Wearable terminal, notification method, and program

Abstract

The purpose of the present disclosure is to notify a user regarding promotion of an action at a timing when a Bp value is as large as possible.　The present disclosure is a wearable terminal that notifies a user regarding a message for promoting an action, the wearable terminal comprising: a sensor unit that detects an operation performed by a user and obtains sensor information; an estimation unit that, on the basis of the sensor information, estimates a prescribed value, which is the magnitude of the granularity of a present timing that is not annoying for the user; a prediction unit that predicts the frequency of occurrence of the prescribed value from occurrence distribution data pertaining to the prescribed value based on schedule information pertaining to the user, thereby calculating a prescribed threshold value; and a notification determination unit that, when the prescribed value is equal to or greater than the prescribed threshold value, determines the timing at which the user is notified regarding the message.

Description

Wearable device, notification method, and program

[0001] The present disclosure relates to a technique for reducing the annoyance of reminder notifications intended to encourage a user to take action.

[0002] There are many technologies that encourage healthy behavior by notifying users from the system side. However, most notification methods that encourage such behavior distract users and cause them to feel annoyed. Therefore, many technologies have been developed to reduce the annoyance of system-based actions (see Patent Literature 1).

[0003] Furthermore, notifications to users are expected to be sent within a certain pre-scheduled time range. One of the many techniques for reducing annoyance is to communicate at a non-annoying timing (breakpoint) within the notification's scheduled time range (see Non-Patent Document 1). Furthermore, the breakpoints detected by this technique also have a granularity (see Non-Patent Document 2). If the granularity is the Bp value, the larger the Bp value, the less annoyed the notification will be. In this way, the system can find a timing with a larger Bp value within the scheduled time for the notification and send the notification, which is believed to be more effective in reducing the annoyance of the notification.

[0004] JP 2019-185389 A

[0005] Newtson, D. and Engquist, G.: The perceptual organization of ongoing behavior, Journal of Experimental Social Psychology, vol. 12, no. 5, pp. 436-450 (1976). Iqbal, ST and Bailey, BP: Understanding and developing models for detecting and differentiating breakpoints during interactive tasks, Proc. SIGCHI Conference on Human Factors in Computing Systems (CHI '07), pp. 697-706 (2007).

[0006] However, with the conventional method, the system does not necessarily notify the user of the promotion of the action at the timing when the Bp value is large.

[0007] The present invention has been made in consideration of the above points, and has an object to notify a user of promotion of an action at a timing when the Bp value is as large as possible.

[0008] In order to solve the above problem, the disclosure of claim 1 is a wearable device that notifies a user of a message to encourage behavior, the wearable device having: a sensor unit that detects the user's movements and obtains sensor information; an estimation unit that estimates a predetermined value that is the granularity of timing that is currently not bothersome to the user based on the sensor information; a prediction unit that calculates a predetermined threshold by predicting the occurrence frequency of the predetermined value from occurrence distribution data of the predetermined value based on the user's schedule information; and a notification determination unit that determines the timing to notify the message when the predetermined value is equal to or greater than the predetermined threshold.

[0009] As described above, the present invention has the effect of being able to notify the user of the promotion of action at a timing when the Bp value is larger than in the past.

[0010] FIG. 1 is an overall configuration diagram of a communication system according to the present embodiment. FIG. 2 is an electrical hardware configuration diagram of a wearable terminal. FIG. 3 is an electrical hardware configuration diagram of an electronic terminal. FIG. 4 is a functional configuration diagram of a communication system according to a first embodiment. FIG. 5 is a flowchart showing processing or operations in a learning phase according to the first embodiment. As an example, a schematic diagram showing processing for creating annotation data of a Bp value based on video information capturing the user's movements at the time when a wearable terminal is worn on a user's arm. FIG. 6 is a flowchart showing processing or operations in an inference phase according to the first embodiment. FIG. 7 is a flowchart showing processing or operations in an inference phase according to the first embodiment. FIG. 8 is a flowchart showing Bp value estimation processing. As an example, a conceptual diagram showing processing for estimating a Bp value based on the user's movements at the time when a wearable terminal 10a is worn on a user's arm. As an example, a diagram showing an overview of an example in which a notification prompting action is sent between 7:00 and 7:10 and the width of the time step Ts is set to 20 seconds. FIG. 9 is a functional configuration diagram of a communication system according to the first embodiment. FIG. 10 is a flowchart showing processing or operations in a learning phase according to a second embodiment. FIG. 11 is a flowchart showing processing or operations in an inference phase according to the second embodiment. 10 is a flowchart showing processing or operations in the inference phase according to a second embodiment. FIG. 11 is a flowchart showing Bp value data selection processing in the inference phase according to a second embodiment. FIG. 12 is a diagram showing an overview of an example in which a prompt to act notification is sent between 7:00 and 7:10, and the width of the time step Ts is set to 20 seconds according to the second embodiment.

[0011] Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

[0012] [System Configuration of the Embodiment] First, an outline of the configuration of a communication system of the present embodiment will be described using Fig. 1. Fig. 1 is a diagram showing the overall configuration of a communication system according to the present embodiment. As shown in Fig. 1, the communication system 1 of the present embodiment is constructed by a wearable terminal 10 and an electronic terminal 30.

[0013] The wearable terminal 10 is a smart watch or the like worn by a user. The electronic terminal 30a is a smartphone, a tablet terminal, a notebook PC (Personal Computer), or the like. The wearable terminal 10 and the electronic terminal 30a can communicate with each other via short-range wireless communication. The short-range wireless communication is communication performed using technologies such as Wi-Fi, Bluetooth (registered trademark), and NFC (Near Field Communication). Note that wired communication may be used instead of wireless communication.

[0014] For example, the wearable device 10 estimates a Bp (Breakpoint) value using a Long Short-Term Memory (LSTM) network based on information from various sensors that can detect user movement and are mounted on the wearable device 10 during a scheduled time Sd (e.g., between 7:00 and 7:10) when a notification encouraging behavior (e.g., "Walk 200 steps") is scheduled in advance. Then, based on the recorded data of the estimated Bp value, the wearable device 10 infers (predicts) the probability (frequency) of subsequent Bp value occurrence from the distribution of Bp value occurrences during the scheduled time Sd, and notifies a message encouraging behavior at a timing when the Bp value is larger.

[0015] [Hardware Configuration] <Hardware Configuration of Wearable Terminal> Next, the electrical hardware configuration of the wearable terminal 10 will be described with reference to Fig. 2. Fig. 2 is a diagram showing the electrical hardware configuration of the user terminal.

[0016] As shown in FIG. 2, the wearable terminal 10 is a computer and includes a CPU 101, a ROM 102, a RAM 103, an imaging device 104, an external device connection I / F (Interface) 105, a communication device 106, a display device 107, an input device 108, a media I / F 109, a bus line 110, and various sensors 111.

[0017] Of these, the CPU 101 controls the overall operation of the wearable terminal 10. The ROM 102 stores programs such as IPL used to drive the CPU 101. The RAM 103 is used as a work area for the CPU 101.

[0018] The imaging device 104 is a type of imaging means that captures an image of a subject and obtains image data under the control of the CPU 101. Examples of such sensors include a CMOS (Complementary Metal Oxide Semiconductor) sensor and a CCD (Charge Coupled Device) sensor.

[0019] The external device connection I / F 105 is an interface for connecting various external devices by wire or wirelessly. In this case, the external devices are the electronic terminal 30 and the like.

[0020] The communicator 106 is a device that communicates with the electronic terminal 30 via short-range wireless communication.

[0021] The display device 107 is a type of display means such as a liquid crystal display or an organic electroluminescence (EL) display that displays various images. The display device 107 is a display or the like.

[0022] The input device 108 is a type of input means for selecting and executing various instructions, selecting a processing target, moving a cursor, etc. The input device 108 is a touch panel, input buttons, or the like.

[0023] The media I / F 109 controls reading and writing (storing) of data from and to a recording medium 109m such as a flash memory, etc. The recording medium 109m includes DVDs and Blu-ray Discs (registered trademarks).

[0024] The various sensors 111 include a gyro sensor, an acceleration sensor, an electronic compass, a GPS receiving device, and the like.

[0025] The bus line 110 is an address bus, a data bus, etc. for electrically connecting the components such as the CPU 101 shown in FIG.

[0026] <Hardware Configuration of Electronic Terminal> Next, the electrical hardware configuration of the electronic terminal 30 will be described with reference to Fig. 3. Fig. 3 is a diagram showing the electrical hardware configuration of the electronic terminal.

[0027] As shown in FIG. 3, the electronic terminal 30 is a computer and includes a CPU 301, a ROM 302, a RAM 303, an imaging device 304, an external device connection I / F (Interface) 305, a network I / F 306, a display device 307, an input device 308, a media I / F 309, a bus line 310, a communicator 311, and an imaging device 312.

[0028] Of these, the CPU 301 controls the overall operation of the electronic terminal 30. The ROM 302 stores programs such as IPL used to drive the CPU 301. The RAM 303 is used as a work area for the CPU 301.

[0029] The external device connection I / F 305 is an interface for connecting various external devices via wire or wirelessly. In this case, the external devices include a display device, a speaker, a keyboard, a mouse, a USB memory, a printer, and a wearable computer.

[0030] The network I / F 306 is an interface for communicating data with other devices via a communication network such as the Internet.

[0031] The display device 307 is a type of display means, such as a liquid crystal display or organic electroluminescence (EL) display, that displays various images. The display device 307 is a display or the like.

[0032] The input device 308 is a type of input means for selecting and executing various instructions, selecting a processing target, moving a cursor, etc. The input device 308 is a touch panel, input buttons, or the like.

[0033] The media I / F 309 controls reading and writing (storing) of data from and to a recording medium 309m such as a flash memory, etc. The recording medium 309m includes DVDs and Blu-ray Discs (registered trademarks).

[0034] The communicator 311 is a device that communicates with the wearable terminal 10 via short-range wireless communication.

[0035] The imaging device 312 is a type of imaging means that captures an image of a subject and obtains image data under the control of the CPU 101. For example, a complementary metal oxide semiconductor (CMOS) sensor or a charge coupled device (CCD) sensor may be used.

[0036] [Functional Configuration of the Communication System of the First Embodiment] Next, the functional configuration of the communication system 1 of the first embodiment will be described with reference to Fig. 4. Fig. 4 is a diagram showing the functional configuration of the communication system of the first embodiment. Note that the wearable terminal 10a of the first embodiment is an example of the wearable terminal 10 shown in Fig. 1. Also, the electronic terminal 30a of the first embodiment is an example of the electronic terminal 30 shown in Fig. 1.

[0037] <Functional Configuration of Wearable Terminal> First, the functional configuration of the wearable terminal 10a will be described.

[0038] As shown in Fig. 4, the wearable device 10a has a communication unit 11, a sensor unit 13, a Bp (breakpoint) value estimation unit 14, a Bp value prediction unit 17, a notification determination unit 18, and a notification unit 19. Each of these units has a function realized by an instruction from the CPU 101 in Fig. 2 based on a program. Furthermore, the ROM 102 or RAM 103 shown in Fig. 2 contains a message information storage unit 12, a Bp value estimation model storage unit 15, and a Bp value data storage unit 16.

[0039] The communication unit 11 transmits and receives data (information) to and from the electronic terminal 20 via short-range wireless communication.

[0040] The message information storage unit 12 stores information about the notification message and information about the timing of the notification.

[0041] The sensor unit 13 is realized by processing the various sensors 111 according to instructions from the CPU 101, and acquires various sensor information for detecting breakpoints.

[0042] The Bp value estimation unit 14 uses the machine learning model m1 stored in the Bp value estimation model storage unit 15 to learn and estimate the magnitude of the breakpoint.

[0043] The Bp value estimation model storage unit 15 stores, as a machine learning model m1, data on the structure of a long short-term memory (LSTM) and weight data of the LSTM.

[0044] The Bp value data storage unit 16 stores the Bp value data estimated by the Bp value estimation unit 14 .

[0045] The Bp value prediction unit 17 predicts a future Bp value based on the Bp value data stored in the Bp value data storage unit 16 .

[0046] The notification determining unit 18 determines whether to notify the user of a message to encourage action.

[0047] The notification unit 19 notifies the user of a message for promoting an action when the notification determining unit 18 has determined to notify the user. The notification method may be output by voice, text, image, vibration, or the like.

[0048] <Functional Configuration of Electronic Terminal> Next, the functional configuration of the electronic terminal 50 will be described.

[0049] 4, the electronic terminal 50 has a communication unit 31. This is a function realized by instructions from the CPU 301 in Fig. 3 based on a program. Furthermore, a message information storage unit 32 is built in the ROM 302 or RAM 303 in Fig. 3.

[0050] The communication unit 31 transmits and receives data (information) to and from the wearable terminal 10a using short-range wireless technology or the like.

[0051] The message information storage unit 32 stores message information for health promotion notifications and information on the timing of notifications.

[0052] [Processing or Operation of First Embodiment] The processing of the Bp value estimating unit 14 is divided into a learning phase and an inference phase.

[0053] In the learning phase, while the user is moving around while wearing the wearable device 10a, the machine learning model m1 including weight data for estimation is stored in the Bp value estimation model storage unit 15 based on various sensor information acquired by the sensor unit 13 and annotation data of breakpoint magnitudes based on video information that shows the user's movements. Note that the video information is acquired from the imaging device 104 shown in Fig. 2 or the imaging device 312 shown in Fig. 3.

[0054] In the inference phase, the trained machine learning model m1 is used to infer the Bp value from various pieces of sensor information acquired by the sensor unit 13 at regular intervals (time steps Ts).

[0055] This will be explained in more detail below.

[0056] <Learning Phase> Fig. 5 is a flowchart showing the process or operation of learning the machine learning model m1 in the learning phase according to the first embodiment. Fig. 6 is a schematic diagram showing the process of creating annotation data for the Bp value based on video data capturing the user's movements while wearing the wearable device 10a on the user's wrist, as an example. Note that the wearable device 10a acquires sensor information and video data showing the user's movements while the user is wearing the wearable device 10a.

[0057] S111: The Bp value estimation unit 14 acquires various pieces of sensor information from the various sensors 111.

[0058] S112: The Bp value estimation unit 14 performs preprocessing to convert various sensor information into values ​​in the range of 0 to 1, for example.

[0059] S113: The Bp value estimation unit 14 calculates the feature amount of the sensor information for each fixed time period (time step Ts).

[0060] S114: The Bp value estimation unit 14 acquires user movement information shown in image data from the imaging device 304. The user movement information is information indicating the user's behavior, as shown in FIG.

[0061] S115: The Bp value estimator 14 generates annotation data indicating whether a breakpoint has occurred at regular intervals (time steps Ts) based on subjective information (see Non-Patent Document 2). This "subjective information" indicates whether a breakpoint has occurred and is input subjectively by the user (annotator). In this case, the Bp value estimator 14 generates annotation data indicating whether a breakpoint has occurred at regular intervals (time steps Ts) based on video information showing changes in the user's behavior or subjective information from the user or a third party as described in Non-Patent Document 2, from the behavior of the user wearing the wearable device 10a.

[0062] S116: In this case, the Bp value estimation unit 14 creates annotation data including information on at least the start time of the time step, the end time of the time step, and the Bp value (magnitude value of the breakpoint) in the time step.

[0063] S117: The Bp value estimation unit 14 generates weight data (trained machine learning model m2) using LSTM (Long Short Term Memory) to predict the presence or absence of breakpoints in the annotation data from the sensor information of the time step. Specifically, the ... using LSTM (Long Short Term Memory) to predict the granularity of breakpoints corresponding to the input length L of the LSTM for the various sensors 111 mounted on the wearable device 10a for the time step t n From each time step in the past (t n-L ~t n-L-1 ) data, each feature (mean, median, maximum, minimum, or standard deviation, etc.) is input, and sThe LSTM network (machine learning model m1) is trained based on the annotation data so that the presence or absence of a breakpoint and, if a breakpoint is present, the granularity of the breakpoint can be estimated from 0 (none) to 100 (large). The Bp value estimation unit 14 then stores the trained machine learning model m2, which is the result of this training, in the Bp value estimation model storage unit 15. Note that the granularity of the breakpoint does not have to be between 0 and 100, but may be any range of values, or may be a category instead of a numerical value.

[0064] 7 and 8, a flowchart of Bp value calculation in the inference phase using the weight data of the LSTM stored in the Bp value estimation model storage unit 15 will be described. 7 and 8 are flowcharts showing processing or operations in the inference phase according to the first embodiment.

[0065] S131: The communication unit 31 of the electronic terminal 30a extracts, from the message information storage unit 32, message data (information) for promoting an action.

[0066] S132: The communication unit 31 transmits the message data to the wearable terminal 10a.

[0067] S133: The communication unit 11 of the wearable terminal 10a receives the message data.

[0068] S134: The communication unit 11 stores the message data in the message information storage unit 12.

[0069] S135: The Bp value estimating unit 14 performs Bp estimation processing. The Bp estimation processing will now be described in detail with reference to Fig. 9. Fig. 9 is a flowchart showing the Bp value estimation processing.

[0070] S151: The Bp value estimation unit 14 first obtains information on the schedule time Sd of the prompt to act from the message information storage unit 12.

[0071] S152: The Bp value estimation unit 14 starts a scheduled time confirmation process. In this case, the Bp value estimation unit 14 confirms whether the current time Ts is within the scheduled time Sd.

[0072] S153: If it is within the scheduled time Sd, the Bp value estimation unit 14 acquires various sensor information from the sensor unit 13.

[0073] S154: The Bp value estimation unit 14 performs preprocessing to convert the various sensor information into values ​​in the range of 0 to 1, for example.

[0074] S155: The Bp value estimation unit 14 reads out the weight data of the trained machine learning model m2 from the Bp value estimation model storage unit 15.

[0075] S156: The Bp value estimation unit 14 estimates the Bp value for each time step t using the trained machine learning model m2. n The feature quantities (mean, median, maximum, minimum, standard deviation, etc.) of information such as acceleration that can tell the user's movement are input, and the current time step t n In this case, the Bp value is estimated for each time step t n is the time step t of the current time obtained from the sensor unit 13 n Based on the LSTM input length L, the past time step t n-l From the time step t before the current time n-1 Shows up to.

[0076] The Bp value may be estimated based on a heart rate sensor, a touch sensor, a camera sensor, or the like, as other examples of the various sensors 111, as long as the Bp value can be estimated from information about the movement of the wearable terminal 10a.

[0077] S157: This completes the schedule time confirmation process.

[0078] Next, the processing or operation will be described by returning to FIG.

[0079] S136: The Bp value estimation unit 14 stores the estimated Bp value in the Bp value data storage unit 16.

[0080] S137: The Bp value prediction unit 17 generates distribution data (hereinafter simply referred to as "distribution data") of Bp values ​​at the schedule time Sd for which the same previous prompt notification was scheduled and stored in the Bp value data storage unit 16. FIGS. 10 and 11 show an example in which a prompt notification is sent between 7:00 and 7:10, with the time step Ts set to 20 seconds. FIG. 10 is a conceptual diagram showing, as an example, a process in which the wearable device 10a is worn on the user's wrist and the Bp value is estimated based on the user's movements at that time. FIG. 11 is a diagram showing an example in which a prompt notification is sent between 7:00 and 7:10, with the time step Ts set to 20 seconds.

[0081] Here, the Bp value prediction unit 17 calculates (predicts) the occurrence frequency of the Bp value from the current time to the end of the schedule time Sd based on the past Bp value data from the current time to the end of the schedule time Sd stored in the Bp value data storage unit 16. Then, the Bp value prediction unit 17 calculates (predicts) the occurrence frequency of the Bp value from the calculated occurrence frequency to the preset upper probability threshold P threshold Based on this, the threshold value B threshold_Sd Calculate.

[0082] S138: The notification determination unit 18 determines whether the Bp value calculated by the Bp value estimation unit 14 is B threshold_Sd Determine whether the Bp value is greater than or equal to B threshold_Sd If it is not equal to or greater than this (NO), the process returns to step S135.

[0083] S139: Bp value is B threshold_Sd In the above cases (S138; YES), the notification unit 19 notifies the user of a message encouraging behavior such as health promotion.

[0084] Effect of the First Embodiment As described above, according to the present embodiment, in order to send a notification at a timing within the scheduled time when the Bp value (predetermined value) is larger, when a breakpoint is detected, the distribution of Bp values ​​within the scheduled time range of the notification is inferred (predicted) and compared, thereby making it possible to send a notification at a timing when the Bp value is higher (when the frequency of occurrence is higher than the threshold value).

[0085] Furthermore, the wearable device 10a estimates the Bp value in advance from information about the movement of the wearable device 10a (from the various sensors 1111) and records the estimation results, thereby calculating the occurrence frequency of the Bp value for each scheduled time period for notifications to encourage the user to take action. This allows the Bp value threshold to be dynamically determined according to the occurrence frequency information, making it possible to notify the user at more appropriate timing.

[0086] Second Embodiment Next, a second embodiment will be described. Note that a wearable terminal 10b of the second embodiment is an example of the wearable terminal 10 shown in Fig. 1. Also, an electronic terminal 30b of the second embodiment is an example of the electronic terminal 30 shown in Fig. 1.

[0087] The communication system 2 of the second embodiment illustrates an example in which the distribution of Bp values ​​in particular varies significantly depending on lifestyle habits in the communication system 1 of the first embodiment. The wearable device 10b of the second embodiment selects distribution data of Bp values ​​to refer to when determining an appropriate threshold, and predicts the probability of occurrence of a Bp value from the distribution of Bp values ​​related to the selected distribution data, thereby notifying a message promoting healthy behavior at a time when the Bp value is larger.

[0088] Specifically, the first embodiment described above shows a method for notifying at a more appropriate timing by creating and estimating a Bp value distribution from records of past Bp values ​​at a scheduled time Sd for which message notification is scheduled in advance. In contrast, the second embodiment enables more appropriate threshold setting by selecting information from records of Bp values ​​to use from patterns that significantly change the distribution.

[0089] The electrical configuration of the second embodiment is the same as that of the first embodiment, and therefore a description thereof will be omitted.

[0090] [Functional Configuration of Communication System of Second Embodiment] Next, each functional configuration of the communication system 1 of the second embodiment will be described with reference to Fig. 12. Fig. 12 is a diagram showing each functional configuration of the communication system of the second embodiment. Note that in the second embodiment, the same components as those in the first embodiment are denoted by the same reference numerals, and their description will be omitted.

[0091] <Functional Configuration of Wearable Device> As shown in Fig. 12, the wearable device 10b has a Bp value data selection unit 23 in addition to the functional configuration of the wearable device 10a of the first embodiment. These units are functions realized by instructions from the CPU 101 of Fig. 2 based on a program. Furthermore, in the ROM 102 or RAM 103 shown in Fig. 2, in addition to the storage units or memory units of the wearable device 10a of the first embodiment, a schedule information storage unit 21 and a Bp value cluster estimation model memory unit 22 are configured.

[0092] The schedule information storage unit 21 stores the user's daily schedule information Si. The schedule information Si includes information on the weather, date, time, day of the week, and plans registered in the scheduler during the schedule time Sd, or information on plans registered in the scheduler before and after the above times. Note that the schedule information Si may also include information on whether the user is at work, information on holidays, information on the daylight saving time period, etc., as long as it is information that can reveal the user's lifestyle pattern.

[0093] The Bp value cluster estimation model storage unit 22 stores a Bp value cluster estimation model M1 as a machine learning model.

[0094] The Bp value data selection unit 23 uses the Bp value cluster estimation model (machine learning model) M1 stored in the Bp value cluster estimation model storage unit 22 to select appropriate distribution data B based on the schedule information Si that can be acquired before (in the past) the schedule time Sd. distribution For example, if a person works from Monday to Friday and has Saturday and Sunday off, the rhythm of their life is different for these two life patterns, so distribution data B suitable for both work days and off days is selected. distribution Make a selection.

[0095] [Processing or Operation of Second Embodiment] The processing of the Bp value data selection unit 23 is divided into a learning phase and an inference phase.

[0096] <Learning Phase> Fig. 13 is a flowchart showing the processing or operation in the learning phase according to the second embodiment. Fig. 13 shows the processing or operation performed by the Bp value data selection unit 23 in the learning phase.

[0097] In the learning phase, the machine learning process performed by the Bp value estimating unit 14 is the same as in the first embodiment, and therefore a description thereof will be omitted.

[0098] S211: The Bp value data selection unit 23 selects the occurrence distribution data B distribution Obtain distribution data B distribution includes information on the Bp value for each time step in the schedule time Sd.

[0099] S212: The Bp value data selection unit 23 selects each distribution data B distribution The schedule information Si corresponding to the schedule information Si is acquired from the schedule information storage unit 21.

[0100] S213: The Bp value data selection unit 23 selects the distribution data B distribution Clustering is performed on the data using an unsupervised machine learning model M1 such as the X-means method. The cluster of distribution patterns obtained here is called cluster C. distribution It is called.

[0101] S214: The Bp value data selection unit 23 selects cluster C of the obtained distribution pattern. distribution Machine learning is performed on the machine learning model M1 so that it is possible to select which cluster (work day, holiday day, etc.) it is from the schedule information Si linked to each data, and the weight data of the trained machine learning model (Bp value cluster estimation model) M2 is stored in the Bp value cluster estimation model memory unit 22.

[0102] <Inference Phase> FIGS. 14 and 15 are flowcharts showing the processing or operation in the inference phase according to the second embodiment.

[0103] S231: The communication unit 31 of the electronic terminal 30 a extracts message data (information) for action promotion from the message information storage unit 32 , and extracts schedule information from the schedule information storage unit 33 .

[0104] S232: The communication unit 31 transmits the message data and schedule information to the wearable terminal 10a.

[0105] S233: The communication unit 11 of the wearable terminal 10a receives the message data and the schedule information.

[0106] S234: The communication unit 11 stores the message data in the message information storage unit 12 and stores the schedule information in the schedule information storage unit 21.

[0107] S235: The Bp value estimating unit 14 performs Bp estimation processing. Note that this processing is the same as S135, and therefore a description thereof will be omitted.

[0108] S236: The Bp value estimation unit 14 stores the estimated Bp value in the Bp value data storage unit 16.

[0109] S237: The Bp value data selection unit 23 performs a Bp value data selection process.

[0110] The Bp value data selection process in the inference phase will now be described in detail with reference to Fig. 16. Fig. 16 is a flowchart showing the Bp value data selection process in the inference phase.

[0111] S271: The Bp value data selection unit 23 acquires the schedule information Si corresponding to the schedule time Sd of the promotion notification from the schedule information storage unit 21.

[0112] S272: The Bp value data selection unit 23 acquires weight data of the trained machine learning model M2 from the Bp value cluster estimation model storage unit 22.

[0113] S273: The Bp value data selection unit 23 uses the weight data of the trained machine learning model M2, inputs the schedule information Si, and selects the cluster C distribution Then, the Bp value data selection unit 23 estimates the estimated C distribution Distribution data B corresponding to distribution Select .

[0114] FIG. 17 is a diagram showing an outline of an example in which a prompt to act notification is sent between 7:00 and 7:10 and the width of the time step Ts is set to 20 seconds according to the second embodiment.

[0115] Here, the Bp value data selection unit 23 generates distribution data for two clusters using the trained machine learning model M2, and when cluster c1 (for example, Saturday and Sunday) is estimated from the schedule information Si, the data of cluster c2 (for example, weekdays) is not used, and the distribution data B distribution Only using B threshold_Sd On the other hand, when the Bp value data selection unit 23 estimates cluster c2 from the schedule information Si, it does not use the data of cluster c1, but uses the distribution data B distribution Only using B threshold_Sd is calculated and a decision is made as to whether or not to notify.

[0116] In this case, the Bp value data selection unit 23 selects the distribution data B distribution Distribution data B of Bp values ​​that were judged to be in the same category when categorizing using the trained machine learning model M2 distribution Distribution data B of a given Bp value among the group distribution The Bp value data selection unit 23 uses the trained machine learning model M2 to predict, before the scheduled time Sd, which category (lifestyle pattern) the distribution data of the Bp values ​​for the day will fall into based on the input schedule information Si.

[0117] This completes the process S237 in FIG.

[0118] The Bp value data selection unit 23 does not use the trained machine learning model M2, but selects the distribution data Bp values ​​to be used based on setting information set in advance for each user. distribution In this case, the setting information may be, for example, information on a lifestyle pattern indicating the days of the week on which working days and days off are performed.

[0119] Next, the processing or operation of FIG. 15 will be described.

[0120] S238: The Bp value prediction unit 17 uses the distribution data B selected by the Bp value data selection unit 23 distribution Based on this, the threshold value B threshold_Sd Calculate.

[0121] S239: The notification determination unit 18 determines whether the Bp value calculated by the Bp value estimation unit 14 is B threshold_Sd Determine whether the Bp value is greater than or equal to B threshold_Sd If there are no more (NO), the process returns to step S235.

[0122] S240: Bp value is B threshold_Sd In the above cases (S239; YES), the notification unit 19 notifies the user of a message to promote healthy behavior.

[0123] [Effects of the Second Embodiment] As described above, the second embodiment provides the same effects as the first embodiment.

[0124] Furthermore, in the second embodiment, by selecting the recorded information of the Bp value to be used from patterns that significantly change the distribution, it becomes possible to set a more appropriate threshold value.

[0125] [Supplementary Note] The present invention is not limited to the above-described embodiment, and may have the following configurations or processes (operations).

[0126] (1) The wearable terminal 10 and the electronic terminal 30 can be realized by a computer and a program, but this program can also be recorded on a (non-temporary) recording medium or provided via the communication network 100.

[0127] (2) Each of the CPUs 101 and 301 may be a single CPU or may be multiple CPUs.

[0128] REFERENCE SIGNS LIST 1 Communication system 10, 10a, 10b Wearable terminal 30, 30a, 30b Electronic terminal 11 Communication unit 13 Sensor unit 14 Bp value estimation unit (an example of an estimation unit) 17 Bp value prediction unit (an example of a prediction unit) 18 Notification determination unit 19 Notification unit 23 Bp value data selection unit (an example of a selection unit) m2 Trained machine learning model (an example of a first trained machine learning model) M2 Trained machine learning model (an example of a second trained machine learning model)

Claims

1. A wearable device that notifies a user of a message to encourage behavior, comprising: a sensor unit that detects the user's movements and obtains sensor information; an estimation unit that estimates a predetermined value that is the granularity of timing that is currently not bothersome to the user based on the sensor information; a prediction unit that calculates a predetermined threshold by predicting the occurrence frequency of the predetermined value from occurrence distribution data of the predetermined value based on the user's schedule information; and a notification determination unit that determines the timing to notify the message when the predetermined value is equal to or greater than the predetermined threshold.

2. The wearable terminal of claim 1, wherein the estimation unit performs the estimation using a trained machine learning model for inputting the sensor information and outputting a value of the granularity size.

3. A wearable terminal as described in claim 1, comprising a selection unit that selects predetermined occurrence distribution data from a plurality of occurrence distribution data based on the user's lifestyle pattern, and the prediction unit predicts the occurrence frequency of the predetermined value from the predetermined occurrence distribution data selected by the selection unit.

4. A notification method executed by a wearable device that notifies a user of a message to encourage behavior, wherein the wearable device executes the following: a sensor information acquisition process that detects the user's movements and obtains sensor information; an estimation process that estimates a predetermined value, which is the granularity of a timing that is currently not bothersome to the user, based on the sensor information; a prediction process that calculates a predetermined threshold by predicting the occurrence frequency of the predetermined value from occurrence distribution data of the predetermined value based on the user's schedule information; and a notification determination process that determines the timing to notify the message when the predetermined value is equal to or greater than the predetermined threshold.

5. A program for causing a computer to execute the method according to claim 4.