces the risk of falling may include equipment that prevents the user from becoming seriously ill. Specifically, the response execution unit 114 instructs the terminal device 52 (portable terminal device 54) to display, or instructs the staff or the like to send an e-mail or the like. At the same time, the response execution unit 114 may present a room layout plan (step G108). For example, the response execution unit 114 may request that the distance between the bed and the headrest be 20 cm or more, that the height of the bed be 40 cm or less, and that no obstacle be placed within 30 cm from the side of the bed. , is displayed on the display unit 130 .

Further, the response execution unit 114 may present the appropriateness of the room layout. As an example, if the user has left hemiplegia, advice is displayed as to whether or not the user can exit the living room from the right side of the bed apparatus. For example, the response execution unit 114 displays a message such as increasing the distance between the bed and the headrest by a certain distance, setting the bed to an appropriate height, and not placing an obstacle within a certain distance from the side of the bed. 130.

[1.3.8 Eighth operation example]
An eighth operation example will be described.

The surrounding environment acquisition unit 102 acquires one or a plurality of items such as position, size, shape, color, etc. as information related to obstacles (mainly fixtures here) from the image data captured by the camera device 20. and output as features.

As a result, the prediction unit 112 determines that the risk is high from the following feature amounts, and the response execution unit 114 performs appropriate response. Some examples are described below.

(first example)
The prediction unit 112 predicts the risk of falling using the first feature amount or the first explanatory variable. For example, the prediction unit 112 compares the appropriate living room layout (initial position) with the current layout of the living room, and determines whether the layout difference is within a reference range. At this time, the following cases can be considered as the determination method.
(1) Deviation from the standard layout For high-risk patients, nurses cannot reach from the bed (do not allow them to pick up objects), and cannot reach when getting up from the bed (unstable support). In order to create a layout that does not allow the use of objects, the layout is used as a reference layout. The prediction unit 112 predicts a case where an object that was out of reach from the bed or an object that was out of reach in a sitting position comes into reach for a patient at risk. judge. The prediction unit 112 may also determine whether there are any objects that should be within reach, such as shoes, walking sticks, nurse call buttons, and the like.
(2) When there is something unexpected For example, things on the bed (tissue boxes, earphones, etc.) fall, or family members use a folding chair that is folded and stored in the gap between the bedside table and the wall. It is determined that it is left as it is.

An appropriate room layout may be stored in the storage unit 120 in advance, or may be set by a staff member or the like. A suitable room layout may be obtained from an external server.

The response execution unit 114 performs notification processing or gives advice based on the result of comparing the appropriate living room layout and the current living room layout. For example, the response execution unit 114 instructs the display unit 130 to give advice on the type of the article outside the reference range and to which side X cm the article should be moved.

(Second example)
The prediction unit 112 detects that the user is caught by a side table or the like and tries to stand up when the user stands up from the bed. Specifically, the prediction unit 112 directly refers to the image data acquired by the camera device 20, detects that the user's hand is on the side table when standing up, and determines that the risk is high. In addition, the prediction unit 112 detects that the distance between the position of the user in the edge-sitting state and the equipment at the end of the user's outstretched hand is less than a predetermined distance, and determines that the risk is high. do.

When the prediction unit 112 predicts that the risk is high, the countermeasure execution unit 114 issues an instruction to notify that effect, an instruction to automatically move equipment, or an instruction to lock the equipment. For example, in the case of the peripheral device 60 having an IoT unit built into the side table, the response execution unit 114 instructs the peripheral device 60 to move or to perform a lock operation.

(Third example)
The prediction unit 112 detects whether the curtains are closed and whether the user is trying to move with one hand (for example, using a cane) in the multi-bed room. For example, when the user uses a cane, the prediction unit 112 predicts that the risk of falling is high because the user opens and closes the curtain with one hand.

When the prediction unit 112 predicts that the risk of falling over is high, for example, when the user is about to stand up from the bed apparatus or when the user is about to leave the living room, the response execution unit 114 114 instructs the peripheral device 60 to automatically open the curtain. Further, when the user is holding a cane and is about to touch the curtain, the response execution unit 114 may forcibly execute the notification process.

(Fourth example)
The prediction unit 112 predicts that the risk of falling over is high when it is detected that the floor in the living room is wet. Specifically, the prediction unit 112 identifies that the floor is wet based on the reflectance of the floor from the image data based on the first feature amount. In addition, the surrounding environment acquisition unit 102 analyzes the image data, determines the cleaning staff from their clothes, skeleton, shoes, etc., and identifies that the cleaning staff has cleaning tools such as mops, thereby performing floor cleaning work. to get what has been done. The prediction unit 112 detects that the floor is wet, including the fact that the cleaning work has been performed and the elapsed time since the cleaning work was performed.

Then, when detecting that the floor is wet, the prediction unit 112 predicts that the risk of overturning is high. If the prediction unit 112 predicts that the risk of falling over is high, the response execution unit 114 executes notification processing when the user tries to get out of the bed apparatus.

The response executing unit 114 determines whether the user leaves the bed based on the output from the posture acquisition unit 104 and the motion acquisition unit 106 as the determination of leaving the bed. Further, the response execution unit 114 may analyze the image data and determine that the user leaves the bed when the user's body position is closer to the side rail.

In addition, the response execution unit 114 not only informs that the floor is wet, but also makes an announcement warning that the floor is slippery for a certain period of time after the cleaning staff leaves. may

(Fifth example)
The ambient environment acquisition unit 102 acquires the brightness of the image data, acquires the brightness of the room (distinction between day and night), and outputs it as a first feature amount. From the first feature amount, the prediction unit 112 predicts that the risk of falling is high when the brightness of the image data is less than a predetermined threshold.

The response execution unit 114 changes the setting of the bed sensor when the prediction unit 112 predicts that the risk of falling is high. Specifically, the response execution unit 114 raises the setting of the notification level of the sensor for getting out of bed by one level, thereby transitioning to a situation in which notification is easy.

Further, even when the predicting unit 112 has reduced the risk of falling (that is, when the brightness of the image data exceeds a predetermined threshold value), the response executing unit 114 changes the setting of the bed leaving sensor. good. In this case, the response execution unit 114 lowers the setting of the notification level of the sensor for getting out of bed by one step, thereby transitioning to a situation in which unnecessary notification is not performed.

Further, the response execution unit 114 may issue an instruction to control the foot light. For example, the response execution unit 114 may instruct to automatically turn on footlights leading to a destination such as a restroom.

(Sixth example)
Based on the posture of the user specified from the second feature quantity output by the posture acquisition unit 104, the prediction unit 112 predicts that the user's hand is near the setting switch and that the user will operate the setting switch. Predict high risk if unauthorized person. Here, various switches such as a setting switch for the sensor device 30 and an operation switch for operating the bed device 5 can be considered as the setting switch for determining the risk. The staff or the like may set in advance in the system which switches the user should not operate.

When the predicting unit 112 predicts that the risk of falling down is high, the countermeasure executing unit 114 automatically switches the operation of the setting switch to disabled. This makes it possible to prevent the user from arbitrarily changing the sensor settings or turning off the sensor device 30 .

[1.3.9 Ninth Operation Example]
A ninth operation example will be described.

That is, the posture acquisition unit 104 acquires one or a plurality of information regarding the posture of the user, such as the positions of the head, shoulders, and hands, from the image data captured by the camera device 20, and obtains the position of the head, shoulders, and hands as the second feature amount. Output.

As a result, the prediction unit 112 determines that the risk is high from the following feature amounts, and the response execution unit 114 performs appropriate response. Some examples are described below.

(first example)
Based on the second feature amount or the second explanatory variable, the prediction unit 112 predicts that the risk of falling is high when the user's forward bending angle exceeds a certain threshold.

The response execution unit 114 executes notification processing when the prediction unit 112 predicts that the risk of falling is high. For example, the response execution unit 114 instructs a nearby staff member or the like to make an announcement to support the user. It should be noted that whether the staff or the like is close may be determined based on position information from a beacon, mobile terminal, or the like owned by the staff or the like. Further, the response execution unit 114 may predict the position of the staff or the like based on the treatment history and issue an instruction to notify the staff.

(Second example)
The prediction unit 112 calculates the probability of leaving the bed from the eye movement of the user identified based on the second feature amount. For example, the prediction unit 112 recognizes the line-of-sight tendency when the user leaves the bed, based on the line-of-sight movement of the user, such as looking at the destination or looking at the feet on the bed. Then, the prediction unit 112 predicts that the risk of falling over is high when the probability of the user leaving the bed is equal to or greater than the threshold.

The response execution unit 114 executes notification processing when the prediction unit 112 predicts that the risk of falling is high. Specifically, the response execution unit 114 issues an instruction to notify that "lower the bed height", and an instruction to automatically lower the floor height of the bed apparatus 5 or to the minimum floor height.

(Third example)
The prediction unit 112 detects that the position of the user's head specified based on the second feature amount is shifted from the prescribed position by a threshold value or more. For example, when the position of the user's head deviates from the specified position downward (toward the feet) by 10 cm or more, it is predicted that the risk of falling is high.

The response execution unit 114 executes notification processing when the prediction unit 112 predicts that the risk of falling is high. For example, the response execution unit 114 instructs the terminal device of the staff member in charge to notify, so that the staff member or the like can correct the position of the user. Thereby, it is possible to prevent the user from sliding down on the bed apparatus.

(Fourth example)
The prediction unit 112 acquires the position of the user's hand on the side rail based on the second feature amount. Then, the prediction unit 112 predicts that the position of the user's hand is the position of the treatment set for each equipment (if the equipment is an assistance bar, the position set for each assistance bar), the position of the user on the bed apparatus 5 It is judged whether it is appropriate or not by comparing with the position of the body. When the position of the user's hand deviates from the appropriate position by a threshold value or more, the prediction unit 112 predicts that the gripping position of the user's hand is inappropriate and that there is a high risk of falling.

The response execution unit 114 executes notification processing when the prediction unit 112 predicts that the risk of falling is high. For example, the response execution unit 114 announces the contents such as "If you grip the right (left) side by X cm, it will be easier to stand up" for the hand whose grip position is inappropriate.

(Fifth example)
The prediction unit 112 determines whether or not the user will change direction after standing up based on the second special quantity, based on the shoulder position of the skeleton and the first feature quantity output from the surrounding environment acquisition unit 102. Predict. The prediction unit 112 predicts that there is a high risk of falling when the user is likely to change direction after getting up.

The response execution unit 114 executes notification processing when the prediction unit 112 predicts that the risk of falling is high. For example, the response execution unit 114 instructs the staff to notify the user of the action, or instructs the user to make an announcement such as "be careful when turning".

[1.3.10 Tenth Operation Example]
A tenth operation example will be described.

The motion acquisition unit 106 and the biometric information acquisition unit 108 acquire, for example, one or a plurality of user motions and biometric information (sleep/awake state, etc.) as information related to user behavior. The motion acquisition unit 106 outputs the acquired information about the motion as a third feature amount, and the biometric information acquisition unit 108 outputs the acquired information about the biometric information as a fourth feature value.

As a result, the prediction unit 112 determines that the risk is high from the following feature amounts, and the response execution unit 114 performs appropriate response. Some examples are described below.

(first example)
The prediction unit 112 detects the time from the user's wakefulness to the time when the user takes action and the tendency of movement based on the fourth feature amount. In other words, the prediction unit 112 detects the movement of the user from sleep to getting out of the bed apparatus. Then, the prediction unit 112 predicts that the risk of falling is high by referring to the prediction dictionary DB 1230 .

That is, the prediction unit 112 calculates the average time from awakening to getting out of bed for each user. This makes it possible to identify the timing (time, elapsed time) from awakening to getting out of bed. The prediction unit 112 determines the order of priority for each user according to the timing of leaving the bed and the risk of falling. Then, based on the prediction of the user's movement to get out of the bed apparatus, it is possible to notify the staff or the like of the order of priority. The prediction unit 112 may predict that the user has a high risk of falling if the user wakes up in a shorter time than the average time from awakening to getting up.

The response execution unit 114 executes notification processing when the prediction unit 112 predicts that the risk of falling is high. For example, the response execution unit 114 instructs to automatically adjust the sensor settings according to the tendency from awakening to behavior, instructs to automatically control the bed device 5, and automatically controls the peripheral device 60. give instructions. Specifically, the response execution unit 114 issues an instruction to raise the notification level of the bed sensor by one level, or issues an instruction to lock the side table with the peripheral device 60 .

(Second example)
The prediction unit 112 refers to the time information since the user left the bed last time together with the image data (video data). As a result, for example, the prediction unit 112 can predict that if the user used the toilet 15 minutes ago and went to the bathroom, the probability of getting out of bed this time is low.

The response execution unit 114 changes the settings of each device according to the result of the prediction unit 112 . For example, the response execution unit 114 may change the notification condition to a lower risk side even for a user who sets the usage sensor when getting up, if it is within X minutes since the user left the bed last time. Specifically, in the case of the bed-leaving sensor, the easiness of sensor ringing (the speed at which the sensor rings from the state in which the user is lying on the bed to the time the user leaves the bed) is determined by (1) getting up, (2) ) sitting on the edge, and (3) leaving the bed. At this time, changing the notification condition to a higher risk side means that, for example, if the current setting is (3) getting out of bed, the sensor will sound more easily (1) when getting up. Changing the notification condition to a lower risk side means that, for example, if the current setting is (1) getting up, the sensor will sound less likely and (3) it will sound when getting out of bed.

[1.3.11 Eleventh Operation Example]
An eleventh operation example will be described.

The control unit 100 (surrounding environment acquisition unit 102, posture acquisition unit 104, motion acquisition unit 106) acquires whether the user has taken medicine. In addition, the control unit 100 (disease information acquisition unit 110) acquires the type of medicine that the user is taking.

When the user takes medicine, the prediction unit 112 predicts that the risk of tumbling is high if the type of medicine that the user takes is likely to cause dizziness.

Correspondence execution unit 114 changes the settings for each device according to the result of prediction unit 112 . For example, since the risk of falling falls increases within X hours after taking the medicine, the response execution unit 114 changes the bed sensor to the higher risk side within that time. In addition, the response execution unit 114 may also make an announcement to the effect that "Please be careful".

[1.3.12 Twelfth operation example]
A twelfth operation example will be described.

The bed apparatus 5 is provided with a lamp indicating the setting state of the sensor. Then, the controller 100 turns on the lamp of the sensor according to the state of the sensor device 30 .

For example, the control unit 100 lights the lamp in green when the body movement of the user can be detected by the sensor device 30, and lights the lamp in red when it cannot be detected.

The response execution unit 114 analyzes the image data and notifies that the setting is incorrect when the color of the lamp is red. Also, the response execution unit 114 may automatically correct the settings of the sensor device 30 .

[1.3.13 13th operation example]
A thirteenth operation example will be described.

The prediction unit 112 determines whether the risk of falling is high based on the following feature amounts.

・Second feature amount: Of the user's posture output from the posture acquisition unit 104, feature points such as the trajectory of the center of gravity, the deviation of the shoulder position from the direction of movement, the height of the head from the floor, and the position of the feet relative position to each other.

- First feature amount: the position of the equipment output from the surrounding environment acquisition unit 102 . Specifically, the surrounding environment acquisition unit 102 detects equipment by matching with the learned equipment image data. If the user touches the equipment, it is detected as being in use. Multiply the importance coefficient for each fixture.

・Third feature quantity: the trajectory of the center of gravity of the user output from the motion acquisition unit 106, the deviation of the shoulder position from the direction of movement, the height of the head from the floor, and the relative positions of the feature points as the positions of the feet

From these feature amounts, one or more feature amounts are selected, and the prediction unit 112 calculates explanatory variables. Then, the prediction unit 112 predicts the user's fall risk based on the explanatory variables.

The response execution unit 114 executes notification processing when the prediction unit 112 predicts that the risk of falling is high.

[1.3.14 Fourteenth Operation Example]
A fourteenth operation example will be described.

The prediction unit 112 determines whether the risk of falling is high based on the following feature amounts.

・First feature value: The surrounding environment acquisition unit 102 calculates the height of the user and estimates the deviation from the appropriate bed height corresponding to the height ・Fifth feature value: The disease information acquisition unit 110, The user's disease information, treatment details (treatment, medication, use of assistive devices), and assessment are acquired, and each item is multiplied by a coefficient.

From these feature amounts, one or more feature amounts are selected, and the prediction unit 112 calculates explanatory variables. Then, the prediction unit 112 predicts the user's fall risk based on the explanatory variables.

The response execution unit 114 executes notification processing when the prediction unit 112 predicts that the risk of falling is high.

[1.3.15 Fifteenth Operation Example]
A fifteenth operation example will be described.

The prediction unit 112 determines whether the risk of falling is high based on the following feature amounts.

First feature amount: the height, back angle, and knee angle of the bed device 5 acquired by the surrounding environment acquisition unit 102 from the bed control device 40 . Also, the height of the user is calculated by the surrounding environment acquisition unit 102, and the deviation from the appropriate bed height corresponding to the height is calculated. Also, the surrounding environment acquisition unit 102 acquires the amount of light from the image data or the peripheral device 60 (illuminance sensor). Multiply the factor by the brightness. Also, the surrounding environment acquisition unit 102 acquires fixtures/obstacles from the image data. For example, assist bars, shelves, futons, mattresses, partitions, cabinets, curtains, wheelchairs, IV poles, mat sensors, shoes, and other users are captured in shape (edge) and color. Then, the surrounding environment acquiring unit 102 multiplies a coefficient for each of the acquired fixtures/obstacles.

・Fifth feature value: User's disease information, treatment details (treatment, medication, use of assistive device), and assessment acquired by the disease information acquisition unit 110, and each item is multiplied by a coefficient.

From these feature amounts, one or more feature amounts are selected, and the prediction unit 112 calculates explanatory variables. Then, the prediction unit 112 predicts the user's fall risk based on the explanatory variables.

The response execution unit 114 executes notification processing when the prediction unit 112 predicts that the risk of falling is high.

[1.3.16 Sixteenth Operation Example]
The sixteenth operation example explains that the correspondence execution unit 114 can further perform the following operation control by communicating with the peripheral device 60 .

(1) Lift Device When the peripheral device 60 is built in the lift device, the controller 100 can acquire the following information and control the peripheral device 60 (lift device).

• Guiding the user through the correct procedure for applying the sling.
・When the staff unloads the user from the bed device onto the wheelchair, it is determined whether the unloading position is appropriate based on the positional relationship between the seat surface of the wheelchair and the user's buttocks. If it is not in the correct position, announce how far it should be moved.
• Determine if the selected sling is appropriate for the user.

・Judge whether the position of the wheelchair placed next to the bed is appropriate from the operable range of the lift and the lifting position of the user. If the wheelchair or lift needs to be moved, announce how far and in which direction it should be moved.
・Analyze the lift usage status from the image data and automatically operate the bed device and the lift device. For example, when the sling is finished, the bed device and the lift device are automatically moved. Then move the arm of the lift to the position where the wheelchair is.
・Acquire whether the posture of the user suspended on the lift is appropriate. If the user is in a position where pressure is applied locally, an announcement is made as to how to put a cushion in that area.

(2) Wheelchair Whether or not the wheelchair is properly locked will be judged from the hand movements of the staff. Also, if the wheelchair is not locked, an announcement is made to the staff to lock the wheelchair.

(3) Medication Whether or not the user is taking medicine at a fixed time is determined by analyzing the camera image and the imaging time. For example, if the user does not take medicine or throws it away, an alarm is output.

(4) Sensor device Acquires the time when the user wakes up and goes to bed, and offers greetings such as good night and good morning.

(5) Curtain The curtain is automatically opened by detecting from the patient's movement that the user is about to leave the room or has left the bed. It also automatically closes the curtains when it detects that the person has entered the living room or lies on his or her back.

(6) Windows Determine the brightness or the time of day and open the blinds if it is morning to wake up the user. Also, the brightness or the time of day is determined, and if it is night, the blinds are closed to encourage the user to sleep.

(7) Cabinet When a drink is taken from a plastic bottle or the like, it detects that the drink has run out and notifies the staff.

(8) Overbed table Determines that the meal is over and notifies the user to put down the tableware. In addition, it acquires that the bed (mattress, etc.) is dirty during meals and notifies the staff.

(9) IV pole Detects tangled IV tubes and notifies staff, etc. in case of danger. In addition, when the remaining amount of medicine for the infusion is decreasing, the scheduled end time is calculated and notified to the staff and the like.

[2. Second Embodiment]
A second embodiment will be described. 1st Embodiment is embodiment which performs various processes in the control apparatus 10. FIG. The second embodiment is an embodiment in which various processes are executed by a server device. In this embodiment, only parts different from the first embodiment will be explained, and the same reference numerals will be given to the configurations having common functions, and the explanation thereof will be omitted.

FIG. 16 is a partial diagram showing an overview of the system 2 of the second embodiment. In the second embodiment, the control device 15 is connected to the network NW. Various devices other than the camera device 20 are connected to the control device 15 . The server device 70 also executes a fall prediction process and the like. The server device 70 is connected to the network NW.

FIG. 17 is a diagram for explaining functions of the second embodiment. The processing performed by the control device 10 of the first embodiment is implemented by the server device 70 .

That is, the living room R has a control device 200 having a control unit, and each device is connected to the control device 200 . Note that the display device 210 is the display unit 130 of the first embodiment, the operation device 220 is the operation unit 135 of the first embodiment, the notification device 230 is the notification unit 140 of the first embodiment, and the communication device 240 is , and the communication unit 150 of the first embodiment. Each device may be formed integrally with the control device 200, or may be configured as a separate device.

In the second embodiment, data acquired by various devices are transmitted to the server device 70 . For example, image data captured by the camera device 20 is transmitted from the control device 200 to the server device 70 . Server device 70 stores the received image data in image data storage area 1240 . Similarly, a signal or the like indicating body movement acquired by the sensor device 30 is also transmitted to the server device 70 .

Then, in the server device 70, the prediction unit 112 predicts the probability (risk) of the risk of falling based on the received various signals and information. Then, the response execution unit 114 executes notification processing and the like on the notification device 230 .

As described above, according to the present embodiment, the functions described in the first embodiment can be realized by a server device or a cloud. In the present embodiment, all the functions realized in the first embodiment are described as being realized by the server device 70, but only some of the functions realized by the control device 10 of the first embodiment are It may be implemented by the server device 70 . For example, the biological information acquisition unit 108 and the disease information acquisition unit 110 may be realized by the control device 200 and other functions may be realized by the server device 70 .

[3. Modification]
Although the embodiments of the present invention have been described in detail above with reference to the drawings, the specific configuration is not limited to these embodiments, and designs and the like within the scope of the scope of the claims can be applied without departing from the gist of the present invention. include.

In addition, the program that operates in each device in the embodiment is a program that controls the CPU and the like (a program that causes the computer to function) so as to implement the functions of the above-described embodiments. The information handled by these devices is temporarily stored in a temporary storage device (for example, RAM) during processing, then stored in various ROM, HDD, and SSD storage devices, and read out by the CPU as necessary. , correction and writing are performed.

When distributed to the market, the program can be stored in a portable recording medium for distribution, or transferred to a server computer connected via a network such as the Internet. In this case, of course, the storage device of the server computer is also included in the present invention.

Also, the above-described embodiments can be realized by, for example, smartphones and tablets. For example, a staff member or the like installs a smartphone on the footboard of the bed apparatus 5 and takes an image of the bed apparatus 5 with a camera built into the smartphone. Further, information is acquired by connecting the sensor device 30, the bed control device 40, and the like by short-range wireless communication.

By installing an application capable of realizing the functions realized by the control unit 100 on the smartphone, the system of the above-described embodiment can be realized on the smartphone.

1 system 10 control device 100 control unit 120 storage unit 20 camera device 30 sensor device 40 bed control device 50 server device 502 electronic chart DB
60 Peripherals

Claims (8)

A system including a bed device provided with a camera device,
an acquisition unit that acquires information about the user's posture, information about the user's motion, information about the user 's floor location, and information about the environment around the user based on the image data captured by the camera device; and,
a prediction unit that predicts the risk of falling of the user based on the information acquired by the acquisition unit;
a notification unit that performs notification according to the prediction result of the prediction unit;
system with. The acquisition unit further acquires information on the wakefulness state and sleep state of the user,
Information on the environment around the user includes information on the state of the bed apparatus and information on obstacles as mitigation factors for reducing the risk of falling of the user.
The system of claim 1. the state information of the bed apparatus includes the operating state of the bed apparatus or the height of the bottom portion of the bed apparatus;
the information about the obstacle includes the type, motion, or positional state of the obstacle;
3. The system of claim 2, wherein the mitigation factors include whether or not equipment that reduces the risk of falls of the user is recognizable and whether or not staff are present in the vicinity of the user. When the acquisition unit recognizes that a staff member is present around the user as the mitigation factor, the prediction unit determines the fall risk of the user as the mitigation factor when there is no staff member around the user. 4. A system according to claim 2 or 3, which predicts low compared to . Control to change the height or bottom of the bed apparatus, change the brightness of the room, or change the setting of the sensor when the prediction unit predicts that the risk of falling of the user is higher than a threshold. further comprising a control unit for
5. A system according to any one of claims 1-4. When the predicting unit predicts that the user's fall risk is higher than a threshold, a device control unit provided for the obstacle controls the operation of the obstacle.
6. A system according to any one of claims 1-5. further comprising a bed exit detection device that detects the bed exit of the user from the bed apparatus;
The acquisition unit further acquires facial expression information related to the facial expression of the user as second information,
The prediction unit predicts the user's fall risk and the user's physical condition based on the user's facial expression information and the user's motion information,
When it is predicted that the user is in poor physical condition, the setting of the bed-leaving detection device is changed so that the user's bed-leaving is easily detected.
System according to any one of claims 1 to 6. The bed apparatus is provided with a headboard on the head side and a footboard on the foot side,
8. A system according to any preceding claim, wherein the camera device is mounted on at least one of the headboard and/or the footboard.

Images