Bed exit prediction notification device and non-volatile storage medium

Abstract

The purpose of the present invention is to notify a caregiver of a bed leaving, for example, 30 minutes before a caregiver leaves the bed. A bed leaving prediction device (server device) (10) is connected to a portable information processing terminal (40) of a caregiver, an environmental sensor (32-34) for detecting environmental values ​​such as the temperature in a room, a human body sensor (31), and a bed-mounted sensor (35) via an electronic communication line network (60). A bed leaving prediction processing unit (115) calculates a bed leaving prediction value indicating the degree of possibility of a caregiver leaving the bed after a second time from the current time point based on a plurality of environmental values ​​detected during a period between a current time point and a time point before a first time point, the output of the human body sensor, and the output of the bed-mounted sensor. A bed leaving notification processing unit (117) compares the bed leaving prediction value with a threshold value, and when the bed leaving prediction value exceeds the threshold value, sends a bed leaving notification indicating that the caregiver will leave the bed after the second time point has passed to the portable information processing terminal.

Description

Technical Field

[0001] The present invention relates to a bed leaving prediction notification device and a program. Background Art

[0002] In nursing facilities, elderly people and other caregivers often fall and suffer fractures when getting out of bed after waking up. In addition, the risk of falling during night shifts when there are fewer people also puts a high psychological burden on caregivers.

[0003] Currently, many facilities have introduced a system in which sensors such as pressure-sensitive pads are installed in each bed to detect when a care receiver leaves the bed and notify the caregiver's portable terminal of the detection.

[0004] However, the existing system only detects after the fact that the care recipient has left the bed, and the caregiver cannot personally witness the person leaving the bed and provide care for the person. Summary of the invention

[0005] Problems to be solved by the invention

[0006] It is desirable to notify the caregiver of the bed leaving, for example, 30 minutes before the care receiver leaves the bed.

[0007] Methods for solving problems

[0008] The bed leaving prediction server device of this embodiment is connected to an information processing terminal, an environmental sensor, an infrared sensor, and a bed sensor via an electronic communication line network. The information processing terminal is an information processing terminal used by a caregiver who takes care of the care recipient. The environmental sensor detects at least one environmental value of the temperature, humidity, and illumination in the room where the care recipient lives. The infrared sensor is configured to include the bed in the room within the sensing range. The bed sensor detects the movement of the care recipient on the bed. The calculation unit repeatedly calculates the movement index corresponding to the movement of the care recipient on the bed based on the output of the infrared sensor. The calculation unit repeatedly calculates the pulse of the care recipient, the posture index for distinguishing the sleeping posture, the sitting posture, and the standing posture, the turning over index for distinguishing the left turning over, the left movement, the right movement, the right turning over, and the no turning over, the body movement index for distinguishing the body movement and the quiet state, and the sleep index for distinguishing the sleeping and waking up based on the output of the bed sensor. The calculation unit calculates a bed leaving prediction value indicating the degree of possibility that the care receiver will leave the bed after a second time has passed since the prediction calculation processing time point, based on a plurality of environmental values ​​detected during a period between the prediction calculation processing time point and a time point before the first time point, and a plurality of exercise amount indices, a plurality of pulses, a plurality of posture indices, a plurality of turning indices, a plurality of body movement indices, and a plurality of sleep indices calculated based on the output of the bed sensor during the period. The notification processing unit compares the bed leaving prediction value with a threshold value, and when the bed leaving prediction value exceeds the threshold value, sends a bed leaving notification indicating that the care receiver will leave the bed after the second time has passed to the information processing terminal; the bed sensor has a plurality of air tubes and a plurality of microphones, the internal spaces of the plurality of air tubes are connected to form a single circulation path, the plurality of microphones are discretely arranged inside the plurality of air tubes, and detect sound signals generated by the circulation of air inside the plurality of air tubes, the air tubes are divided by partitions at a plurality of locations, the center of each partition is slightly open, and the microphones are installed at the edges of the openings. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 1 is a diagram showing an overall system including the bed exit prediction notification device according to the present embodiment.

[0010] Figure 2 Is set with Figure 1 Schematic diagram of a human body sensing sensor inside a room.

[0011] Figure 3 It is shown Figure 1 Schematic diagram of the bed-mounted sensor.

[0012] Figure 4 yes Figure 1 Functional structure diagram of the bed exit prediction notification device.

[0013] Figure 5It is shown Figure 1 Flowchart of a bed exit prediction notification processing procedure of a bed exit prediction notification device.

[0014] Figure 6 yes Figure 5 Supplementary figure of the treatment order.

[0015] Figure 7 It is shown Figure 5 FIG. 4 is a diagram showing an example of a display screen of an information processing terminal carried by a nursing staff corresponding to step S8.

[0016] Figure 8 It is shown Figure 5 Flowchart of a modified example of the bed exit prediction notification process. DETAILED DESCRIPTION

[0017] Hereinafter, the bed leaving prediction notification device of the present embodiment will be described with reference to the accompanying drawings. In addition, bed leaving prediction is defined as estimating the possibility that the care receiver will leave the bed in a relatively long time of about 30 minutes, and is clearly distinguished from detecting the care receiver leaving the bed after the fact, and further detecting obvious precursors and signs of leaving the bed such as the care receiver raising the upper body on the bed or sitting at the end of the bed. It is known that if the precursors and signs of leaving the bed are detected, the care receiver will leave the bed in a few seconds or at most a few minutes.

[0018] like Figure 1 As shown, the bed exit prediction notification device (information processing device) 10 as a server device is connected to the electronic communication line network 60. As the electronic communication line network 60, a public communication line network (Internet line network) is typically used, but it is not limited to this. An information processing device 20 for comprehensive management of nursing facilities, a plurality of portable information processing terminals 40 such as smart phones carried by a plurality of nursing staff, etc. are connected to the bed exit prediction notification device 10 via the electronic communication line network 60. An application for cooperating with the bed exit prediction notification device 10 is installed in the portable information processing terminal 40. The sensor group 30 is connected to the bed exit prediction notification device 10 via the electronic communication line network 60 and the gateway device 50. The sensor group 30 and the gateway device 50 are connected by a short-range wireless communication method such as Bluetooth (registered trademark) or by wire.

[0019] As the sensor group 30, a human body sensing sensor 31, a temperature sensor 32, a humidity sensor 33, an illuminance sensor 34, and a bed sensor (bed sensor) 35 are provided. Figure 2As shown in the example, the human body sensor 31, the temperature sensor 32, the humidity sensor 33, and the illuminance sensor 34 are integrated into, for example, an indoor sensor 36. The human body sensor 31 is typically an infrared sensor. The indoor sensor 36 is installed on a wall or the like in a manner that the sensing range (dashed line) of the human body sensor 31 includes a bed (here, a bed) 70 in the room. The output of the human body sensor 31 can be used to calculate an exercise index reflecting the exercise amount of the care recipient on the bed 70.

[0020] The bed sensor 35 is a bed sensor that detects the heartbeat, turning over, and other movements of the care receiver on the bed 70. Figure 3 As shown, for example, four flexible air tubes 351-354 are connected vertically and horizontally in a manner that the internal spaces are connected to form a single circulation path. The air tubes 351-354 are divided by partitions 355-358 at multiple locations. The center of each partition 355-358 is slightly open, and microphones 359-362 are installed at the edge of the opening. With the beating, turning over and other movements of the care recipient, the internal air circulates in the circulation path of the air tubes 351-354. The microphones 359-362 detect the sound signals generated at the openings of the partitions 355-358 accompanying this circulation.

[0021] like Figure 4 As shown, the bed leaving prediction notification device 10 has a control unit 101 responsible for overall control. A storage unit 103, a timing circuit 105, a terminal communication unit 107, a gateway communication unit 109, a human body sensor output processing unit 111, a bed-mounted sensor output processing unit 113, a bed leaving prediction processing unit 115, and a bed leaving notification processing unit 117 are connected to the control unit 101 via a control / data bus 102.

[0022] In addition, as the actual device structure related to the control unit 101, the human body sensor output processing unit 111, the bed sensor output processing unit 113, the bed leaving prediction processing unit 115, and the bed leaving notification processing unit 117, it is composed of a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a RAM (Random Access Memory), and a ROM (Read Only Memory). The storage unit 103 composed of an HDD (Hard Disk Drive) and an SSD (Solid State Drive) pre-stores a bed leaving general program, a human body sensor output processing program, a bed sensor output processing program, a bed leaving prediction program, and a bed leaving notification program. When starting, each program is loaded from the storage unit 103 to the RAM to execute the general processing, the human body sensor output processing, the bed sensor output processing, the bed leaving prediction processing, and the bed leaving notification processing.

[0023] The storage unit 103 stores various data of the sensor group 30, and also stores an association table that associates identification numbers (IDs) of a plurality of portable information processing terminals 40 carried by a plurality of caregivers who are respectively responsible for a plurality of caregivers or rooms with the plurality of caregivers or rooms. In addition, the application ID installed in the portable information processing terminal 40 is associated with the identification number (ID). Information communication is bidirectionally performed between the bed exit prediction notification device 10 and the portable information processing terminal 40 through the application ID.

[0024] Human body sensor output processing unit 111 repeatedly calculates an exercise index reflecting the amount of exercise of the care receiver on bed 70 at a predetermined period based on the per-unit-time integral value of the peak value of the detection signal of human body sensor 31, the area of ​​the signal waveform portion per unit time, and other waveform feature quantities.

[0025] When the person being cared for on the bed 70 turns over, beats, etc., the air moves in the circulation path in the air tubes 351-354 along with the changes in the center of gravity and body pressure distribution. The movement is detected by the microphones 359-362. The bed-mounted sensor output processing unit 113 calculates the following values ​​based on the peak value of the output signal of the microphones 359-362, its time change, mutual time lag and other parameters: the pulse (heart rate) of the person being cared for, the value for distinguishing the posture of sleeping / sitting up / standing up (posture index); the value for distinguishing turning over to the left / moving left / moving right / turning over to the right / no turning over (turning over index); the value for distinguishing the state of body movement / rest (body movement index); and the value for distinguishing the state of sleeping / waking up (sleep index).

[0026] The bed leaving prediction processing unit 115 calculates a bed leaving prediction value of the degree of possibility of the care receiver leaving the bed 70 expressed as a percentage (%) after a second time period, for example, at least 10 minutes, typically 30 minutes, from the current time point based on the following data detected and calculated within a period between the current time point (prediction calculation processing time point) and a time point earlier than the current time point by a first time period, for example, at least 10 minutes, typically 30 minutes, the current time point and a time point earlier than the current time point (30 minutes), the data being:

[0027] During this period, the temperature sensor 32 repeatedly detects a plurality of temperatures.

[0028] During this period, the humidity sensor 33 repeatedly detects a plurality of humidity levels.

[0029] During this period, the multiple illuminances repeatedly detected by the illuminance sensor 34 are

[0030] A plurality of exercise amount indices indicating the amount of exercise related to the care receiver are calculated by the human sensor output processing unit 111 based on a plurality of detection signals repeatedly detected by the human sensor 31 during the period.

[0031] The pulse (heart rate) of the care receiver calculated by the bed-mounted sensor output processing unit 113 based on a plurality of detection signals repeatedly detected by the bed-mounted sensor 35 during this period,

[0032] A plurality of posture indices indicating different postures of the care receiver are calculated by bed-mounted sensor output processing unit 113 based on a plurality of detection signals repeatedly detected by bed-mounted sensor 35 during the period.

[0033] A plurality of turning indicators for distinguishing the turning of the care receiver are calculated by the bed-mounted sensor output processing unit 113 based on a plurality of detection signals repeatedly detected by the bed-mounted sensor 35 during the period.

[0034] A plurality of body movement indices for distinguishing body movement and rest of the care receiver calculated by bed-mounted sensor output processing unit 113 based on a plurality of detection signals repeatedly detected by bed-mounted sensor 35 during the period,

[0035] A plurality of sleep indices for distinguishing between sleep and wakefulness of the care receiver are calculated by the bed-mounted sensor output processing unit 113 based on a plurality of detection signals repeatedly detected by the bed-mounted sensor 35 during the period. The higher the bed leaving prediction value, the higher the possibility of leaving the bed after the second time has passed, and the lower the bed leaving prediction value, the lower the possibility of leaving the bed after the second time has passed.

[0036] In addition, preferably, the characteristic values ​​and basic statistics of the temperature, humidity, illumination, exercise index, pulse index, posture index, turning index, body movement index, and sleep index during the period (30 minutes) are obtained, and the bed leaving prediction value is calculated based on the characteristic values ​​and basic statistics. In addition, as a process for calculating the bed leaving prediction value based on the characteristic values ​​and basic statistics, artificial intelligence (AI) technology is actually used.

[0037] That is, the bed leaving prediction processing is preferably implemented by a neural network. More preferably, the bed leaving prediction processing is implemented by a multi-level neural network (deep learning). A program for learning is stored in the storage unit 103, and the learning program (learning unit) uses the bed leaving index as training data (training data) to enable the neural network to learn together with the information of the care recipient, the environmental values ​​such as temperature, humidity, and illumination detected by the environmental sensor installed in the room where the care recipient lives, the exercise index, the pulse, the posture index, the turning over index, the body movement index, and the sleep index.

[0038] As characteristic values ​​and basic statistics, the average, standard deviation, minimum value, maximum value, and three quartiles (25%, 50%, and 75%) within the period can be exemplified. In fact, in the estimation process of the bed exit prediction value, at least one basic statistic specified for each indicator is applied. As characteristic values ​​and basic statistics actually applied to the estimation process of the bed exit prediction value, it is preferred to apply them in appropriate combination for each care receiver, and further for temperature, humidity, illumination, exercise index, pulse, posture index, turning over index, body movement index, and sleep index according to the characteristics of the care receiver's sleep, waking up, and bed exit.

[0039] The bed leaving prediction processing unit 115 compares the estimated bed leaving prediction value with the threshold value pre-stored in the storage unit 103. When the bed leaving prediction value exceeds the threshold value, the bed leaving notification processing unit 117 transmits a bed leaving notification indicating that the care receiver will leave the bed 70 after a second time period, typically 30 minutes, which is the same as the first time period, has passed since the current time point (the prediction calculation processing time point) to the portable information processing terminal 40 via the terminal communication unit 107. When the bed leaving prediction value is below the threshold value, the bed leaving notification is not transmitted to the portable information processing terminal 40.

[0040] The portable information processing terminal 40 that has received the bed leaving notification displays a message that the care receiver will leave the bed 70 in 30 minutes, and at the same time displays three buttons corresponding to the command sent (fed back) to the bed leaving prediction notification device 10 in a manner that distinguishes the following three situations: the situation where the caregiver who has arrived at the room has already provided care for the care receiver to leave the bed (first situation), the situation where the caregiver has waited for a while after arriving at the room but the care receiver has not left the bed (second situation), or the situation where the care receiver has left the bed before the caregiver arrives at the room (third situation). When the caregiver clicks any of the three buttons, information indicating any of the first to third situations is sent from the portable information processing terminal 40 to the bed leaving prediction notification device 10.

[0041] When the control unit 101 of the bed exit prediction notification device 10 receives the first condition (bed exit care has been performed) as the bed exit prediction result from the portable information processing terminal 40, the threshold value stored in the storage unit 103 is maintained at the current value. On the other hand, when the control unit 101 receives the second condition (not leaving the bed) as the bed exit prediction result from the portable information processing terminal 40, the control unit 101 controls the storage unit 103 to update the threshold value stored in the storage unit 103 to a higher value obtained by adding a predetermined value, such as 5%, to the current value. When the control unit 101 receives the third condition (already leaving the bed) as the bed exit prediction result from the portable information processing terminal 40, the control unit 101 controls the bed exit prediction processing unit 115 to update the threshold value applied to the bed exit prediction processing to a lower value obtained by subtracting a predetermined value, such as 5%, from the current value.

[0042] In addition, the initial value of the above-mentioned threshold, the addition value added to the threshold in the second situation, the subtraction value subtracted from the threshold in the third situation, the first time and the second time used in the bed exit estimation process can be set individually for each care receiver by the facility manager via the information processing device 20 for comprehensive management.

[0043] Figure 5 The following is a description of the bed exit prediction notification processing procedure according to the present embodiment. Figure 6 yes Figure 5 Supplementary diagram of the processing sequence. First, in step S1, the bed exit prediction notification processing is started. The processing can be started for each room individually or for multiple rooms in the facility. In addition, the processing can be started manually or automatically according to a set time. With the start of the bed exit prediction notification processing as an opportunity, the detection data of the human body sensing sensor 31, the temperature sensor 32, the humidity sensor 33, the illuminance sensor 34, and the bed-mounted sensor 35 are received by the bed exit prediction notification device 10 via the gateway device 50, and the room identification number and the detection time are associated and stored in the storage unit 103 (S2).

[0044] Based on the detection data of the human body sensor 31, the human body sensor output processing unit 111 calculates the exercise index reflecting the exercise amount of the care receiver on the bed 70, and stores it in the storage unit 103 (S3). In addition, based on the detection data of the microphones 359-362 of the bed-mounted sensor 35, the bed-mounted sensor output processing unit 113 calculates the pulse, posture index, turning index, body movement index, and sleep index of the care receiver, and stores them in the storage unit 103 (S4).

[0045] Next, based on the multiple temperatures, humidity, illumination, exercise amounts, pulses, posture indices, turning indices, body movement indices, and sleep indices generated during the period between the current time point (prediction calculation processing time point) and the time point 30 minutes earlier, the bed leaving prediction processing unit 115 calculates a bed leaving prediction value (%) indicating the possibility that the care receiver will leave the bed 70 at a time point 30 minutes after the current time point (S5). In the bed leaving prediction processing unit 115, the bed leaving prediction value is compared with a threshold value pre-set for the care receiver (S6). In addition, the bed leaving prediction value is repeatedly calculated at a predetermined cycle, for example, a cycle of 5 minutes, and compared with the threshold value.

[0046] When the bed leaving prediction value is below the threshold, that is, when it is determined that the possibility of the care receiver leaving the bed 70 after 30 minutes is low (S6: No), the process returns to step S2. When the bed leaving prediction value exceeds the threshold, that is, when it is determined that the possibility of the care receiver leaving the bed 70 after 30 minutes is high (S6: Yes), a notification of "will leave the bed after 30 minutes" is sent from the bed leaving notification processing unit 117 to the portable information processing terminal 40 of the caregiver in charge of the care receiver through the terminal communication unit 107 together with the room number and name information of the care receiver (S7). When the bed leaving prediction value is below the threshold, it is preferable that the notification of "will leave the bed after 30 minutes" and the notification of "will leave the bed after 30 minutes is low" are not sent to the portable information processing terminal 40.

[0047] like Figure 7As shown in the example, a message 41 such as "The care receiver will leave the bed in 30 minutes" is displayed on the display screen of the portable information processing terminal 40 together with the room number and name for identifying the care receiver. In the screen, three buttons 42, 43, and 44 are displayed for notifying the bed leaving prediction notification device 10 of any of the following three situations as the bed leaving prediction result: the situation that the caregiver who arrived at the room has already provided care for the care receiver to leave the bed (first situation), the situation that the caregiver waited for a while after arriving at the room but the care receiver did not leave the bed (second situation), and the situation that the care receiver had already left the bed before the caregiver arrived at the room (third situation). When the caregiver clicks any of the three buttons 42, 43, and 44, a notification indicating any of the first to third situations is sent from the portable information processing terminal 40 to the bed leaving prediction notification device 10 as the bed leaving prediction result (S8). In addition, the bed exit prediction results are input together with temperature, humidity, illumination, exercise index, pulse index, posture index, turning index, body movement index, and sleep index (or each characteristic value, basic statistics) through the input layer of artificial intelligence (AI) to improve the estimation accuracy.

[0048] The control unit 101 of the bed leaving prediction notification device 10 determines the bed leaving prediction result based on the reception from the portable information processing terminal 40 (S9). When the bed leaving prediction result is the first condition (bed leaving care has been performed), the bed leaving prediction process and the bed leaving notification process are terminated based on the judgment that the bed leaving prediction is correct (S10). When the bed leaving prediction result is the second condition (not leaving the bed), the current threshold associated with the care receiver is updated to a high value obtained by adding a prescribed value, such as 5%, to the threshold based on the judgment that the bed leaving prediction is incorrect and the bed leaving prediction is too early (S11), and the process returns to step S2 to continue the bed leaving prediction process and the notification process. By increasing the threshold, the accuracy of the next bed leaving prediction can be improved. When the bed leaving prediction result is the third condition (already leaving the bed), the current threshold associated with the care receiver is updated to a low value obtained by subtracting a prescribed value, such as 5%, from the threshold based on the judgment that the bed leaving prediction is incorrect and the bed leaving prediction is too late (S12), and the bed leaving prediction process and the bed leaving notification process (S10) are terminated. By lowering the threshold, the next bed leaving prediction accuracy can be improved. By repeatedly performing bed leaving prediction processing and notification processing over several days or weeks, and then performing threshold updating processing based on the bed leaving prediction results, the bed leaving prediction accuracy can be further improved.

[0049] As described above, based on a variety of information such as the room's environmental information (temperature, humidity, illumination), the care receiver's physiological information (pulse), the care receiver's activity information in bed (exercise index, posture index, turning over index, body movement index), and the care receiver's sleep information (sleep index) over a relatively long period of at least 10 minutes, typically 30 minutes, it is possible to highly accurately predict whether the care receiver will leave the bed after a relatively long period of at least 10 minutes, typically 30 minutes. This allows the caregiver to visit and provide assistance to the care receiver before the care receiver leaves the bed.

[0050] In the above description, when the bed leaving prediction result is the second condition (not leaving the bed) or the third condition (already leaving the bed), the threshold is updated, but the threshold may be maintained without updating the threshold and the message content of the bed leaving notification may be changed. Figure 8 As shown in FIG. 1 , when the bed leaving prediction result is the second condition (not leaving the bed), the elapsed time (the current time point is 30 minutes) recorded in the message content of the bed leaving notification at the current time point "will leave the bed in 30 minutes" is added to the prescribed time, typically 5 minutes, to update the message content of the bed leaving notification to "will leave the bed in 35 minutes". When the bed leaving prediction result is the third condition (already leaving the bed), the elapsed time (the current time point is 30 minutes) recorded in the message content of the bed leaving notification at the current time point "will leave the bed in 30 minutes" is subtracted from the prescribed time, typically 5 minutes, to update the message content of the bed leaving notification to "will leave the bed in 25 minutes". By repeatedly performing the bed leaving prediction process and the notification process over several days or weeks, and then performing the threshold value update process according to the bed leaving prediction result, the bed leaving timing notified can be matched with the actual bed leaving timing.

[0051] Although several embodiments of the present invention have been described, these embodiments are given as examples only and are not intended to limit the scope of the invention. These embodiments can be implemented in various other ways, and various omissions, substitutions, and changes can be made without departing from the scope of the invention. These embodiments or their variations are included in the scope or spirit of the invention, and are included in the invention recorded in the claims and their equivalents.

[0052] Description of Reference Numerals

[0053] 10: Bed exit prediction notification device, 20: Comprehensive management information processing device, 30: Sensor group, 31: Human body sensing sensor, 32: Temperature sensor, 33: Humidity sensor, 34: Illuminance sensor, 35: Bed-mounted sensor, 40: Portable information processing terminal, 101: Control unit, 102: Control / data bus, 103: Storage unit, 105: Timing circuit, 107: Terminal communication unit, 109: Gateway communication unit, 111: Human body sensing sensor output processing unit, 113: Bed-mounted sensor output processing unit, 115: Bed exit prediction processing unit, 117: Bed exit notification processing unit.

Claims

1. A bed leaving prediction server device, connected to an information processing terminal, an environmental sensor, an infrared sensor and a bed sensor via an electronic communication line network, wherein the information processing terminal is an information processing terminal used by a caregiver who cares for a care recipient, the environmental sensor detects at least one environmental value of the temperature, humidity and illumination in a room where the care recipient lives, the infrared sensor is configured to include a bed in the room within a sensing range, and the bed sensor detects the movement of the care recipient on the bed, wherein: The bed leaving prediction server device comprises: a calculation unit that repeatedly calculates an exercise index corresponding to an exercise amount of the care receiver on the bed based on an output of the infrared sensor, and repeatedly calculates a pulse of the care receiver, a posture index for distinguishing between sleeping posture, sitting posture, and standing posture, a turning over index for distinguishing between turning over to the left, moving to the left, moving to the right, turning over to the right, and no turning over, a body movement index for distinguishing between body movement and resting state, and a sleep index for distinguishing between sleeping and waking up based on an output of the bed sensor; a calculation unit for calculating a bed leaving prediction value indicating a degree of possibility that the care receiver will leave the bed after a second time has passed from the prediction calculation processing time point, based on the plurality of environmental values ​​detected during a period between a prediction calculation processing time point and a time point before a first time point, and the plurality of exercise amount indices, the plurality of pulses, the plurality of posture indices, the plurality of turning over indices, the plurality of body movement indices, and the plurality of sleep indices calculated based on an output of the bed sensor during the period; as well as a notification processing unit that, when the bed leaving prediction value exceeds a threshold value, transmits a bed leaving notification indicating that the care receiver will leave the bed after the second time has passed to the information processing terminal; The bed sensor has multiple air tubes and multiple microphones. The internal spaces of the multiple air tubes are connected to form a single circulation path. The multiple microphones are discretely arranged inside the multiple air tubes to detect sound signals generated by the circulation of internal air in the multiple air tubes. The air tubes are separated by partitions at multiple locations. The center of each partition is slightly open, and the microphone is installed at the edge of the opening.

2. The bed exit prediction server device according to claim 1, wherein: It also has a receiving portion, the receiving unit receives, from the information processing terminal, either information indicating a first condition or information indicating a second condition as a bed leaving prediction result, the first condition being a condition in which the caregiver who has received the bed leaving notification has already provided care for the care receiver to leave the bed, and the second condition being a condition in which the care receiver has not left the bed, When the information indicating the first condition is received from the information processing terminal as the bed leaving prediction result, the threshold is maintained, and when the information indicating the second condition is received as the bed leaving prediction result, the threshold is updated to a new threshold obtained by adding a predetermined value to the threshold.

3. The bed exit prediction server device according to claim 1, wherein: It also has a receiving portion, the receiving unit receives, from the information processing terminal, either information indicating a first condition or information indicating a second condition as a bed leaving prediction result, the first condition being a condition in which the caregiver who has received the bed leaving notification has already provided care for the care receiver to leave the bed, and the second condition being a condition in which the care receiver has not left the bed, When information indicating the first condition is received from the information processing terminal as a bed leaving prediction result, a bed leaving notification indicating that the care receiver will leave the bed after the second time has passed is maintained, and when information indicating the second condition is received as a bed leaving prediction result, the bed leaving notification is updated to indicate that the care receiver will leave the bed after a third time longer than the second time has passed.

4. The bed exit prediction server device according to claim 1, wherein: It also has a receiving portion, The receiving unit receives, from the information processing terminal, any one of information indicating a first condition, information indicating a second condition, and information indicating a third condition as a bed leaving prediction result, wherein the first condition is a condition in which the caregiver who has received the bed leaving notification has already provided care for the care receiver to leave the bed, the second condition is a condition in which the care receiver has not left the bed, and the third condition is a condition in which the care receiver has left the bed before the caregiver arrives at the room, When information indicating the first condition is received from the information processing terminal as a bed leaving prediction result, the threshold is maintained; when information indicating the second condition is received as a bed leaving prediction result, the threshold is replaced with a new threshold obtained by adding a prescribed value to the threshold; and when information indicating the third condition is received as a bed leaving prediction result, the threshold is replaced with a new threshold obtained by subtracting the prescribed value from the threshold.

5. The bed exit prediction server device according to claim 1, wherein: It also has a receiving portion, The receiving unit receives, from the information processing terminal, any one of information indicating a first condition, information indicating a second condition, and information indicating a third condition as a bed leaving prediction result, wherein the first condition is a condition in which the caregiver who has received the bed leaving notification has already provided care for the care receiver to leave the bed, the second condition is a condition in which the care receiver has not left the bed, and the third condition is a condition in which the care receiver has left the bed before the caregiver arrives at the room, When information indicating the first condition is received from the information processing terminal as a bed leaving prediction result, a bed leaving notification indicating that the care receiver will leave the bed after the second time has passed is maintained; when information indicating the second condition is received as a bed leaving prediction result, the bed leaving notification is updated to indicate that the care receiver will leave the bed after a third time longer than the second time has passed; and when information indicating the third condition is received as a bed leaving prediction result, the bed leaving notification is updated to indicate that the care receiver will leave the bed after a fourth time shorter than the second time has passed.

6. The bed exit prediction server device according to claim 1, wherein: The threshold value is individually associated with respect to the care receiver.

7. A computer-readable non-volatile storage medium, wherein: The computer stores programs for implementing the following functions: The device is connected to an information processing terminal, an environmental sensor, an infrared sensor, and a bed sensor via an electronic communication line network, wherein the information processing terminal is used by a caregiver who cares for a care recipient, the environmental sensor detects at least one environmental value of the temperature, humidity, and illumination in the room where the care recipient lives, the infrared sensor is configured to include the bed in the room within a sensing range, the bed sensor detects the movement of the care recipient on the bed, the bed sensor has a plurality of air tubes and a plurality of microphones, the internal spaces of the plurality of air tubes are connected to form a single circulation path, the plurality of microphones are discretely arranged inside the plurality of air tubes, and detect sound signals generated by the circulation of the internal air of the plurality of air tubes, the air tubes are separated by partitions at a plurality of locations, the center of each partition is slightly open, and the microphones are installed at the edges of the openings; Based on the output of the infrared sensor, an exercise index corresponding to the exercise amount of the care receiver on the bed is repeatedly calculated; based on the output of the bed sensor, a pulse (heart rate) of the care receiver, a posture index for distinguishing between sleeping posture, sitting posture, and standing posture, a turning over index for distinguishing between turning over to the left, moving to the left, moving to the right, turning over to the right, and no turning over, a body movement index for distinguishing between body movement and resting state, and a sleep index for distinguishing between sleeping and waking up are repeatedly calculated; Calculating a bed leaving prediction value indicating the degree of possibility that the care receiver will leave the bed after a second time has passed from the prediction calculation processing time point, based on the plurality of environmental values ​​detected during a period between a prediction calculation processing time point and a time point before a first time point, and the plurality of exercise amount indices, the plurality of pulses (heart rate), the plurality of posture indices, the plurality of turning over indices, the plurality of body movement indices, and the plurality of sleep indices calculated based on the output of the bed sensor during the period; as well as The bed leaving prediction value is compared with a threshold value, and when the bed leaving prediction value exceeds the threshold value, a bed leaving notification indicating that the care receiver will leave the bed after the second time period has elapsed is transmitted to the information processing terminal.

8. A device for predicting bed exit, wherein: The bed leaving prediction device implements bed leaving prediction processing based on a neural network, and the bed leaving prediction device has: a calculation unit for calculating the exercise amount index, calculating the exercise amount index corresponding to the exercise amount of the care receiver on the bed based on the output of the infrared sensor arranged to include the bed in the room within the sensing range; A bed sensor is provided on the bed, and comprises a plurality of air tubes and a plurality of microphones, wherein the internal spaces of the plurality of air tubes are connected to form a single circulation path, and the plurality of microphones are discretely provided inside the plurality of air tubes to detect sound signals generated by the circulation of the internal air of the plurality of air tubes, wherein the air tubes are divided by partitions at a plurality of locations, and the center of each partition is slightly opened, and the microphones are installed at the edges of the openings; a pulse and index calculation unit, which calculates the pulse of the care receiver, a posture index for distinguishing between sleeping posture, sitting posture, and standing posture, a turning over index for distinguishing between turning over to the left, moving to the left, moving to the right, turning over to the right, and no turning over, a body movement index for distinguishing between body movement and resting state, a sleep index for distinguishing between sleeping and waking up, and a bed leaving index for distinguishing between lying in bed and leaving the bed, based on the output of the bed sensor; a learning unit, using the bed leaving index together with information identifying the care receiver, at least one environmental value of temperature, humidity, and illumination detected by an environmental sensor installed in the room where the care receiver lives, the exercise amount index, the pulse, the posture index, the turning over index, the body movement index, and the sleep index as training data to enable the neural network to learn; as well as The bed leaving prediction unit inputs the environmental value at the current moment, the pulse of the care receiver, the posture index, the turning index, the body movement index, and the sleep index into the neural network that has performed the learning, and outputs a bed leaving prediction value indicating the possibility of the care receiver leaving the bed after a specified time has passed from the current moment.

Images