Prediction device, prediction method, and prediction program

Abstract

To accurately predict the quality of sleeping by using a fragmentation degree indicating the degree of variation in a body activity.SOLUTION: A prediction device comprises: an acquisition part which acquires activity data about a body activity; a fragmentation degree calculation part which calculates a fragmentation degree indicating the degree of variation in the activity based on the activity data; a prediction part which uses a prediction model indicating the correspondence between the fragmentation degree and the quality of sleeping, predicts the quality of sleeping corresponding to the calculated fragmentation degree; and an output part which outputs the predicted quality of sleeping.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] This disclosure relates to a prediction device, a prediction method, and a prediction program. [Background technology]

[0002] Patent Document 1 discloses a biorhythm prediction device that includes: an acquisition unit that acquires a biosignal of the user's ultradian rhythm period and influencing factors including the user's activity level, information on the external environment around the user, and at least one of behavioral indicators related to the user's daily activities; an input unit that receives input of planned actions which are actions that the user plans to perform; an extraction unit that extracts an ultradian rhythm waveform of the user's biosignal from the biosignal, with the influence based on the influencing factors reduced; and a prediction unit that predicts an ultradian rhythm waveform for a target time which is the time when the planned action will be performed, based on the ultradian rhythm waveform and the influencing factors.

[0003] Patent Document 2 discloses an action suggestion device comprising a processing unit and a storage unit, which makes action suggestions for an individual based on an individual's biometric information, wherein the processing unit determines the individual's actions based on the collected individual's biometric information and converts them into action element data, generates the individual's habitual behavior data from the action element data for a predetermined period, generates the individual's action guidelines based on the action element data, the habitual behavior data, and the individual's target indicator data, and makes action suggestions for the individual based on the generated action guidelines.

[0004] Patent Document 3 discloses a biological rhythm adjustment device characterized by comprising: a rhythm curve input means for inputting a biological rhythm curve; an adjustment target setting means for pre-setting a biological rhythm curve to be an adjustment target; an environment information input means for inputting information about the stimulating environment; a stimulus condition determination means for determining stimulus conditions according to the input biological rhythm curve, the set adjustment target, and the input environment information; a stimulus means for providing a stimulus to the living body that changes the biological rhythm curve; and an adjustment result evaluation means for evaluating whether the biological rhythm curve after stimulation has reached the set adjustment target and feeding the evaluation result back to the stimulus condition determination means.

[0005] Patent Document 4 discloses a biological rhythm control system comprising: a physiological response measurement unit that measures physiological responses that reflect fluctuations in the level of arousal; a biological rhythm extraction unit that extracts biological rhythms, which are circadian rhythms (fluctuations with a period of approximately 24 hours) and diurnal rhythms (fluctuations with a period of approximately 1.5 to 2 hours), from the measured physiological responses; an input unit for inputting a control target; a control output determination unit that creates a prediction curve of the rhythm from the extracted biological rhythms, determines a phase control amount from the difference between the created prediction curve and the control target, and determines the parameters of an external stimulus to cause a predetermined phase change using the phase control amount and phase response curve data that shows the phase change in response to an external stimulus and is stored in advance; and a stimulus presentation unit that provides an external stimulus based on the control output of the control output determination unit. [Prior art documents] [Patent Documents]

[0006] [Patent Document 1] Japanese Patent Publication No. 2021-49041 [Patent Document 2] International Publication No. 2010 / 146811 [Patent Document 3] Japanese Patent Application Publication No. 05-3874 [Patent Document 4] Patent No. 2917592 [Overview of the Initiative] [Problems that the invention aims to solve]

[0007] The conventional technologies described above include techniques that predict a user's state by reflecting ultradian rhythm waveforms, circadian rhythm waveforms, and combinations of the user's sleep duration and behavior, and then provide recommended actions and advice. However, they do not consider the degree of fragmentation, which represents the degree of variation in physical activity such as exercise intensity and activity cycles. Since the degree of fragmentation is related to sleep quality, it is important to take into account the degree of physical activity fragmentation in order to accurately predict sleep quality.

[0008] This disclosure aims to provide a prediction device, prediction method, and prediction program that can accurately predict sleep quality using a fragmentation degree that represents the degree of variation in bodily activity. [Means for solving the problem]

[0009] The prediction device according to the first embodiment includes: an acquisition unit that acquires activity data relating to physical activity; a fragmentation degree calculation unit that calculates a fragmentation degree representing the degree of variation of the activity based on the activity data; a prediction unit that predicts the sleep quality corresponding to the calculated fragmentation degree using a prediction model that represents the correspondence between the fragmentation degree and sleep quality; and an output unit that outputs the predicted sleep quality.

[0010] In the prediction device according to the first embodiment, the device may be configured to include: a quality calculation unit that calculates the quality of sleep based on the activity data; an storage unit that stores the acquired activity data, the calculated degree of fragmentation, and the calculated quality of sleep in association with each other; and a model creation unit that creates the prediction model based on the stored activity data, the degree of fragmentation, and the quality of sleep, taking the degree of fragmentation as input and the quality of sleep as output.

[0011] In the prediction device according to the first embodiment, the output unit may output difference data relating to the difference between the time waveform of the activity data corresponding to the optimal sleep quality with the highest sleep quality among the sleep quality stored by the storage unit, and the time waveform of the activity data acquired by the acquisition unit.

[0012] In the prediction device according to the first embodiment, the acquisition unit may acquire behavioral data corresponding to the activity data, the storage unit may store the acquired activity data, the calculated degree of fragmentation, the calculated sleep quality, and the behavioral data in association with each other, and the output unit may output the behavioral data corresponding to the difference.

[0013] In the prediction device according to the first embodiment, the acquisition unit acquires activity data of multiple different users, the storage unit stores the acquired activity data, the calculated degree of fragmentation, the calculated quality of sleep, the behavioral data, and the users in association, and the output unit may output difference data relating to the difference between the deviation score of the degree of fragmentation, the time waveform of the activity data of another user whose behavioral pattern based on the behavioral data is most similar and whose quality of sleep is the highest among the other users, and the time waveform of the activity data acquired by the acquisition unit.

[0014] The prediction method according to the second embodiment includes the following processes: a computer acquires activity data relating to physical activity, calculates a degree of fragmentation representing the degree of variation in the activity based on the activity data, predicts the quality of sleep corresponding to the calculated degree of fragmentation using a prediction model that represents the correspondence between the degree of fragmentation and the quality of sleep, and outputs the predicted quality of sleep.

[0015] The prediction program according to the third embodiment causes a computer to perform a process that includes acquiring activity data relating to physical activity, calculating a degree of fragmentation representing the degree of variation in the activity based on the activity data, predicting the quality of sleep corresponding to the calculated degree of fragmentation using a prediction model that represents the correspondence between the degree of fragmentation and the quality of sleep, and outputting the predicted quality of sleep. [Effect of the Invention]

[0016] According to the present disclosure, there is an effect that the quality of sleep can be accurately predicted by using the fragmentation degree representing the degree of variation in body activity. [Brief Description of the Drawings]

[0017] [Figure 1] It is a configuration diagram of a prediction system. [Figure 2] It is a block diagram showing the hardware configuration of a server. [Figure 3] It is a functional block diagram of a server. [Figure 4] It is a diagram showing an example of the frequency spectrum of activity data. [Figure 5] It is a diagram showing the relationship between the fragmentation degree during wakefulness and sleep stability. [Figure 6] It is a diagram showing the time change of activity data. [Figure 7] It is a flowchart of a prediction model creation process. [Figure 8] It is a flowchart of a prediction process according to the first embodiment. [Figure 9] It is a diagram showing an example of the time waveform of activity data. [Figure 10] It is a diagram showing an example of the time waveform of activity data. [Figure 11] It is a diagram showing an example of difference data. [Figure 12] It is a flowchart of a prediction process according to the second embodiment. [Figure 13] It is a flowchart of a prediction process according to the third embodiment. [Figure 14] It is a diagram showing an example of the display screen of a mobile terminal device. [Figure 15] It is a flowchart of a prediction process according to the fourth embodiment. [Modes for Carrying Out the Invention]

[0018] This embodiment will be described below with reference to the drawings. The same components and processes will be denoted by the same reference numerals throughout the drawings, and redundant explanations will be omitted.

[0019] <First Embodiment>

[0020] Figure 1 shows an example of the system configuration of the prediction system 10 according to the first embodiment. The prediction system 10 is a system that predicts sleep quality. The prediction system 10 has a configuration in which a server 20, which is an example of a prediction device, and multiple portable terminal devices 40, each carried by multiple users 30, are connected via a network 50.

[0021] The mobile terminal device 40 can communicate with activity trackers 60 attached to parts of the user's body, such as the wrist, waist, and chest, for example, by short-range wireless communication.

[0022] The activity tracker 60 measures activity data related to the user 30's physical activity and transmits it to the mobile terminal device 40. The activity data is measured, for example, at predetermined time intervals and transmitted sequentially to the mobile terminal device 40.

[0023] Activity data includes, for example, at least one of activity level and heart rate index. Activity level is a numerical value indicating the intensity of physical activity, calculated, for example, from an accelerometer. Heart rate index is a numerical value related to cardiac activity, such as heart rate. In other words, activity data is data that fluctuates depending on the physical activity of the user 30.

[0024] The mobile terminal device 40 transmits activity data received from the activity tracker 60 to the server 20 via the network 50. The activity tracker 60 measures activity data, for example, at predetermined time intervals, and transmits the data to the mobile terminal device 40 each time it measures activity data. The mobile terminal device 40 transmits the activity data received from the activity tracker 60 to the server 20 each time it receives activity data.

[0025] Figure 2 is a block diagram showing the hardware configuration of server 20. As shown in Figure 2, server 20 includes a controller 21. The controller 21 includes a CPU (Central Processing Unit) 21A, a ROM (Read Only Memory) 21B, a RAM (Random Access Memory) 21C, and an input / output interface (I / O) 21D. The CPU 21A, ROM 21B, RAM 21C, and I / O 21D are connected to each other via a system bus 21E. The system bus 21E includes a control bus, an address bus, and a data bus.

[0026] CPU21A is an example of a computer. Here, "computer" refers to a processor in a broad sense, including general-purpose processors (e.g., CPUs) or specialized processors (e.g., GPUs: Graphics Processing Units, ASICs: Application Specific Integrated Circuits, FPGAs: Field Programmable Gate Arrays, programmable logic devices, etc.).

[0027] Furthermore, the communication unit 22 and the storage unit 23 are connected to I / O 12D.

[0028] The communication unit 22 is an interface for data communication with external devices such as the mobile terminal device 40.

[0029] The memory unit 23 is composed of a non-volatile external storage device such as a hard disk, and stores the prediction program 23A, the prediction model 23B, and the activity database 23C. The CPU 12A reads the prediction program 23A stored in the memory unit 23 into the RAM 21C and executes it.

[0030] Figure 3 is a block diagram showing the functional configuration of the CPU 21A of the server 20. As shown in Figure 3, the CPU 21A functionally comprises an acquisition unit 71, a fragmentation degree calculation unit 72, a prediction unit 73, an output unit 74, a sleep quality calculation unit 75, an storage unit 76, and a model creation unit 77.

[0031] The acquisition unit 71 acquires activity data related to physical activity. Specifically, the acquisition unit 71 acquires activity data measured by an activity tracker 60 worn by the user 30 by receiving it from the mobile terminal device 40 via the network 50.

[0032] The fragmentation degree calculation unit 72 calculates a fragmentation degree, which represents the degree of variation in physical activity, based on the activity data acquired by the acquisition unit 71.

[0033] The degree of fragmentation can be calculated, for example, by analyzing the diurnal variation of activity data during wakefulness, and can be calculated using the following formula, which is an application of the formula described in Reference 1 below. (Reference 1) Witting W, et al., https: / / doi.org / 10.1016 / 0006-3223(90)90523-5

[0034] JPEG0007845111000001.jpg1872...(1)

[0035] Here, IV is the degree of fragmentation. N is the total number of data points for awake activity data. Xi is the activity data for each time point i (i=1,2,...,N) where activity data was measured.

[0036] JPEG0007845111000002.jpg76 This is the average value of all awake activity data.

[0037] Furthermore, fragmentation degree IV can also be calculated based on the frequency spectrum shown in Figure 4, which is obtained by frequency analysis of activity data during wakefulness.

[0038] As shown in Figure 5, there is a correlation where the less fragmentation of physical activity during wakefulness, the higher the quality of sleep (sleep stability).

[0039] Therefore, the prediction unit 73 uses a prediction model 23B that represents the correspondence between the degree of fragmentation and the quality of sleep to predict the quality of sleep corresponding to the degree of fragmentation calculated by the degree of fragmentation calculation unit 72.

[0040] The output unit 74 outputs the sleep quality predicted by the prediction unit 73. For example, the output unit 74 transmits the sleep quality predicted by the prediction unit 73 to the mobile terminal device 40 via the network 50. As a result, the predicted sleep quality is displayed on the display unit of the mobile terminal device 40, and the user 30 can understand how good the quality of their sleep is for the day based on their physical activity that day.

[0041] Predictive model 23B is created by the model creation unit 77. Predictive model 23B is a model that takes fragmentation degree as input and sleep quality as output. Various activity data, fragmentation degree, and sleep quality are required to create predictive model 23B.

[0042] Therefore, the sleep quality calculation unit 75 calculates the sleep quality for that day based on the activity data acquired by the acquisition unit 71, using the following formula which applies the formula described in Reference 1 above.

[0043] JPEG0007845111000003.jpg1972...(2)

[0044] Here, IS represents sleep quality, or sleep stability. A higher value of sleep quality IS indicates higher sleep quality, while a lower value indicates lower sleep quality. p is the number of activity data points during sleep per day. Sleep quality IS is calculated over several days of sleep. X h X is the average value over several days at each time point h (h=1,2,···,p) of the activity data. m is the average value of all sleep activity data. N is the number of activity data points during sleep.

[0045] Here, whether the activity data is wakefulness activity data or sleep activity data can be determined by the following formula described in the publicly known reference 2 below. (Reference 2) te Lindert B, et al., https: / / doi.org / 10.5665 / sleep.2648

[0046] A0 = 0.04E -(8~5) + 0.2E -(4~1) + 4E0 + 0.2E +(1~4) + 0.04E +(5~8) ···(3)

[0047] Here, A0 is a score for determining whether it is wakefulness or sleep. E0 is the activity data for 15 seconds at the determination time for determining whether it is wakefulness or sleep. E n is the activity data of a predetermined section when the two minutes before and after the determination time for determining whether it is wakefulness or sleep are divided into 16 sections every 15 seconds. That is, E -(8~5) is the activity data of the section from 2 minutes before the determination time to 75 seconds before. Also, E -(4~1) is the activity data of the section from 60 seconds before the determination time to 15 seconds before. Also, E +(1~4) is the activity data of the section from 15 seconds after the determination time to 60 seconds after. Also, E +(5~8) is the activity data of the section from 75 seconds after the determination time to 2 minutes after.

[0048] And when the score A0 is less than or equal to a predetermined threshold value T, it is determined as sleep, and when the score A0 is greater than the predetermined threshold value T, it is determined as wakefulness.

[0049] Also, the storage unit 76 stores, for each user, the activity data acquired by the acquisition unit 71, the fragmentation degree calculated by the fragmentation degree calculation unit 72, and the sleep quality calculated by the sleep quality calculation unit 75, in association with each other, as an activity database 23C in the storage unit 23.

[0050] The model creation unit 77 creates a predictive model 23B based on the activity data, fragmentation degree, and sleep quality accumulated by the storage unit 76, taking the fragmentation degree IV as input and the sleep quality IS as output. The model creation unit 77 may, for example, use a known regression analysis method to create a linear or polynomial regression equation as the predictive model 23B, or it may use known machine learning to create the predictive model 23B. Furthermore, when creating the predictive model 23B, parameters such as sleep latency, sleep duration, and the number of awakenings during the night, which are determined based on the activity data, may also be used to create the predictive model 23B, for example, as shown in Figure 6.

[0051] Next, the predictive model creation process executed on the CPU 21A of the server 20 will be explained with reference to the flowchart shown in Figure 7. Note that the process shown in Figure 7 is executed repeatedly each time activity data is received from the mobile terminal device 40.

[0052] In step S100, the CPU 21A acquires activity data transmitted from the mobile terminal device 40.

[0053] In step S101, the CPU 21A calculates the fragmentation degree IV using the above formula (1) based on the activity data acquired in step S100.

[0054] In step S102, the CPU 21A calculates the sleep quality IS for the day based on the activity data acquired in step S100 using the above formula (2).

[0055] In step S103, the CPU 21A associates the activity data acquired in step S100, the fragmentation degree IV calculated in step S101, and the sleep quality IS calculated in step S102, and stores them in the activity database 23C.

[0056] In step S104, the CPU 21A creates a predictive model 23B that takes the fragmentation degree IV as input and the sleep quality IS as output, based on the activity data, fragmentation degree IV, and sleep quality IS accumulated in the activity database 23C in step S103.

[0057] In this way, the prediction model 23B is updated sequentially by executing the prediction model creation process shown in Figure 7 each time activity data is received.

[0058] Next, the prediction process performed by the CPU 21A of the server 20 will be explained with reference to the flowchart shown in Figure 8. Note that the process shown in Figure 8 is executed when the mobile terminal device 40 requests a prediction of sleep quality. Furthermore, the process in Figure 8 is executed in parallel with the process in Figure 7.

[0059] In step S200, the CPU 21A retrieves the activity data for the day of the user of the mobile terminal device 40 that was instructed to predict sleep quality from the activity database 23C.

[0060] In step S201, the CPU 21A calculates the fragmentation degree IV using the above formula (1) based on the activity data acquired in step S200.

[0061] In step S202, the fragmentation degree IV calculated in step S201 is input into the prediction model 23B to obtain the sleep quality corresponding to the calculated fragmentation degree IV, that is, the sleep quality predicted for that day.

[0062] In step S203, the CPU 21A transmits the sleep quality data obtained in step S202 to the mobile terminal device 40 that was instructed to predict the sleep quality.

[0063] As a result, the quality of sleep for the day is displayed on the display unit of the mobile terminal device 40, for example, as a numerical value. Therefore, the user 30 can understand how their sleep quality is affected by their physical activity for the day.

[0064] <Second Embodiment>

[0065] Next, the second embodiment will be described. Note that the configuration of the prediction system 10 is the same as in the first embodiment, so its description will be omitted. Similarly, the prediction model creation process is also the same as in the first embodiment, so its description will be omitted.

[0066] In the second embodiment, the time waveform of each user's daily activity data is stored in the activity database 23C. Figures 9 and 10 show examples of time waveforms of activity data. Figure 9 is an example of a time waveform of activity data with a fragmentation degree of 0.21. Figure 10 is an example of a time waveform of activity data with a fragmentation degree of 0.76.

[0067] The output unit 74 outputs difference data relating to the difference between the time waveform of the activity data corresponding to the optimal sleep quality, which is the highest quality sleep quality among the sleep quality stored in the activity database 23C, and the time waveform of the activity data for that day acquired by the acquisition unit 71.

[0068] Specifically, for example, if the time waveform of activity data corresponding to optimal sleep quality is the time waveform shown in Figure 9, and the time waveform of activity data for that day is the time waveform shown in Figure 10, then the time waveform shown in Figure 11, which is a superimposition of the time waveforms in Figure 9 and Figure 10, is output as difference data. This allows user 30 to easily understand the difference between the time waveform of activity data for optimal sleep quality and the time waveform of activity data for that day, making it easier to take actions to achieve optimal sleep quality.

[0069] Next, the prediction process executed on the CPU 21A of server 20 will be explained with reference to the flowchart shown in Figure 12. Note that steps identical to those shown in Figure 8 are denoted by the same reference numerals and their explanations are omitted.

[0070] The prediction process in Figure 12 includes steps S204 and S205, compared to the prediction process in Figure 8.

[0071] In step S204, the CPU 21A acquires the time waveform of activity data corresponding to the optimal sleep quality and the time waveform of activity data for that day from among the time waveforms of activity data stored in the activity database 23C.

[0072] In step S205, the CPU 21A generates a time waveform by superimposing the time waveform of the activity data corresponding to the optimal sleep quality obtained in step S204 with the time waveform of the activity data for that day, and transmits this as difference data to the mobile terminal device 40. This makes it possible for the user 30 to easily understand the difference between the physical activity on an optimal sleep day and the physical activity on that day, and to encourage them to take actions to achieve optimal sleep quality.

[0073] <Third Embodiment>

[0074] Next, a third embodiment will be described. Note that the configuration of the prediction system 10 is the same as in the first embodiment, so its description will be omitted. Similarly, the model creation process is also the same as in the first embodiment, so its description will be omitted.

[0075] In the third embodiment, the acquisition unit 71 acquires behavioral data corresponding to the activity data. The storage unit 76 stores the acquired activity data, the calculated fragmentation degree, the calculated sleep quality, and the behavioral data in the activity database 23C, associating them with each other.

[0076] Behavioral data is data that associates a pre-assigned behavior label with the time period in which the behavior occurred. For example, if user 30 wakes up at 6:30, the behavior label will be "Wake up" and the time period will be "6:30". Also, if user 30 plays amateur baseball from 19:30 to 20:20, the behavior label will be "Amateur baseball" and the time period will be "19:30 to 20:20".

[0077] Behavioral data is entered, for example, by user 30 into a mobile terminal device 40. This allows the mobile terminal device 40 to transmit the behavioral data to server 20, where activity data, fragmentation level, sleep quality, and behavioral data are associated and stored in the activity database 23C.

[0078] The output unit 74 then outputs behavioral data corresponding to the difference between the time waveform of the activity data corresponding to the optimal sleep quality with the highest sleep quality and the time waveform of the activity data for that day acquired by the acquisition unit 71, i.e., behavioral data representing recommended behaviors.

[0079] Specifically, for example, suppose the time waveform of activity data corresponding to the optimal sleep quality shown in Figure 9 is associated with the behavioral data "played amateur baseball from 19:30 to 20:20". Also, suppose the time waveform of the activity data for that day shown in Figure 10 is associated with the behavioral data "woke up at 6:30".

[0080] In this case, the time waveform shown in Figure 11, which is a superimposition of the time waveforms in Figure 9 and Figure 10, reveals that the activity from 19:30 to 20:20 is different. Therefore, in order to bring the quality of sleep on that day closer to the optimal quality, it can be seen that it is effective to engage in recommended activities such as playing amateur baseball during the 19:30 to 20:20 timeframe, which will result in larger activity data.

[0081] For example, recommended actions based on differences in time of day and activity data may be predetermined. If there is a time period where the difference between the activity data corresponding to the optimal sleep quality (which results in the highest sleep quality) and the activity data for that day exceeds a predetermined threshold, the system may output action data for the recommended action based on the difference in time of day and activity data for that period.

[0082] Next, the prediction process executed on the CPU 21A of server 20 will be explained with reference to the flowchart shown in Figure 13. Note that steps identical to those shown in Figure 12 are denoted by the same reference numerals and their explanations are omitted.

[0083] The prediction process in Figure 13 has an additional step, S206, compared to the prediction process in Figure 12.

[0084] In step S206, the CPU 21A generates a differential time waveform by superimposing the time waveform of the activity data corresponding to the optimal sleep quality obtained in step S204 with the time waveform of the activity data for that day, and transmits the action data corresponding to the generated differential data to the mobile terminal device 40. This makes it easy for the user 30 to understand what specific actions they should take to bring the quality of their sleep for that day closer to the optimal sleep quality.

[0085] Figure 14 shows an example of a display screen shown on the mobile terminal device 40. As shown in Figure 14, the display screen 41 has a display area 42 that displays sleep quality, a display area 43 that displays differential data, and a display area 44 that displays recommended actions.

[0086] In the example in Figure 14, display area 42 shows that the predicted sleep quality for the day is "49" and the optimal sleep quality is "76". Display area 43 shows a time waveform that overlays the time waveform of the activity data corresponding to the optimal sleep quality with the time waveform of the activity data for the day. Display area 44 shows recommended actions. In the example in Figure 14, the time waveform displayed in display area 43 shows that the recommended action at point (1) is "walk one station during your commute" and the recommended action at point (2) is "stretch every hour while working at your desk".

[0087] <Fourth Embodiment>

[0088] Next, a fourth embodiment will be described. Note that the configuration of the prediction system 10 is the same as in the first embodiment, so its description will be omitted.

[0089] In the fourth embodiment, the acquisition unit 71 acquires activity data from multiple different users. The storage unit 76 stores the acquired activity data, the calculated fragmentation degree, the calculated sleep quality, the behavioral data, and the users in the activity database 23C.

[0090] The output unit 74 then outputs difference data relating to the difference between the standard deviation of the degree of fragmentation, the time waveform of the activity data of another user whose behavioral pattern based on behavioral data is most similar and whose sleep quality is the highest among other users, and the time waveform of the activity data acquired by the acquisition unit 71.

[0091] Next, the prediction process performed on the CPU 21A of server 20 will be explained with reference to the flowchart shown in Figure 15. Note that steps identical to those shown in Figure 13 are denoted by the same reference numerals and their explanations are omitted.

[0092] The prediction process in Figure 15 includes steps S207 and S208, compared to the prediction process in Figure 13.

[0093] In step S207, CPU 21A calculates a standard score for the fragmentation degree calculated in step S201, based on the fragmentation degree calculated for all users stored in the activity database 23C.

[0094] In step S208, the CPU 21A refers to the activity database 23C and identifies other users whose behavioral patterns based on their behavioral data are most similar and who have the highest quality sleep. It then transmits to the mobile terminal device 40 the difference data between the time waveform of the identified other user's activity data and the time waveform of the activity data for that day, along with the fragmentation score calculated in step S207.

[0095] As a result, the mobile terminal device 40 displays the deviation score of the degree of fragmentation and the difference data between that and the time waveform of the activity data of other users whose behavioral patterns are most similar and whose sleep quality is the highest. This allows user 30 to easily compare their sleep quality with that of other users.

[0096] It should be noted that the above embodiments are merely illustrative examples illustrating the configuration of the disclosed technology. The disclosed technology is not limited to the specific forms described above, and various modifications are possible within the scope of its technical concept.

[0097] For example, in the above embodiment, the case in which the prediction program 23A is pre-installed in the storage unit 23 was described, but the prediction program 23A may also be stored in a non-volatile non-transitory recording medium or distributed via a network and installed on the server 20 as appropriate.

[0098] Examples of non-volatile, non-transitional recording media include CD-ROMs (Compact Disc Read Only Memory), magneto-optical disks, HDDs (Hard Disk Drives), DVD-ROMs (Digital Versatile Disc Read Only Memory), flash memory, and memory cards. [Explanation of symbols]

[0099] 10 Prediction Systems 20 servers 21 Controllers 21E System Bus 22 Communications Department 23 Memory section 23A Prediction Program 23B Predictive Model 23C Activity Database 30 users 40 Mobile terminal devices 50 Networks 60 Activity meter 71 Acquisition Department 72 Fragmentation degree calculation unit 73 Prediction Section 74 Output section 75 Sleep Quality Calculation Unit 76 Storage Unit 77 Model Creation Department

Claims

1. An acquisition unit that acquires activity data relating to bodily activity while awake, A fragmentation degree calculation unit calculates a fragmentation degree that represents the degree of fluctuation of the activity based on the activity data, A prediction unit that predicts the sleep quality corresponding to the calculated fragmentation level using a prediction model that represents the correspondence between the degree of fragmentation and the quality of sleep, An output unit that outputs the predicted sleep quality, Equipped with, The aforementioned fragmentation degree is calculated by the following formula, where IV is the fragmentation degree, N is the total number of activity data points during wakefulness, and Xi is the activity data at each time point i (i = 1, 2, ..., N) when the activity data was measured. Prediction device.

2. A quality calculation unit calculates the quality of sleep based on the activity data, A storage unit that stores the acquired activity data, the calculated fragmentation degree, and the calculated sleep quality in association with each other, A model creation unit creates a predictive model that takes the degree of fragmentation as input and the quality of sleep as output, based on the accumulated activity data, the degree of fragmentation, and the quality of sleep. The prediction device according to claim 1, comprising:

3. The output unit outputs difference data relating to the difference between the time waveform of the activity data corresponding to the optimal sleep quality with the highest sleep quality among the sleep quality stored by the storage unit, and the time waveform of the activity data acquired by the acquisition unit. The prediction device according to claim 2.

4. The acquisition unit acquires behavioral data corresponding to the activity data, The storage unit stores the acquired activity data, the calculated fragmentation degree, the calculated sleep quality, and the behavioral data in association with each other. The output unit outputs the action data corresponding to the difference. The prediction device according to claim 3.

5. The acquisition unit acquires the activity data of multiple different users, The storage unit stores the acquired activity data, the calculated fragmentation degree, the calculated sleep quality, the behavioral data, and the user in association with each other. The output unit outputs difference data relating to the difference between the standard deviation of the fragmentation degree, the time waveform of the activity data of another user whose behavioral pattern based on the behavioral data is most similar and whose sleep quality is the highest among other users, and the time waveform of the activity data acquired by the acquisition unit. The prediction device according to claim 4.

6. Computers We acquire activity data regarding physical activity while awake. Based on the activity data, a degree of fragmentation representing the degree of variation in the activity is calculated. Using a predictive model that represents the correspondence between the degree of fragmentation and sleep quality, the sleep quality corresponding to the calculated degree of fragmentation is predicted. Outputs the predicted sleep quality. A prediction method that performs a process including the following: The aforementioned fragmentation degree is calculated by the following formula, where IV is the fragmentation degree, N is the total number of activity data points during wakefulness, and Xi is the activity data at each time point i (i = 1, 2, ..., N) when the activity data was measured. Prediction method.

7. On the computer, We acquire activity data regarding physical activity while awake. Based on the activity data, a degree of fragmentation representing the degree of variation in the activity is calculated. Using a predictive model that represents the correspondence between the degree of fragmentation and sleep quality, the sleep quality corresponding to the calculated degree of fragmentation is predicted. Outputs the predicted sleep quality. A prediction program that causes the program to perform a process that includes the following: The aforementioned fragmentation degree is calculated by the following formula, where IV is the fragmentation degree, N is the total number of activity data points during wakefulness, and Xi is the activity data at each time point i (i = 1, 2, ..., N) when the activity data was measured. Prediction program.

Images