Pleasant sleep system, inference device, and learning device

Abstract

To provide a pleasant sleep system capable of improving the accuracy of inferring a bathing condition or a bedtime optimal to respective users, an inference device, and a learning device.SOLUTION: A pleasant sleep system includes: a first inference acquisition part for acquiring a schedule bedtime, attribute information, activity information, and weather information about a user; and a first inference part for inferring a recommended bathing condition from the schedule bedtime, the attribute information, the activity information, and the weather information about the user acquired by the first inference acquisition part by using a first learned model for infer the recommended bathing condition for improving a sleep state of the user on the basis of the scheduled bedtime, the attribute information, the activity information, and the weather information.SELECTED DRAWING: Figure 2

Description

Technical Field

[0001] The present disclosure relates to a sound sleep system, an inference device, and a learning device.

Background Art

[0002] Patent Document 1 discloses a bathing system. In this bathing system, it is possible to notify an appropriate bathing time among bathing conditions by estimating an increase value of deep body temperature and prompting bathing so as to improve the sleep state of the user.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] In the bathing system described in Patent Document 1, appropriate bathing conditions are generated based on the pre-bedtime from the bathing time to the scheduled bedtime. However, there are various factors that affect a person's bedtime other than the pre-bedtime. And the influence of these factors may vary for each user.

[0005] The present disclosure has been made to solve the above problems. An object of the present disclosure is to provide a sound sleep system, an inference device, and a learning device capable of improving the accuracy of inferring optimal bathing conditions or bedtime for each user.

Means for Solving the Problems

[0006] The sleep aid system according to the present disclosure includes a first inference acquisition unit that acquires a scheduled bedtime, attribute information, activity information, and weather information regarding a user, and based on the scheduled bedtime, the attribute information, the activity information, and the weather information, uses a learned first model for inferring recommended bathing conditions for improving the sleep state of the user to infer the recommended bathing conditions from the scheduled bedtime, the attribute information, the activity information, and the weather information regarding the user acquired by the first inference acquisition unit.

[0007] The inference device according to the present disclosure includes an inference acquisition unit that acquires a scheduled bedtime, attribute information, activity information, and weather information regarding a user, and based on the scheduled bedtime, the attribute information, the activity information, and the weather information, uses a learned first model for inferring recommended bathing conditions for improving the sleep state of the user to infer the recommended bathing conditions from the scheduled bedtime, the attribute information, the activity information, and the weather information regarding the user acquired by the inference acquisition unit.

[0008] The learning device according to the present disclosure includes a learning acquisition unit that acquires learning data including an actual bedtime, actual bathing conditions, sleep information indicating a sleep state, attribute information, activity information, and weather information regarding a user, and a generation unit that uses the learning data to generate a learned first model for inferring recommended bathing conditions for improving the sleep state based on the scheduled bedtime, the attribute information, the activity information, and the weather information of the user.

[0009] The inference device according to the present disclosure includes an inference acquisition unit that acquires actual bathing conditions, attribute information, activity information, and weather information regarding a user, and based on the actual bathing conditions, the attribute information, the activity information, and the weather information, uses a learned second model for inferring a recommended bedtime for improving the sleep state of the user to infer the recommended bedtime from the actual bathing conditions, the attribute information, the activity information, and the weather information regarding the user acquired by the inference acquisition unit.

[0010] The learning device according to the present disclosure includes a learning acquisition unit that acquires learning data including actual bedtime, actual bathing conditions, sleep information indicating a sleep state, attribute information, activity information, and weather information regarding a user, and uses the learning data to improve the sleep state of the user based on the actual bathing conditions, the attribute information, the activity information, and the weather information of the user. And a generation unit that generates a learned second model for inferring a recommended bedtime.

Effect of the Invention

[0011] According to the present disclosure, based on various information regarding a user, a bathing condition or a bedtime that improves the sleep state is inferred. Therefore, it is possible to improve the accuracy of inferring the optimal bathing condition or bedtime for each user.

Brief Description of the Drawings

[0012]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Figure 16

Figure 17

Figure 18

Figure 19

Figure 20

Embodiments for Carrying Out the Invention

[0013] Embodiments for carrying out the present disclosure will be described with reference to the accompanying drawings. In each figure, the same or corresponding parts are denoted by the same reference numerals. Redundant descriptions of such parts will be simplified or omitted as appropriate.

[0014] Embodiment 1. FIG. 1 is a configuration diagram of a building to which the sound sleep system in Embodiment 1 is applied.

[0015] The sound sleep system 1 shown in FIG. 1 is used by a user U who lives in a building 1000. The building 1000 includes a living room 1001, a bathroom 1002, and a bedroom 1003. The user U wears a wearable device W on their wrist or the like and carries a mobile terminal S. The mobile terminal S is a device capable of displaying and inputting information, such as a smartphone. A dedicated application used in the sound sleep system 1 is installed on the mobile terminal S.

[0016] The sound sleep system 1 proposes bathing conditions and bedtime so that the user U can get good sleep. The sound sleep system 1 includes a water supply device 2, an air conditioner 3a, a bathroom air conditioner 3b, a display means 4, an input means 5, an activity amount acquisition means 6, a condition acquisition means 7, a sleep acquisition means 8, an environment acquisition means 9, an inference device 10, and a learning device 20.

[0017] The water supply device 2 fills the bathtub B with hot water in the bathroom 1002. The water supply device 2 includes a water heater 2a, a controller 2b, a bathroom remote control 2c, and an out-of-bathroom remote control 2d. The water heater 2a generates hot water. The controller 2b controls the operation of the water heater 2a to adjust the temperature and amount of hot water filled in the bathtub B. The controller 2b can detect that a person is taking a bath in the hot water by a sensor provided in the water heater 2a. The bathroom remote control 2c is provided on the wall surface of the bathroom 1002. The out-of-bathroom remote control 2d is provided on the wall surface of the living room 1001. The bathroom remote control 2c and the out-of-bathroom remote control 2d have buttons and accept settings for control by the controller 2b. The bathroom remote control 2c and the out-of-bathroom remote control 2d have displays and can display various types of information.

[0018] The air conditioner 3a is provided in the bedroom 1003. The air conditioner 3a sucks in the air in the bedroom 1003, generates conditioned air with adjusted temperature and humidity, and blows it into the bedroom 1003. The air conditioner 3a has a bedroom biosensor 3c capable of acquiring the biological information of a person sleeping in the bedroom 1003.

[0019] The bathroom air conditioner 3b is installed in the bathroom 1002. The bathroom air conditioner 3b blows out conditioned air with adjusted temperature and humidity into the bathroom 1002.

[0020] The display means 4 can display various kinds of information. For example, the display means 4 is the mobile terminal S. Specifically, the mobile terminal S functions as the display means 4 by displaying a proposed user interface (hereinafter also referred to as "proposed UI") realized by a dedicated application on the screen. Further, the display means 4 may include a bathroom remote controller 2c, an external bathroom remote controller 2d, a wearable device W, a personal computer possessed by the user U, and the like.

[0021] The input means 5 receives input of information from a person. For example, the input means 5 is the mobile terminal S. Specifically, the mobile terminal S functions as the input means 5 by receiving information through the proposed UI. Further, the input means 5 may include a bathroom remote controller 2c, an external bathroom remote controller 2d, a wearable device W, a personal computer, and the like.

[0022] The input means 5 receives input of conditions under which the water supply device 2 operates, bathing conditions preferred by the user U, the scheduled bedtime of the user U, attribute information of the user U, and the like. The attribute information includes at least one of age, gender, height, weight, residential area, and the preference of the bathing conditions of the user U.

[0023] The activity amount acquisition means 6 can estimate the activity amount, which is an index value of the amount of physical activity performed by the user U during a day, and acquire it as activity information. The activity amount acquisition means 6 may be means for acquiring the estimated activity amount as activity information. The activity information is information indicating the activity amount and the type of activity of the user U on that day. The activity amount increases as the time a person exercises is longer and as the activity level of the person is higher. The activity amount increases as the heart rate or pulse rate of a person per unit time is higher. For example, the activity amount acquisition means 6 is a mobile terminal S. Specifically, the mobile terminal S constantly collects exercise data such as the number of steps a person has walked, the distance traveled, and biological information such as pulse waves. A dedicated application of the mobile terminal S acquires the collected exercise data and estimates the activity amount. Also, the mobile terminal S may have the daily schedule of the user U registered, or the user U may register the results of exercise in the schedule or the like. For example, the schedule may register the time and frequency of a person's desk work, the presence or absence of going out, and the like. The schedule may have the activity level for each day of the week pre-entered. The dedicated application may acquire the information of the schedule as exercise data and estimate the activity amount. Also, the activity amount acquisition means 6 may include a wearable device W.

[0024] The condition acquisition means 7 acquires the actual bathing conditions that are the conditions under which the user U takes a bath. The bathing conditions include at least the water temperature of the hot water in the bathtub B and the bathing time. Further, the bathing conditions may include at least one of the water volume, bathing time, and bathroom temperature. The water volume includes information such as full-body bath and half-body bath. The bathing time is the time when the user U starts taking a bath. The bathing time is the time from when the user U immerses in the hot water until getting out of the bathtub B. Note that the bathing time may be the total time of the time the user U was immersed in the hot water from the start to the end of taking a bath. For example, the condition acquisition means 7 is the controller 2b. Note that the condition acquisition means 7 may include a biosensor provided in the dressing room of the bathroom 1002, a mobile terminal S, a wearable device W, a bathroom remote control 2c, a remote control 2d outside the bathroom, etc. In this case, the mobile terminal S may receive an input of the conditions for taking a bath from the user U via the proposed UI. Also, each device may realize the condition acquisition means 7 as a whole, such as the biosensor acquiring the bathing start time and the controller 2b acquiring the water temperature.

[0025] The sleep acquisition means 8 acquires sleep information indicating the state of the user U during sleep. The sleep information includes at least information related to body movements such as the number of body movements. Further, the sleep information may include sleep states such as the pulse rate, respiratory rate, brain activity value, sleep time, sleep latency, middle awakening time, middle awakening frequency, deep sleep time, proportion of deep sleep time, REM sleep time, proportion of REM sleep time, etc. of the person during sleep. For example, the sleep time, sleep latency, middle awakening time, middle awakening frequency, deep sleep time, proportion of deep sleep time, REM sleep time, proportion of REM sleep time are calculated from the pulse rate, number of body movements, respiratory rate, etc. which are the biological information of the person during sleep. Also, the sleep information may include a sleep score which is an index value indicating the effectiveness of sleep for one night. The more effective and comfortable the sleep the user U gets, the higher the sleep score. The sleep score can be calculated from the pulse rate, number of body movements, respiratory rate, brain activity value, sleep time, sleep latency, middle awakening time, middle awakening frequency, deep sleep time, proportion of deep sleep time, REM sleep time, proportion of REM sleep time, etc. of the person during sleep.

[0026] The sleep acquisition means 8 has a sensor capable of detecting the state of the user U during sleep. For example, the sleep acquisition means 8 is the bedroom biosensor 3c of the air conditioner 3a. The sleep acquisition means 8 may include a wearable device W. Further, the sleep acquisition means 8 may include a mobile terminal S, a biosensor (not shown) provided in the bedroom 1003, etc.

[0027] The environment acquisition means 9 acquires and outputs weather information such as temperature and atmospheric pressure. For example, the environment acquisition means 9 is an external server that transmits weather information.

[0028] The inference device 10 performs a first inference process by inputting various information of the day into the first model, and infers recommended bathing conditions for the user U. Thereafter, the inference device 10 performs a second inference process by inputting various information of the day into the second model, and infers the recommended bedtime for the user U. For example, the inference device 10 is provided in a building different from the building 1000. The inference device 10 can communicate with each device of the sound sleep system 1. Note that the inference device 10 may be the mobile terminal S of the user U, or may be a computer provided in a building different from the building 1000.

[0029] For example, the learning device 20 is provided in a building different from the building 1000. The learning device 20 can communicate with each device of the sound sleep system 1. The learning device 20 acquires and accumulates various information regarding the user U on a daily basis in association with each other. The learning device 20 performs a first learning process using the accumulated information to generate a learned first model. The learning device 20 performs a second learning process using the accumulated information to generate a learned second model.

[0030] When it gets dark, user U operates the mobile terminal S, which is the input means 5, to input the planned bedtime and infer the recommended bathing conditions. At this time, the mobile terminal S as the inference device 10 uses the first model based on the activity information acquired by the activity acquisition means 6 on this day, the weather information acquired by the environment acquisition means 9, the planned bedtime, etc., to infer the recommended bathing conditions that can improve the sleep state, that is, enhance the quality of sleep. The mobile terminal S, as the display means 4, displays the inferred recommended bathing conditions. The recommended bathing conditions include a combination of hot water temperature, hot water volume, bathing time, bathing duration, and bathroom temperature. The water supply device 2 fills the bath with hot water at the hot water temperature and hot water volume included in the recommended bathing conditions so as to be in time for the bathing time included in the recommended bathing conditions. User U takes a bath based on the recommended bathing conditions.

[0031] After user U finishes taking a bath, the condition acquisition means 7 acquires the actual bathing conditions, which are the conditions under which user U actually took a bath. Then, the mobile terminal S, as the inference device 10, uses the second model based on the activity information, weather information, actual bathing conditions, etc. of this day to infer the recommended bedtime that can improve the sleep state, that is, enhance the quality of sleep. For example, user U goes to bed referring to the recommended bedtime. At this time, the actual bedtime is acquired by the sleep acquisition means 8. The sleep information of user U during sleep is acquired by the sleep acquisition means 8. Various information such as the activity information of this day, the weather information of this day, the actual bathing conditions of this day, the actual bedtime of this day, and the sleep information of this day are stored in the learning device 20 in an associated manner.

[0032] In this way, since the first model and the second model are generated based on the information about user U, the bathing conditions or bathing time that can most improve the sleep quality for user U can be inferred.

[0033] Next, the sound sleep system 1 will be described in more detail with reference to FIGS. 2 and 3. FIG. 2 is a functional block diagram of the sound sleep system according to Embodiment 1. FIG. 3 is a diagram showing an overview of the neural network model. FIG. 4 is a diagram showing a general relationship between seasons and suitable bathing conditions. FIG. 5 is a diagram showing a general relationship between age and suitable bathing conditions.

[0034] FIG. 2 shows, as an example, the functional configuration when the inference device 10 is not the mobile terminal S but the mobile terminal S is the display means 4 and the input means 5. In this example, for example, the inference device 10 is provided in the same building as the learning device 20.

[0035] The inference device 10 includes a first inference section 11 and a second inference section 12 as functions. The first inference section 11 is a function for executing the first inference process. The first inference section 11 includes, as functions, a first inference acquisition unit 13, a first model storage unit 14, and a first inference unit 15. The second inference section 12 is a function for executing the second inference process. The second inference section 12 includes, as functions, a second inference acquisition unit 16, a second model storage unit 17, and a second inference unit 18.

[0036] The first inference acquisition unit 13 acquires, as first inference data, the scheduled bedtime of the user U on a certain day, the attribute information of the user U, the one-day activity information, and the one-day weather information in association with each other. The scheduled bedtime and the attribute information are acquired from the input means 5. The activity information is acquired from the activity amount acquisition means 6. The weather information is acquired from the environment acquisition means 9.

[0037] The first model storage unit 14 stores the learned first model. The first model is a model for inferring recommended bathing conditions for improving the sleep state based on the scheduled bedtime, the attribute information, the one-day activity information, and the one-day weather information.

[0038] The first inference unit 15 infers recommended bathing conditions from the first model stored in the first model storage unit 14 and the first inference data acquired by the first inference acquisition unit 13. At this time, the first inference unit 15 may infer a plurality of recommended bathing conditions in which at least one of the hot water temperature and the amount of hot water is different as the recommended bathing conditions.

[0039] The second inference acquisition unit 16 acquires, as second inference data, the actual bathing conditions of the user U on a certain day, the attribute information of the user U, the one-day activity information, and the one-day weather information in association with each other. The actual bathing conditions are acquired from the condition acquisition means 7. The attribute information is acquired from the input means 5. The activity information is acquired from the activity amount acquisition means 6. The weather information is acquired from the environment acquisition means 9.

[0040] The second model storage unit 17 stores a learned second model. The second model is a model for inferring a recommended bedtime for improving the sleep state based on the actual bathing conditions, the attribute information, the one-day activity information, and the one-day weather information.

[0041] The second inference unit 18 infers a recommended bedtime from the second model stored in the second model storage unit 17 and the second inference data acquired by the second inference acquisition unit 16.

[0042] The display control unit 19 causes the display means 4 to display the recommended bathing conditions inferred by the first inference unit 15 and the recommended bedtime inferred by the second inference unit 18. When the time for bathing included in the recommended bathing conditions is reached or the time is before the specified time from the bathing time, the display control unit 19 causes the display means 4 to display a notification prompting bathing. When the recommended bedtime is reached or the time is before the specified time from the bedtime, the display control unit 19 causes the display means 4 to display a notification prompting bedtime.

[0043] The learning device 20 includes a learning acquisition unit 21, a storage unit 22, a first generation unit 23, and a second generation unit 24 as functions.

[0044] The learning acquisition unit 21 acquires, as learning data, the actual bedtime, the actual bathing conditions, the attribute information, the one-day activity information, and the one-day weather information on a certain day, in association with each other. The learning acquisition unit 21 causes the storage unit 22 to store the acquired learning data for a certain day. A plurality of pieces of learning data for each day are stored in the storage unit 22.

[0045] The first generation unit 23 executes a first learning process to generate a first model by machine learning. Specifically, the first generation unit 23 uses the learning data stored in the storage unit 22 and uses a learning algorithm related to machine learning to generate a learned first model. The first generation unit 23 may update the first model stored in the first model storage unit 14 to the latest first model.

[0046] The second generation unit 24 executes a second learning process to generate a second model by machine learning. Specifically, the second generation unit 24 uses the learning data stored in the storage unit 22 and uses a learning algorithm related to machine learning to generate a learned second model. The second generation unit 24 may update the second model stored in the second model storage unit 17 to the latest second model.

[0047] Known algorithms such as supervised learning, unsupervised learning, and reinforcement learning can be applied as the learning algorithms adopted by the first generation unit 23 and the second generation unit 24. As an example, the first generation unit 23 and the second generation unit 24 learn, according to a neural network model, recommended bathing conditions for improving the sleep state or recommended bedtime for improving the sleep state by so-called supervised learning. Here, supervised learning refers to a method in which learning data, which is a pair of input and result (label), is given to the first generation unit 23 and the second generation unit 24, so that certain features in the learning data are learned, and the result is inferred from the input.

[0048] Figure 3 shows a three-layer neural network as an overview of the neural network model. The neural network is composed of an input layer X1-X3 consisting of a plurality of neurons, an intermediate layer Y1-Y2 consisting of a plurality of neurons, and an output layer Z1-Z3 consisting of a plurality of neurons. The intermediate layer is also referred to as a hidden layer and may be one layer or two or more layers. In the case of a three-layer neural network with one intermediate layer, when a plurality of inputs are input to the input layer X1-X3, the values are multiplied by weights W1 (w11-w16) and then input to the intermediate layer Y1-Y2. The output from the intermediate layer Y1-Y2, which is the result of the input, is multiplied by weights W2 (w21-w26) and output from the output layer Z1-Z3. The final output result changes depending on the values of weights W1 and W2.

[0049] In this example of the first learning process, among the learning data stored in the storage unit 22, those with sleep information better than a specified level, for example, those with a sleep score higher than the sleep threshold, are selected and used for learning. The neural network in the first generation unit 23 learns the "bathing conditions" that improve the sleep state through so-called supervised learning according to the learning data created based on the combination of the "bathing conditions" of the actual results, the "bedtime", the "activity information", the "attribute information", and the "weather information" acquired by the learning acquisition unit 21 and stored in the storage unit 22. The "bathing conditions" are the conditions output as the "recommended bathing conditions". That is, the neural network adjusts the weights W1 and W2 so that the result output from the output layer approaches the "bathing conditions", which are the correct data (results), by inputting the planned "bedtime", "activity information", "attribute information", and "weather information" to the input layer, and thus performs learning. The first generation unit 23 generates and outputs a learned model by executing the above learning. When the "attribute information" includes preferences for bathing conditions, a learned model that can infer recommended bathing conditions close to the preferred conditions is generated.

[0050] Note that the first learning process may be performed without selecting learning data. In this case, even when using the reinforcement learning method, the first generation unit 23 may perform learning by adjusting weights so that the sleep information output together with the "bathing condition" becomes better, for example, so that the sleep score becomes higher.

[0051] In this example of the second learning process, among the learning data stored in the storage unit 22, those with sleep information better than a specified level, for example, those with a sleep score higher than the sleep threshold, are selected and used for learning. The neural network in the second generation unit 24 learns the "bedtime" for improving the sleep state by so-called supervised learning according to the learning data created based on the combination of the actual "bathing condition", the actual "bedtime", "activity information", "attribute information", and "weather information" acquired by the learning acquisition unit 21 and stored in the storage unit 22. The "bedtime" is the condition output as the "recommended bedtime". That is, the neural network adjusts the weights W1 and W2 so that the result output from the output layer approaches the "bedtime", which is the correct data (result), by inputting the actual "bathing condition", "activity information", "attribute information", and "weather information" into the input layer, and thus performs learning. The second generation unit 24 generates and outputs a learned model by executing the above learning.

[0052] Note that the second learning process may be performed without selecting learning data. In this case, even when using the reinforcement learning method, the second generation unit 24 may perform learning by adjusting weights so that the sleep information output together with the "bedtime" becomes better, for example, so that the sleep score becomes higher.

[0053] In addition, in this example, an example in which supervised learning is applied to the learning algorithm in the first generation unit 23 and the second generation unit 24 has been described, but the learning algorithm is not limited to this. For example, methods such as reinforcement learning, unsupervised learning, and semi-supervised learning may be applied to the learning algorithm. Further, as the learning algorithm, deep learning that learns the extraction of the feature amount itself may be applied. Further, in the first generation unit 23 and the second generation unit 24, machine learning may be executed according to other known methods, for example, genetic programming, functional logic programming, support vector machine, and the like.

[0054] In the first inference process, on the basis of calculating bathing conditions that improve the quality of sleep, after appropriately raising the change in the deep body temperature of a person by bathing, conditions for falling asleep when the deep body temperature drops are considered. In this case, the scheduled bedtime, activity information, attribute information, and weather information are related to the recommended bathing conditions as follows.

[0055] The shorter the time from the bathing time to the scheduled bedtime, the more difficult it is for the deep body temperature to rise, that is, the bathing conditions such that the hot water temperature becomes lower and the bathing time becomes shorter can be inferred. Since the deep body temperature does not rise much, it is possible to suppress as much as possible the awakening of brain activity due to the rise in the deep body temperature, and an effect of making it easier to fall asleep immediately after bathing can be expected.

[0056] On the other hand, the longer the time from the bathing time to the scheduled bedtime, the easier it is for the deep body temperature to rise, that is, the bathing conditions that satisfy at least one of the hot water temperature becoming higher, the amount of hot water being larger, and the bathing time becoming longer can be inferred. Generally, the greater the rise in the deep body temperature, the greater the subsequent decrease in the deep body temperature. By falling asleep at the timing when the deep body temperature drops significantly, the quality of sleep can be improved. In the case of this condition where the time to wait for the drop in the deep body temperature can be set, by setting the bathing conditions for raising the deep body temperature significantly, an effect of promoting deep sleep and improving the quality of sleep can be expected.

[0057] It is assumed that the greater the activity level during the day, the higher the level of physical fatigue. In this case, in order to relieve the accumulated fatigue, it can be inferred that the conditions for a full-body bath, i.e., the bath temperature is high and the amount of hot water is large. On the other hand, when the activity level during the day is low but the activity type involves a lot of desk work, it is assumed that the mental fatigue is high and the autonomic nerves are easily disturbed. In this case, in order to relieve mental fatigue and calm the autonomic nerves, it can be inferred that the conditions for a half-body bath, i.e., the bath temperature is low, the amount of hot water is small, and the bathing time is long.

[0058] When the seasons in the weather information are different, the assumed average temperature of the day is different. Fig. 4 shows a graph depicting the recommended relationship between the bath temperature, which is a bathing condition, and the bathing time for each season. The horizontal axis represents the bath temperature. The vertical axis represents the bathing time. The solid-line graph k is the relationship between the recommended bath temperature and the bathing time in winter. The dashed-line graph m is the relationship between the recommended bath temperature and the bathing time in summer. The dotted-and-dash-line graph l is the relationship between the recommended bath temperature and the bathing time in the intermediate period between winter and summer.

[0059] As shown in Fig. 4, in any season, in order for the deep body temperature to rise, the lower the bath temperature, the longer the bathing time is recommended. And when comparing the graphs k, l, and m, at the same bath temperature, the colder the season, the longer the bathing time is recommended. Similarly, at the same bathing time, the colder the season, the higher the bath temperature is recommended.

[0060] Therefore, it can be inferred that the lower the average temperature of the season, the higher the bath temperature and the longer the bathing time. Also, regarding the temperature and weather of the day, when the user is likely to feel cold or the weather is like rainy days, it can be inferred that the bathing conditions are such that the deep body temperature rises more significantly.

[0061] When the age in the attribute information is different, generally the ease of change in deep body temperature is different. Specifically, the higher the age, the less likely the deep body temperature is to change. Also, the higher the age, the more physically weakened, and thus the greater the burden on the body due to bathing.

[0062] Figure 5 shows a graph depicting the recommended relationship between the bath water temperature and the bathing time, which are bathing conditions, for each age and each season. The relationships between the solid line, the broken line, and the dashed-dotted line and the seasons are the same as those in the graph shown in Figure 4. Also, for each season, the line thicknesses are different such that they become thicker in the order of a, b, and c. Graph c represents the recommended bathing conditions for people older than those targeted by graph b. Graph b represents the recommended bathing conditions for people older than those targeted by graph a. As also shown in Figure 5, even in the same season, it can be inferred that the bathing conditions are such that the higher the age, the lower the bath water temperature and the shorter the bathing time. In this case, the condition is such that the deep body temperature drops below a certain level when going to sleep. Furthermore, the condition is such that the physical burden is reduced. Also, although not shown, for the same reason, it can be inferred that the bathing conditions are such that the higher the age, the less the amount of bath water.

[0063] In the second inference process, when calculating the bedtime that improves the quality of sleep, the condition is considered such that one falls asleep when the deep body temperature drops after rising due to bathing. In this case, the actual bathing conditions, activity information, attribute information, and weather information are related to the recommended bedtime as follows.

[0064] From the actual bathing conditions, the amount of increase in the deep body temperature can be estimated. Also, even if the specific amount of increase is not estimated, it can be estimated whether the bathing conditions were such that the deep body temperature increased significantly. In this case, the greater the bathing conditions with a large amount of increase in the deep body temperature, the later the bedtime can be inferred. This is to ensure sufficient time for the deep body temperature to drop after rising.

[0065] The more active one is during the day, the higher the physical fatigue level, so an earlier bedtime can be inferred. This is to relieve physical fatigue earlier.

[0066] The older one is, the earlier the recommended bedtime. That is, the younger one is, the later bedtime can be inferred. This is to ensure sufficient time for the deep body temperature to drop steadily.

[0067] The worse the weather or the further the temperature is from the median of the season, the earlier bedtime can be inferred. This is because the fatigue level on the body caused by the weather is significant.

[0068] Next, with reference to FIGS. 6 to 12, examples of the information displayed on the display means 4 and the information input by the input means 5 will be described. FIGS. 6 to 12 are diagrams showing examples of the proposed UI displayed in the sound sleep system according to the first embodiment.

[0069] FIG. 6 is a proposed UI for inputting settings such as attribute information. The setting of attribute information is performed in advance. The proposed UI accepts the scheduled bedtime bath for each day or day of the week.

[0070] The attribute information includes age and gender as user information. Although not shown, the user information may include the residential area of the user. Also, in the proposed UI, the input of the activity level, which is activity information, is accepted as user information.

[0071] Furthermore, the attribute information may include the user's preferred bathing conditions. That is, as the "basic bathing conditions", the input of a combination of the hot water temperature, bathing time, amount of hot water, and the presence or absence of bathroom heating, which the user prefers, is accepted.

[0072] As shown in FIG. 7, when the time becomes earlier than the specified time from the bath time included in the recommended bathing conditions, the display means 4 displays a notification prompting bathing. At this time, as shown in FIG. 8, the proposed UI displays the current time, the already input scheduled bedtime, and the recommended bathing conditions as "recommended bathing", respectively.

[0073] In the proposed UI, as the input means 5, it is possible to input the hot water supply conditions of the water heater 2. When "manual setting" is selected, the screen transitions to a screen where the user inputs the hot water supply conditions including the hot water temperature and the amount of hot water. When "recommended hot water supply" is selected, a command to supply hot water at the hot water temperature and the amount of hot water included in the recommended bathing conditions inferred by the inference device 10 is transmitted from the mobile terminal S, which is the input means 5, to the water heater 2. In this case, the water heater 2 supplies hot water at the hot water temperature and the amount of hot water included in the recommended bathing conditions.

[0074] FIGS. 9 and 10 show examples different from FIG. 8 of the proposed UI that accepts input of the recommended bathing conditions.

[0075] The inference device 10 may infer a plurality of recommended bathing conditions. FIGS. 11 and 12 show examples in which a plurality of recommended bathing conditions are displayed on the proposed UI. In the proposed UI of FIG. 11, "[recommended 1] hot water supply" and "[recommended 2] hot water supply" are presented. In the proposed UI of FIG. 12, "[full body bath] hot water supply" and "[half body bath] hot water supply" are presented. The user selects one of the presented multiple conditions. In this case, the water heater 2 supplies hot water based on the selected bathing conditions.

[0076] Next, the learning process and the inference process will be described with reference to FIGS. 13 and 14, respectively. FIG. 13 is a flowchart showing an outline of the first learning process performed by the learning device in the first embodiment. FIG. 14 is a flowchart showing an outline of the first inference process performed by the inference device in the first embodiment.

[0077] The first learning process shown in the flowchart of FIG. 13 starts at a prescribed cycle such as once a day. Note that the operation of the flowchart may start at any timing such as when there is a separate instruction by the input means 5 or when learning data for one day is added.

[0078] In step S001, the learning acquisition unit 21 acquires learning data from the storage unit 22. Thereafter, in step S002, the first generation unit 23 executes the first learning process to generate a first model. Thereafter, in step S003, the first generation unit 23 stores the generated first model in the first model storage unit 14. Thereafter, the operation of the flowchart ends.

[0079] Note that similar processing may be performed not with the first learning process but with the second learning process. In this case, the second generation unit 24 generates a second model by the second learning process and stores the generated second model in the second model storage unit 17.

[0080] The first inference process shown in the flowchart of FIG. 14 starts when an operation to display recommended bathing conditions is input by the input means 5. Note that the first inference process, which is the operation of the flowchart, may start at any timing such as when there is a separate instruction by the input means 5 or when a prescribed time set every day arrives.

[0081] In step S101, the first inference acquisition unit 13 acquires first inference data. Thereafter, in step S102, the first inference unit 15 infers recommended bathing conditions based on the first model and the first inference data as the first inference process. Thereafter, in step S103, the first inference unit 15 outputs information on the inferred recommended bathing conditions. Thereafter, in step S104, the display control unit 19 causes the recommended bathing conditions output to be displayed on the proposed UI of the display means 4. Thereafter, the operation of the flowchart ends.

[0082] Note that the same process may be performed in the second inference process instead of the first inference process. In this case, when the hot water supply device 2 or the like detects that the user has finished bathing, the second inference process, which is the operation of the flowchart, is started. Note that the second inference process may be started at any timing such as when there is a separate instruction by the input means 5 or when the specified time set every day is reached.

[0083] The second inference acquisition unit 16 acquires second inference data. The second inference unit 18 infers and outputs a recommended bedtime based on the second model and the second inference data as the second inference process. The display control unit 19 causes the recommended bedtime output to be displayed on the proposal UI.

[0084] Next, with reference to FIGS. 15 to 17, the operations performed by the sound sleep system 1 after the first inference process is performed will be described. FIG. 15 is a flowchart showing a first example of the operations performed by the sound sleep system in Embodiment 1. FIG. 16 is a flowchart showing a second example of the operations performed by the sound sleep system in Embodiment 1. FIG. 17 is a flowchart showing a third example of the operations performed by the sound sleep system in Embodiment 1.

[0085] In the first example of FIG. 15, an example is shown in which a hot water filling command performed by the hot water supply device 2 is manually input by the proposal UI.

[0086] Steps S101 to S103 in the flowchart are the same as the flowchart in FIG. 14.

[0087] After step S103, the operation of step S201 is performed. In step S201, the display control unit 19 determines whether or not the time has reached the preparation time, which is a specified time before the bath time included in the recommended bath conditions. If the time has not reached the preparation time in step S201, the operation of step S201 is repeated.

[0088] In step S201, when it is the preparation time, the operation in step S202 is performed. In step S202, the display control unit 19 causes the display means 4 to display a notification to prompt bathing.

[0089] Thereafter, in step S203, the display control unit 19 determines whether an operation to start hot water supply to the hot water supply device 2 has been input to the proposed UI.

[0090] In step S203, when an operation to start hot water supply is input, the operation in step S204 is performed. In step S204, the hot water supply device 2 starts hot water filling based on the input conditions, in this case, the recommended bathing conditions. Thereafter, the operation of the flowchart ends.

[0091] In step S203, when an operation to start hot water supply is not input, the operation in step S025 is performed. In step S205, the display control unit 19 determines whether a prescribed waiting time has elapsed since the notification in step S202.

[0092] In step S205, when the waiting time has not elapsed, the operations after step S203 are performed. In step S205, when the waiting time has elapsed, the operations after step S101 are performed. That is, the first inference process is performed again.

[0093] In the second example of FIG. 16, it is an example in which the command for hot water filling performed by the hot water supply device 2 is automatically performed according to the inferred recommended bathing conditions.

[0094] Steps S101 to S103 in the flowchart are the same as the flowchart in FIG. 14.

[0095] After step S103, the operation in step S301 is performed. In step S301, the hot water supply device 2 starts hot water filling based on the conditions output in step S103, that is, the recommended bathing conditions.

[0096] After that, in step S302, the bathroom air conditioner 3b starts operating so as to reach the bathroom temperature included in the conditions output in step S103, that is, the recommended bathing conditions.

[0097] After that, the operation of the flowchart ends.

[0098] Note that either one of the operations in step S301 and step S302 may be omitted.

[0099] In the third example of FIG. 17, after the operation of the first example, it is an example in which the second inference process is performed.

[0100] Steps S101 to S103 and steps S201 to S205 in the flowchart are the same as the flowchart in FIG. 15.

[0101] After step S204, the operation of step S401 is performed. In step S401, the second inference unit 18 determines whether the bathing of the user has been detected. The bathing of the user may be detected by detecting the user in the bathroom, the change in the water level of the hot water in the bathtub B, etc. If the bathing of the user is not detected in step S401, the operation of step S401 is repeated.

[0102] If the bathing of the user is detected in step S401, the operation of step S402 is performed. In step S402, the second inference unit 18 determines whether the bathing of the user has ended, that is, whether the user has finished bathing. If the user has not finished bathing in step S402, the operation of step S402 is repeated.

[0103] If the user has finished bathing in step S402, the operation of step S403 is performed. In step S403, the inference device 10 and the learning device 20 acquire the actual bathing time, hot water temperature, hot water volume, bathroom temperature, etc. as the actual bathing conditions.

[0104] After that, in step S404, the second inference acquisition unit 16 of the inference device 10 acquires second inference data from each device. After that, in step S405, the second inference unit 18 infers a recommended bedtime as a second inference process based on the second model and the second inference data. After that, in step S406, the second inference unit 18 outputs information on the inferred recommended bedtime. After that, in step S407, the display control unit 19 causes the recommended bedtime output to be displayed on the proposed UI of the display means 4. After that, the operation of the flowchart ends.

[0105] According to the first embodiment described above, the sound sleep system 1 includes a first inference acquisition unit 13 and a first inference unit 15. The first inference unit 15 infers recommended bathing conditions based on first inference data including a scheduled bedtime, attribute information, activity information, and weather information, and a first model. The first model is a model corresponding to the user. In this way, bathing conditions that improve the sleep state are inferred based on various types of information regarding the user. As a result, the accuracy of inferring optimal bathing conditions for each user can be improved. Even when the inference device 10 includes the first inference acquisition unit 13 and the first inference unit 15, similarly, the accuracy of inferring optimal bathing conditions for each user can be improved.

[0106] In addition, the sound sleep system 1 and the inference device 10 include a second inference acquisition unit 16 and a second inference unit 18. The second inference unit 18 infers a recommended bedtime using second inference data including actual bathing conditions, attribute information, activity information, and weather information, and a second model. The second model is a model corresponding to the user. In this way, a bedtime that improves the sleep state is inferred based on various types of information regarding the user. As a result, the accuracy of inferring an optimal bedtime for each user can be improved.

[0107] Furthermore, after inferring recommended bathing conditions and after the user has finished bathing, a recommended bedtime may be inferred based on the actual bathing conditions. Even if recommended bathing conditions that are preferable for the user are inferred, the user may not take a bath as recommended. In this case, by inferring the recommended bedtime based on the actual bathing conditions, the bedtime is inferred under conditions that are more in line with the actual situation. As a result, the accuracy of inferring the optimal bedtime for each user can be improved.

[0108] Also, the sound sleep system 1 and the learning device 20 include a learning acquisition unit 21 and a first generation unit 23. The first generation unit 23 performs machine learning based on learning data including the actual bedtime, actual bathing conditions, sleep information, attribute information, activity information, and weather information regarding the user, thereby generating a learned first model. Based on the first model learned based on various types of information specific to the user, recommended bathing conditions corresponding to the user can be inferred. Therefore, the accuracy of inferring the optimal bathing conditions for each user can be improved.

[0109] Also, the first generation unit 23 generates a learned first model by performing machine learning using a neural network model. When optimizing the parameters of a known model formula with a plurality of types of information included in the learning data as in the prior art, when adjusting the parameters with one type of information, another type of information may have an impact. In such a machine learning method of adjusting the parameters of a known model formula, even if based on a plurality of types of information, the accuracy may not be improved well. As shown in the present embodiment, by performing machine learning using a neural network model, the first model is generated after taking into account the mutual relationship of each piece of information. As a result, the accuracy of inferring the optimal bathing conditions for each user can be further improved.

[0110] In addition, the sound sleep system 1 and the learning device 20 include a learning acquisition unit 21 and a second generation unit 24. The second generation unit 24 performs machine learning based on learning data including the actual bedtime, actual bathing conditions, sleep information, attribute information, activity information, and weather information of the user, thereby generating a learned second model. Based on the second model learned based on various information specific to the user, a recommended bedtime corresponding to the user can be inferred. Therefore, it is possible to improve the accuracy of inferring the optimal bedtime for each user.

[0111] In addition, the second generation unit 24 generates a learned second model by performing machine learning using a neural network model. In this case, similar to the case where the first generation unit 23 performs machine learning using a neural network model, the second model is generated while taking into account the mutual relationship of each information. As a result, it is possible to further improve the accuracy of inferring the optimal bathing conditions for each user.

[0112] In addition, the sleep information at least includes information regarding the body movement of the user. The state of body movement is an index that accurately represents the sleep state of the user such as the sleep stage. Since the sleep information includes information regarding body movement, it is possible to further improve the accuracy of inferring the optimal bathing conditions and bedtime for each user.

[0113] In addition, the recommended bathing conditions at least include the hot water temperature of the hot water filled in the bathtub B and the bathing time. In order to improve the sleep state, the change in deep body temperature due to bathing is an important factor. The hot water temperature is an important factor particularly for raising the deep body temperature. The bathing time is an important factor particularly when determining how much the deep body temperature is raised in relation to the scheduled bedtime. Since these factors are included in the bathing conditions, it is possible to further improve the accuracy of inferring the optimal bathing conditions for each user.

[0114] In addition, the attribute information includes at least one of the user's age, gender, place of residence, preferences for bathing conditions, height, and weight. Age, gender, height, and weight are important factors in estimating the ease of change in deep body temperature. The place of residence is an important factor in estimating the physical load of the day and the deep body temperature before bathing. Preferences for bathing conditions are an important factor in determining the recommended bathing conditions. Even if recommended bathing conditions that do not match the user's preferences are presented, the user may not take a bath under the presented conditions. By including at least one of these factors in the attribute information, the accuracy of inferring the optimal bathing conditions or bedtime for each user can be further improved.

[0115] In addition, the activity information includes at least the user's activity level and type for the day. In particular, the activity level is estimated from information obtained by the activity amount acquisition means 6, activity types, activity levels registered in the proposal UI, etc., and the user's schedule. The activity level and its type are related to the degree of physical fatigue. Therefore, for each user, the accuracy of inferring the optimal bathing conditions or bedtime for that day can be further improved.

[0116] In addition, the sound sleep system 1 includes a display means 4. The recommended bathing conditions or recommended bedtime are displayed on the display means 4. Therefore, the user can easily know the recommended bathing conditions or recommended bedtime.

[0117] In addition, the display means 4 notifies that it is necessary to take a bath at a time before the specified time from the bath time included in the recommended bathing conditions. Therefore, the user can take a bath under the recommended bathing conditions more reliably.

[0118] In addition, the sound sleep system 1 further includes a hot water supply device 2. When the recommended bathing conditions are inferred, the hot water supply device 2 fills the bathtub B with hot water at the hot water temperature included in the recommended bathing conditions. Therefore, the user can take a bath under the recommended bathing conditions without performing a separate operation.

[0119] Note that the learning data, the first inference data, and the second inference data may further include, as one of the input data, the amount and time of drinking alcohol and the amount and time of meals. For example, such information is acquired by each device when the user inputs it to the proposal UI via the input means 5. In this way, by increasing the input data, the sound sleep system 1 can infer the recommended bathing conditions and the recommended bedtime in consideration of the daily differences in the user's physical condition. As a result, the inference accuracy of the conditions for improving the quality of sleep can be improved.

[0120] Note that the learning data, the first inference data, and the second inference data may further include, as one of the input data, biological information including the user's pulse wave before bathing. For example, the biological information is acquired by a biological sensor provided in the dressing room, a biological sensor provided in the wearable device W, or the like. By increasing the input data, the sound sleep system 1 can infer the recommended bathing conditions and the recommended bedtime in consideration of the daily differences in the user's physical condition.

[0121] Note that the inference device 10 may estimate the scheduled bedtime based on the accumulated actual bedtime. The inference device 10 may perform the first inference process using the estimated scheduled bedtime. For this reason, the trouble of the user inputting the scheduled bedtime can be saved. In addition, a scheduled bedtime closer to the actual situation than the scheduled bedtime received from the user can be used for inference. As a result, the inference accuracy of the bathing conditions for improving the quality of sleep can be improved.

[0122] Note that the first model storage unit 14 and the second model storage unit 17 may be provided in a device different from the inference device 10 such as the learning device 20.

[0123] Note that the first learning process and the second learning process may be performed using a known model formula including a plurality of parameters in which the actual "bathing conditions", the actual "bedtime", "activity information", "attribute information", and "weather information" are respectively reflected. In this case, each parameter included in the model formula is changed so as to improve the quality of sleep. Specifically, for example, as learning data, among the plurality of pieces of information stored in the storage unit 22, those including sleep information in which each index included in the sleep information has a good result are selected. Based on the learning data, each parameter is determined.

[0124] For example, when the amount of information stored in the storage unit 22 is small, in the machine learning based on the neural network model as shown in the first embodiment, the output accuracy of the learned model may be low. Thus, when the amount of information stored in the storage unit 22 is small, the learning device 20 executes machine learning based on a known model formula to generate the first model and the second model, and after the amount of information stored in the storage unit 22 becomes sufficiently large, the first model and the second model may be generated by machine learning based on the neural network model.

[0125] Embodiment 2. FIG. 18 is a configuration diagram of a building to which the sound sleep system in the second embodiment is applied. Note that the same reference numerals are given to the same or corresponding parts as those in the first embodiment, and the description of those parts is omitted.

[0126] In the second embodiment, the learning data, the first inference data, and the second inference data may further include, as input data, at least one of a warming degree which is an index value related to a person's deep body temperature and a body burden degree which is an index value of the burden on the body due to bathing.

[0127] The warming degree is calculated based on at least one of the estimated value of the deep body temperature of a person after bathing and the difference between the estimated values of the deep body temperature of the person before and after bathing. The greater the absolute value of the deep body temperature, or the greater the increase in the deep body temperature, the greater the warming degree. The physical burden degree is calculated from the pulse rate or respiratory rate of the user. For example, when the pulse rate is more than a specified ratio with respect to a specified reference value, the physical burden degree increases. For example, when the respiratory rate is more than a specified ratio with respect to a specified reference value, the physical burden degree increases.

[0128] In Embodiment 2, the sound sleep system 1 further includes a biological sensor 30. The biological sensor 30 non - contact detects the pulse wave of a person as biological information. For example, the biological sensor 30 is provided in the dressing room. Also, the biological sensor 30 is provided in the wearable device W.

[0129] For example, the inference device 10 or the learning device 20 calculates the warming degree and the physical burden degree based on the biological information detected by the biological sensor 30. The inference device 10 includes the warming degree and the physical burden degree in the first inference data and the second inference data. The inference device 10 performs the first inference process and the second inference process using such data. The learning device 20 includes the warming degree and the physical burden degree in the learning data. The learning device 20 performs the first learning process and the second learning process using such data.

[0130] According to Embodiment 2 described above, the learning data, the first inference data, and the second inference data further include the warming degree and the physical burden degree. The recommended bathing conditions are inferred for the purpose of appropriately changing the deep body temperature of a person. Also, the recommended bathing conditions are inferred so as not to increase the physical fatigue degree of a person too much. The recommended bedtime is inferred based on the change in the deep body temperature of a person and the physical fatigue degree of the person. In this way, by adding at least one of the index value of the deep body temperature due to bathing and the physical burden degree to the input data, the sound sleep system 1 can infer the recommended bathing conditions and the recommended bedtime in consideration of the state of the user due to bathing.

[0131] Embodiment 3 FIG. 19 is a functional block diagram of the sound sleep system according to Embodiment 3. The same or corresponding parts as those in Embodiments 1 and 2 are denoted by the same reference numerals, and the description thereof is omitted.

[0132] In Embodiment 3, the sound sleep system 1 further includes a specifying means 40 for specifying a user from a plurality of persons. The specifying means 40 includes at least one of the wearable device W, a biological sensor provided in the bathroom 1002 or the dressing room, a weighing scale having a function of specifying an individual provided in the dressing room, and the portable terminal S. In the wearable device W, an individual can be specified by biometric authentication. The portable terminal S can specify an individual by receiving an input from the proposed UI.

[0133] In the sound sleep system 1, the inference device 10 and the learning device 20 collect each piece of information in a state where an individual is specified. The inference device 10 acquires first inference data and second inference data associated with an individual. The learning device 20 acquires learning data associated with an individual and stores it in association with the individual. The first model and the second model are models corresponding to an individual associated with the learning data.

[0134] According to Embodiment 3 described above, the sound sleep system 1 further includes a specifying means 40. For this reason, the sound sleep system 1 can also infer and propose bathing conditions and bedtime suitable for the user to a user living in a building where a plurality of residents live.

[0135] Next, an example of the hardware constituting the inference device 10 will be described with reference to FIG. 20. FIG. 20 is a hardware configuration diagram of the inference device of the sound sleep system according to Embodiments 1 to 3.

[0136] Each function of the inference device 10 can be realized by a processing circuit. For example, the processing circuit includes at least one processor 100a and at least one memory 100b. For example, the processing circuit includes at least one dedicated hardware 200.

[0137] When the processing circuit includes at least one processor 100a and at least one memory 100b, each function of the inference device 10 is realized by software, firmware, or a combination of software and firmware. At least one of the software and the firmware is described as a program. At least one of the software and the firmware is stored in at least one memory 100b. The at least one processor 100a realizes each function of the inference device 10 by reading and executing the program stored in the at least one memory 100b. The at least one processor 100a is also referred to as a central processing unit, a processing device, an arithmetic device, a microprocessor, a microcomputer, or a DSP. For example, the at least one memory 100b is a non-volatile or volatile semiconductor memory such as a RAM, a ROM, a flash memory, an EPROM, or an EEPROM, a magnetic disk, a flexible disk, an optical disk, a compact disk, a mini disk, or a DVD.

[0138] When the processing circuit includes at least one dedicated hardware 200, the processing circuit is realized by, for example, a single circuit, a composite circuit, a programmed processor, a parallel-programmed processor, an ASIC, an FPGA, or a combination thereof. For example, each function of the inference device 10 is realized by the processing circuit respectively. For example, each function of the inference device 10 is realized by the processing circuit collectively.

[0139] Regarding each function of the inference device 10, part of it may be realized by dedicated hardware 200, and the other part may be realized by software or firmware. For example, the function of the first model storage unit 14 may be realized by a processing circuit as dedicated hardware 200, and for functions other than the function of the first model storage unit 14, at least one processor 100a may read and execute a program stored in at least one memory 100b to realize them.

[0140] In this way, the processing circuit as a computer realizes each function of the inference device 10 by hardware 200, software, firmware, or a combination thereof.

[0141] Although not shown in the figure, each function of the learning device 20 and the mobile terminal S is also realized by a processing circuit equivalent to the processing circuit that realizes each function of the inference device 10.

[0142] Note that at least a part of each function of the inference device 10 may be realized on a cloud server. In this case, the processing circuit is composed of a plurality of sub-circuits. The plurality of sub-processing circuits are respectively provided in a plurality of devices that constitute the cloud server. The plurality of devices that constitute the cloud server may each be provided in a different building.

[0143] Summarizing the above description, the possible configurations of the technology according to the present disclosure include the following configurations shown as appendices. (Appendix 1) A first inference acquisition unit that acquires the scheduled bedtime, attribute information, activity information, and weather information regarding the user; Using a learned first model for inferring recommended bathing conditions for improving the sleep state of the user based on the scheduled bedtime, the attribute information, the activity information, and the weather information, the first inference acquisition unit acquires the scheduled bedtime, the attribute information, the activity information, and the weather information regarding the user, and infers the recommended bathing conditions from them. A first inference unit; A sound sleep system comprising the same. (Appendix 2) A second inference acquisition unit that acquires the actual bathing conditions, the attribute information, the activity information, and the weather information regarding the user; When the recommended bathing conditions are inferred by the first inference unit and it is detected that the user has taken a bath, based on the actual bathing conditions, the attribute information, the activity information, and the weather information, a second inference unit that infers the recommended bedtime for improving the user's sleep state using a learned second model for inferring the recommended bedtime from the actual bathing conditions, the attribute information, the activity information, and the weather information regarding the user acquired by the second inference acquisition unit; The sound sleep system according to appended note 1, further comprising the above. (Appended note 3) Display means for displaying the recommended bedtime inferred by the second inference unit; The sound sleep system according to appended note 2, further comprising the above. (Appended note 4) A learning acquisition unit that acquires learning data including the actual bedtime, the actual bathing conditions, sleep information indicating the sleep state, the attribute information, the activity information, and the weather information regarding the user; A second generation unit that generates the learned second model for inferring the recommended bedtime based on the actual bathing conditions, the attribute information, the activity information, and the weather information of the user using the learning data; The sound sleep system according to appended note 2 or appended note 3, further comprising the above. (Appended note 5) The sound sleep system according to appended note 4, wherein the second generation unit generates the learned second model by performing machine learning using a neural network model. (Appended note 6) A learning acquisition unit that acquires learning data including the actual bedtime, the actual bathing conditions, sleep information indicating the sleep state, the attribute information, the activity information, and the weather information regarding the user; A first generation unit that generates the learned first model for inferring the recommended bathing conditions based on the scheduled bedtime, the attribute information, the activity information, and the weather information of the user using the learning data; The sound sleep system according to any one of appendices 1 to 5, further comprising (Appendix 7) The sound sleep system according to appendix 6, wherein the first generation unit generates the learned first model by performing machine learning using a neural network model. (Appendix 8) The sound sleep system according to any one of appendices 4 to 7, wherein the sleep information includes information related to the body movement of the user. (Appendix 9) The sound sleep system according to any one of claims 1 to appendix 7, wherein the recommended bath conditions include the water temperature of the hot water in the bathtub and the bath time. (Appendix 10) The sound sleep system according to any one of appendices 1 to 9, wherein the attribute information includes at least one of the user's age, gender, residential area, preference for bath conditions, height, and weight. (Appendix 11) The sound sleep system according to any one of appendices 1 to 10, wherein the activity information is information indicating the user's activity amount and type on that day. (Appendix 12) The sound sleep system according to appendix 11, wherein the activity amount in the activity information is information obtained by an activity amount acquisition means for detecting the user's activity amount, the registered activity type, the exercise intensity, and estimated from the user's schedule. (Appendix 13) Display means for displaying the recommended bath conditions inferred by the first inference unit The sound sleep system according to any one of appendices 1 to 12, further comprising (Appendix 14) The sound sleep system according to appendix 13, wherein the display means notifies that bathing is to be promoted at a time earlier than a specified time from the bath time included in the recommended bath conditions. (Appendix 15) A water supply device for filling the bathtub with hot water so that the water temperature is the water temperature included in the recommended bath conditions when the recommended bath conditions are inferred by the first inference unit The sound sleep system according to any one of Appendices 1 to 14, further comprising (Appendix 16) An inference acquisition unit that acquires a scheduled bedtime, attribute information, activity information, and weather information regarding a user, Based on the scheduled bedtime, the attribute information, the activity information, and the weather information, using a learned first model for inferring recommended bathing conditions for improving the sleep state of the user, the inference acquisition unit acquires the scheduled bedtime, the attribute information, the activity information, and the weather information regarding the user, and an inference unit that infers the recommended bathing conditions from the above, An inference device comprising (Appendix 17) A learning acquisition unit that acquires learning data including an actual bedtime, actual bathing conditions, sleep information indicating a sleep state, attribute information, activity information, and weather information regarding a user, Using the learning data, a generation unit that generates a learned first model for inferring recommended bathing conditions for improving the sleep state based on the scheduled bedtime, the attribute information, the activity information, and the weather information of the user, A learning device comprising (Appendix 18) An inference acquisition unit that acquires actual bathing conditions, attribute information, activity information, and weather information regarding a user, Based on the actual bathing conditions, the attribute information, the activity information, and the weather information, using a learned second model for inferring a recommended bedtime for improving the sleep state of the user, the inference acquisition unit acquires the actual bathing conditions, the attribute information, the activity information, and the weather information regarding the user, and an inference unit that infers the recommended bedtime from the above, An inference device comprising (Appendix 19) A learning acquisition unit that acquires learning data including an actual bedtime, actual bathing conditions, sleep information indicating a sleep state, attribute information, activity information, and weather information regarding a user, A generation unit that generates a learned second model for inferring a recommended bedtime for improving the user's sleep state based on the actual bathing conditions, the attribute information, the activity information, and the weather information of the user using the learning data. A learning device comprising the same.

Explanation of Signs

[0144] 1 Sound sleep system, 2 Hot water supply device, 2a Hot water supply machine, 2b Controller, 2c Bathroom remote control, 2d Remote control outside the bathroom, 3a Air conditioner, 3b Bathroom air conditioner, 3c Bedroom biosensor, 4 Display means, 5 Input means, 6 Activity amount acquisition means, 7 Condition acquisition means, 8 Sleep acquisition means, 9 Environment acquisition means, 10 Inference device, 11 First inference section, 12 Second inference section, 13 First inference acquisition section, 14 First model storage section, 15 First inference section, 16 Second inference acquisition section, 17 Second model storage section, 18 Second inference section, 19 Display control section, 20 Learning device, 21 Learning acquisition section, 22 Storage section, 23 First generation section, 24 Second generation section, 30 Biosensor, 40 Identification means, 100a Processor, 100b Memory, 200 Hardware, 1000 Building, 1001 Living room, 1002 Bathroom, 1003 Bedroom, S Mobile terminal, U User, W Wearable device

Claims

1. A first inference acquisition unit that acquires a scheduled bedtime, attribute information, activity information, and weather information regarding a user; Based on the scheduled bedtime, the attribute information, the activity information, and the weather information, using a learned first model for inferring recommended bath conditions for improving the sleep state of the user, the first inference acquisition unit acquires the scheduled bedtime, the attribute information, the activity information, and the weather information regarding the user, and a first inference unit that infers the recommended bath conditions from the above; A sound sleep system comprising the above.

2. A second inference acquisition unit that acquires actual bath conditions, attribute information, activity information, and weather information regarding the user; When the recommended bath conditions are inferred by the first inference unit and it is detected that the user has taken a bath, based on the actual bath conditions, the attribute information, the activity information, and the weather information, using a learned second model for inferring a recommended bedtime for improving the sleep state of the user, a second inference unit that infers the recommended bedtime from the actual bath conditions, the attribute information, the activity information, and the weather information regarding the user acquired by the second inference acquisition unit; The sound sleep system according to claim 1, further comprising the above.

3. Display means for displaying the recommended bedtime inferred by the second inference unit; The sound sleep system according to claim 2, further comprising the above.

4. A learning acquisition unit that acquires learning data including actual bedtime, actual bath conditions, sleep information indicating a sleep state, attribute information, activity information, and weather information regarding the user; A second generation unit that generates the learned second model for inferring the recommended bedtime based on the actual bath conditions, the attribute information, the activity information, and the weather information of the user using the learning data; The sound sleep system according to claim 2, further comprising the above.

5. The sound sleep system according to claim 4, wherein the second generation unit generates the learned second model by performing machine learning using a neural network model.

6. A learning acquisition unit that acquires learning data including actual bedtime, actual bath conditions, sleep information indicating a sleep state, attribute information, activity information, and weather information regarding the user; A first generation unit that generates the learned first model for inferring the recommended bathing conditions based on the scheduled bedtime, the attribute information, the activity information, and the weather information of the user using the learning data. The sound sleep system according to claim 1, further comprising the above.

7. The sound sleep system according to claim 6, wherein the first generation unit generates the learned first model by performing machine learning using a neural network model.

8. The sound sleep system according to any one of claims 4 to 7, wherein the sleep information includes information regarding the body movement of the user.

9. The sound sleep system according to any one of claims 1 to 7, wherein the recommended bathing conditions include the water temperature of the hot water filled in the bathtub and the bathing time.

10. The sound sleep system according to any one of claims 1 to 7, wherein the attribute information includes at least one of the user's age, gender, residential area, preference for bathing conditions, height, and weight.

11. The sound sleep system according to any one of claims 1 to 7, wherein the activity information is information indicating the amount and type of the user's activity on that day.

12. The sound sleep system according to claim 11, wherein the amount of activity in the activity information is information obtained by an activity amount acquisition means for detecting the user's activity amount, the registered activity type, the activity level, and is estimated from the user's schedule.

13. Display means for displaying the recommended bathing conditions inferred by the first inference unit. The sound sleep system according to any one of claims 1 to 7, further comprising the above.

14. The sound sleep system according to claim 13, wherein the display means notifies that bathing is to be promoted at a time earlier than a specified time from the bathing time included in the recommended bathing conditions.

15. A hot water supply device that fills the bathtub with hot water so as to reach the water temperature included in the recommended bathing conditions when the recommended bathing conditions are inferred by the first inference unit. The sound sleep system according to any one of claims 1 to 7, further comprising the above.

16. An inference acquisition unit that acquires the scheduled bedtime, attribute information, activity information, and weather information regarding the user. Based on the scheduled bedtime, the attribute information, the activity information, and the weather information, an inference unit that infers the recommended bathing conditions for improving the user's sleep state using a learned first model, from the scheduled bedtime, the attribute information, the activity information, and the weather information regarding the user acquired by the inference acquisition unit. An inference device comprising the same.

17. A learning acquisition unit that acquires learning data including the actual bedtime, the actual bathing conditions, sleep information indicating the sleep state, attribute information, activity information, and weather information regarding the user. A generation unit that generates a learned first model for inferring recommended bathing conditions for improving the sleep state based on the scheduled bedtime, the attribute information, the activity information, and the weather information of the user using the learning data. A learning device comprising the same.

18. An inference acquisition unit that acquires the actual bathing conditions, attribute information, activity information, and weather information regarding the user. An inference unit that infers the recommended bedtime for improving the user's sleep state using a learned second model based on the actual bathing conditions, the attribute information, the activity information, and the weather information, from the actual bathing conditions, the attribute information, the activity information, and the weather information regarding the user acquired by the inference acquisition unit. An inference device comprising the same.

19. A learning acquisition unit that acquires learning data including the actual bedtime, the actual bathing conditions, sleep information indicating the sleep state, attribute information, activity information, and weather information regarding the user. A generation unit that generates a learned second model for inferring the recommended bedtime for improving the user's sleep state based on the actual bathing conditions, the attribute information, the activity information, and the weather information of the user using the learning data. A learning device comprising the same.

Images