Bathing navigation program, bathing navigation method, bathing navigation device, and bathing navigation system

Abstract

To propose a technology that uses a simple predictive model to accurately predict when core body temperature has risen above a target level, and to recommend that bathers exit the bath before the timing deviates from a physiologically safe period. [Solution] The bathing navigation device that executes the bathing navigation program 30 acquires bathing environment information 31 and bather's physical information 33. Based on the bathing environment information 31 and bather's physical information 33, the bathing navigation device 2 extracts feature data x that is significant in the rise of core body temperature. The bathing navigation device 2 substitutes the extracted feature data x into a prediction model and calculates a prediction probability, which is the probability that the bather's core body temperature will rise to or above the target amount. If the calculated prediction probability is greater than or equal to the cutoff value cv, the device recommends that the bather get out of the bath.

Description

Technical Field

[0001] The technical field disclosed in this specification relates to a bath navigation program, a bath navigation method, a bath navigation device, and a bath navigation system that provide information related to bathing.

Background Art

[0002] In Japan, there has been a bathing culture of soaking in the hot water in a bathtub since ancient times. Bathing in hot water is performed for the purpose of warming the body, cleaning the body, and relaxing. On the other hand, the number of bathing accidents has been increasing year by year. Although warnings about bathing accidents have been issued through various media since autumn, there is no sign of a decrease. Therefore, there is a need for a technology that prompts a bather to get out of the hot water (get out of the hot water in the bathtub) at a physiologically safe timing.

[0003] As guidelines for proposing getting out of the hot water at a physiologically safe timing, two timings are known. The first timing is when the deep body temperature has risen by 0.5°C. The deep body temperature is the temperature at the center of the body (hereinafter referred to as the "core part"). The deep body temperature is measured, for example, by ear temperature, tympanic membrane temperature, sublingual temperature, esophageal temperature, rectal temperature, etc. The second timing is when the bather feels sweaty subjectively (when there is a subjective feeling of sweating). Although it has been found that these two timings occur at almost the same bathing time, it is difficult to constantly measure the deep body temperature during bathing, so it is difficult to represent the subjective feeling of sweating as an objective index. Therefore, a technology that predicts the deep body temperature itself and prompts getting out of the hot water has been proposed.

[0004] For example, the bathing navigation system described in Patent Document 1 estimates the bather's core temperature or the change in core temperature using a prediction formula, based on the bather's tympanic temperature (core temperature, an example of deep body temperature) at the start of bathing (when the bather begins to immerse themselves in the water) and bathing environment data that increases the bather's tympanic temperature (for example, bathing time (time spent immersed in the water), water temperature, submerged body surface area, non-submerged body surface area, and bathroom temperature). Based on the estimated core temperature or change in core temperature, the bathing navigation system prompts the bather to get out of the bath using a display or speaker before the change in core temperature reaches a predetermined amount.

[0005] For example, Patent Document 2 discloses a bathing device that estimates the core body temperature of a bather using a human body heat model based on the temperature of the water (water temperature) stored in the bathtub, and provides notification based on the estimated core body temperature.

[0006] For example, Patent Document 3 discloses a notification system that detects physical quantities of a bather related to core body temperature (e.g., skin temperature, heart rate, respiratory rate), calculates an estimated core body temperature based on the rate of change or amount of change of the detected physical quantities, the temperature and amount of water in the bathtub, and the bathing time, and provides notification based on the calculated estimated value.

[0007] For example, the core body temperature estimation device described in Patent Document 4 acquires water temperature data from a water temperature sensor installed in the bathtub. The core body temperature estimation device estimates the core body temperature based on the water temperature data, using the acquired water temperature data and a body model that calculates the amount of heat in a full-body bath. The core body temperature estimation device notifies an external party when the core body temperature exceeds a predetermined value. [Prior art documents] [Patent Documents]

[0008] [Patent Document 1] Patent No. 4522755 [Patent Document 2] Japanese Patent Publication No. 2023-115825 [Patent Document 3] Patent No. 7238596 [Patent Document 4] Japanese Patent Publication No. 2023-50891 [Overview of the project] [Problems that the invention aims to solve]

[0009] Core body temperature during bathing is influenced not only by bathing environment information such as water temperature, bathroom temperature, water level, and bathing time, but also by age-related physical decline and gender differences. Therefore, when estimating core body temperature, it is necessary to consider the bather's physical information, such as age, height, weight, and build.

[0010] The technologies described in Patent Documents 1 to 4 estimate core body temperature by considering biological information and bathing environment information, but they do not consider the bather's physical information. Therefore, for example, if the bathing environment information is the same, the prediction result could be the same whether the bather is young or elderly. In other words, there was a possibility that the prediction would completely lack consideration for elderly people who are at high risk of bathing accidents. As a result, for example, if the bather is elderly, the timing of prompting them to get out of the bathtub could deviate from a physiologically safe timing.

[0011] Furthermore, core body temperature changes due to the combined action of various conditions. The technologies described in Patent Documents 1, 3, and 4 predict core body temperature itself based on tympanic membrane temperature, heart rate, respiratory rate, bathing time, water temperature, submerged body surface area, non-submerged body surface area, and bathroom temperature. However, the types of data available are insufficient for accurate prediction of core body temperature, resulting in low prediction accuracy. The technology described in Patent Document 2 predicts core body temperature itself by considering the heat balance from the skin to the core. However, the prediction results of the technology described in Patent Document 2 are easily influenced by the bather's body temperature before bathing. Specifically, if a bather bathes multiple times between entering and leaving the bathroom, the effect of the previous bath remains in the bather's body temperature. For example, if the initial core body temperature at the start of the first bath is 37°C, the initial core body temperature at the start of the second bath will be higher than 37°C. The technology described in Patent Document 2 requires predicting the rise in core body temperature with high accuracy by using a complex prediction model at regular intervals from the start of bathing for each bathing session, and further determining whether the core body temperature has risen to a temperature rise threshold from the start of bathing. This places an excessive computational load on the device and requires high computational processing power. Therefore, there is room for improvement in technology that predicts changes in a bather's core body temperature and provides information to the bather. [Means for solving the problem]

[0012] The bathing navigation program developed to solve the above problems is (1) a bathing navigation program that can be executed by a bathing navigation device and provides information to a bather who is immersed in the bathwater, wherein the bathing navigation device includes a bathing environment information acquisition process that acquires bathing environment information which is information about the bather's bathing environment, a bather body information acquisition process that acquires bather body information which is information about the bather's body, and the bathing environment information acquired in the bathing environment information acquisition process and the bathing information acquired in the bather body information acquisition process The system is configured to perform a prediction probability calculation process, which involves extracting feature data significant to an increase in core body temperature based on the person's physical information, substituting the extracted feature data into a prediction model for calculating a prediction probability, which is the probability that the bather's core body temperature will exceed a target increase, and calculating the prediction probability; and, if the prediction probability calculated in the prediction probability calculation process is above a threshold, recommending that the bather get out of the bath; and, if the prediction probability calculated in the prediction probability calculation process is below the threshold, not recommending that the bather get out of the bath.

[0013] The bathing navigation program with the above configuration considers not only bathing environment information but also the bather's physical information to extract feature data, and then substitutes the extracted feature data into a prediction model to calculate the prediction probability. Therefore, the bathing navigation program has higher prediction accuracy compared to predictions that do not consider the bather's physical information. For example, even with the same bathing environment information, differences in age and gender will result in differences in the calculated prediction probability. In other words, it becomes possible to calculate a prediction probability that takes into account the elderly, who are at a higher risk of bathing accidents. The bathing navigation program compares the calculated prediction probability with a threshold and recommends that the bather get out of the bath. For example, whether the bather is elderly or young, male or female, it can recommend that the bather get out of the bath before deviating from a physiologically safe timing. Furthermore, the bathing navigation program with the above configuration calculates the prediction probability, which is the probability that the core body temperature will rise above a target amount, rather than the core body temperature itself, using a prediction model. Therefore, the prediction result is less affected by the bather's body temperature before entering the bath. Therefore, the prediction model used to calculate the prediction probability is simpler than a prediction model that considers the heat balance from the skin to the core, making it possible to predict the rise in core body temperature with high accuracy without placing an excessive computational load on the device or requiring high computing power. Thus, a bathing navigation program with the above configuration can accurately predict changes in core body temperature using a simple prediction model and can advise the bather to exit the bath before deviating from a physiologically safe timing. Furthermore, since the bathing navigation program with the above configuration calculates the prediction probability without using biometric information that fluctuates wildly due to changes in posture caused by body movement during bathing or hydrostatic pressure on the body, it can advise the bather to exit the bath from immediately after the start of bathing before deviating from a physiologically safe timing, and can monitor even short bathing sessions.

[0014] (2) In the bathing navigation program described in (1), it is preferable that the threshold is set to a value such that the recall rate and the accuracy rate are equal.

[0015] According to the bathing navigation program with the above configuration, it is expected that the timing of recommending that bathers get out of the bath will coincide with the timing when bathers become aware of sweating or feel the effects of heat. Therefore, bathers who are recommended to get out of the bath will be able to get the effects of heat from the water and get out of the bath.

[0016] (3) In the bathing navigation program described in (1) or (2), it is preferable that the bathing navigation device is instructed to perform a detection process to detect when the bather is discharging water, and a first warning process which, after recommending that the bather discharge water in the first recommendation process, warns the bather to discharge water if it is determined that a warning condition is met, and does not warn the bather to discharge water if it is determined that the warning condition is not met, wherein the warning condition is that a predetermined time has elapsed after the execution of the first recommendation process while the detection process has not detected the discharging of water.

[0017] According to the bathing navigation program configured as described above, after predicting that the core body temperature has risen above the target amount and recommending that the bather get out of the bath, a warning to get out of the bath is issued to the bather after a predetermined time has elapsed. This prevents bathers whose core body temperature has risen above the target amount from continuing to bathe at a time that is significantly outside of a physiologically safe period.

[0018] (4) In the bathing navigation program described in (3), it is preferable that the predetermined time differs depending on the water temperature of the bathwater into which the bather enters, as detected by the water temperature sensor of the bathing navigation device.

[0019] Core body temperature tends to rise more easily when the water temperature is high during bathing. The bathing navigation program with the above configuration will warn the bather to exit the bath at different times depending on the water temperature. Therefore, it can warn bathers to exit the bath when their core body temperature rises above the target amount at an appropriate time in accordance with the changes in core body temperature.

[0020] (5) In the bath navigation program according to any one of (1) to (4), when the prediction probability calculated in the prediction probability calculation process is not greater than or equal to the threshold value, and the hot water supply time of the bather exceeds the hot water supply warning time set before the timing deviating from the physiologically safe timing, it is preferable to execute a second warning process for warning the bather.

[0021] For example, when the bather is an elderly person whose deep body temperature is difficult to rise, the prediction probability may not reach the threshold value easily, and the bather may not be advised to supply hot water. Even in such a case, the bath navigation program with the above configuration gives a warning of hot water supply when the hot water supply time exceeds the hot water supply warning time. Therefore, according to the bath navigation program with the above configuration, it is possible to prompt the bather to supply hot water before significantly deviating from the physiologically safe timing while adapting to the physical functions of the bather.

[0022] (6) In the bath navigation program according to (5), when the hot water supply time exceeds the hot water supply advice time set at a timing before the hot water supply warning time, it is preferable to execute a second advice process for advising the bather to supply hot water.

[0023] Even when the bath navigation program with the above configuration predicts that the deep body temperature has not reached the target rise amount, by advising the bather to supply hot water before the hot water supply time exceeds the hot water supply warning time, the bather can be made aware that the hot water supply time is long, and safe hot water supply can be promoted for the bather.

[0024] (7) In the bath navigation program according to any one of (1) to (6), it is preferable to execute, in the bath navigation device, a storage process for storing bath-related information regarding hot water supply in the memory of the bath navigation device, and an update process for updating a prediction model using the bath-related information stored in the memory.

[0025] The bathing navigation program with the above configuration can accumulate bathing-related information in the memory, and by updating the prediction model using the accumulated bathing-related information, the prediction model can be refined and the prediction accuracy can be improved.

[0026] A system, method, device capable of realizing the functions of the bathing navigation program having the above configuration, and a storage medium storing the bathing navigation program are also novel and useful.

Effects of the Invention

[0027] According to the above technology, it is a technology for predicting the change in the deep body temperature of a bather and providing information to the bather, which can accurately predict that the deep body temperature has reached or exceeded the target increase amount using a simple prediction model, and can realize a technology for advising the bather to drain hot water before deviating from a physiologically safe timing.

Brief Description of the Drawings

[0028] [Figure 1] It is a diagram for explaining an example of use of the bathing navigation device according to the first embodiment. [Figure 2] It is a diagram for explaining the schematic configuration of the bathing navigation device according to the first embodiment. [Figure 3] It is a diagram for explaining an example of feature quantity data. [Figure 4] An example of a determination cross-tabulation table is shown. [Figure 5] An example of a scatter diagram of data for constructing a prediction model is shown. [Figure 6] It is a flowchart for explaining an example of the procedure of hot water drainage navigation processing. [Figure 7] It is a flowchart for explaining an example of the procedure of prediction probability calculation processing. [Figure 8] It is a diagram for explaining an example of a notification. [Figure 9] It is a flowchart for explaining an example of the procedure of re-warning processing. [Figure 10] It is a sequence diagram for explaining an example of update processing. [Figure 11] This diagram illustrates the schematic configuration of the bathing navigation device according to the second embodiment. [Figure 12] This is a flowchart illustrating an example of the procedure for hot water supply navigation processing. [Figure 13] This flowchart illustrates an example of the procedure for calculating predicted probabilities. [Figure 14] This is a diagram illustrating an example of feature data. [Figure 15] An example of a scatter plot of data used to build a predictive model is shown. [Figure 16] This figure shows the data used for verification. [Figure 17A] This figure shows time-series data of predicted probabilities using biometric information for each bathing activity. [Figure 17B] This figure shows time-series data of predicted probabilities for each bathing activity, without using biometric information. [Figure 18A] This figure shows time-series data of predicted probabilities using biometric information for each bathing activity. [Figure 18B] This figure shows time-series data of predicted probabilities for each bathing activity, without using biometric information. [Modes for carrying out the invention]

[0029] (First Embodiment) Hereinafter, the bathing navigation program, bathing navigation system, bathing navigation method, and bathing navigation device disclosed herein will be described with reference to the drawings. The first embodiment discloses a navigation system comprising a bathing navigation device capable of executing a bathing navigation program that provides information to a bather.

[0030] <Bath Navigation System> For example, as shown in Figure 1, the bathing navigation system 1 includes a bathing navigation device 2, which is communicatively connected to a wearable terminal 5, a smartphone 6, and a hot water supply and heating system 8. The bathing navigation device 2 is installed, for example, in a bathroom 7 of a detached house or an apartment building. A bathtub 71 is installed in the bathroom 7, and hot water 72 supplied from the hot water supply and heating system 8 is filled into the bathtub 71. The bathing navigation device 2 can provide bathing information to a bather M who is immersed in the hot water 72 of the bathtub 71.

[0031] In this specification, as shown in Figure 1, the state in which bather M is immersed in the water 72 of bathtub 71 is referred to as "entering the bath." The act of bather M leaving the water 72 of bathtub 71 is referred to as "exiting the bath." The time that bather M is immersed in the water 72 is referred to as "bathing time." All actions that bather M performs in the bathroom 7 are referred to as "bathing."

[0032] The bathing navigation device 2 consists of a main unit 21 and a sensor device 23 connected via a cable 22. The main unit 21 is mounted on the wall 73 of the bathroom 7. The sensor device 23 is mounted on the inner wall of the bathtub 71 at a position where the user is immersed in the hot water 72.

[0033] <Outline configuration of the bathing navigation device> Referring to Figure 2, the general configuration of the bathing navigation device 2 will be explained. The main unit 21 of the bathing navigation device 2 has a controller 10 which includes a CPU 11 and a memory 12. The main unit 21 has a user interface (hereinafter referred to as "user IF") 13, a communication interface (hereinafter referred to as "communication IF") 14, an audio output unit 15, a bathroom temperature sensor 16, and a timing unit 17, which are electrically connected to the controller 10. The CPU 11 is an example of a "controller". The memory 12 is an example of a "memory accessible by the bathing navigation device".

[0034] The sensor device 23 includes a water temperature sensor 28 for measuring water temperature and a water pressure sensor 29 for measuring water pressure. The sensor device 23 transmits water temperature data, including the water temperature value measured by the water temperature sensor 28, and water pressure data, including the water pressure value measured by the water pressure sensor 9, to the main unit 21 via cable 22.

[0035] The CPU 11 executes various processes according to the program read from memory 12 and based on user operations. Note that the controller 10 in Figure 2 is a general term for the hardware and software used to control the bathing navigation device 2, and does not necessarily represent a single piece of hardware actually present in the main unit 21.

[0036] Memory 12 stores various programs, including the bathing navigation program 30, and various data, including the bathing file 41. Memory 12 is also used as a temporary storage area.

[0037] The bathing navigation program 30 is a program that controls the operation of the bathing navigation device 2. The bathing navigation program 30 stores a prediction model that has been built in advance by machine learning. The prediction model is a prediction formula for calculating the prediction probability, which is the probability that the core body temperature of the bather M will rise to or above a target amount from the start of bathing (0 minutes into bathing). The target rise in this configuration is 0.4°C. The prediction model uses feature data that can be extracted as significant data for the rise in core body temperature, based on bathing environment information 31, which is information about the bathing environment, biometric information 32, which is information indicating the state of the bather M, and bather body information 33, which is information about the body of the bather M. If the prediction probability calculated by the prediction model is above a threshold, the bathing navigation program 30 recommends that the bather M get out of the bath; if the calculated prediction probability is below the threshold, it does not recommend that the bather M get out of the bath. This bathing navigation process will be described later. The bathing navigation program 30 also performs an update process to update the prediction model. The update process will be described later.

[0038] A bathing file 41 is created for each bath. For example, if bather M bathes twice between entering and leaving bathroom 7, a bathing file 41 is created for both the first and second baths. Also, if bather M and another bather bathe together, a bathing file 41 is created for both bather M's bath and the other bather's bath.

[0039] The bathing file 41 stores bathing-related information. In other words, the bathing navigation device 2 stores bathing logs in the bathing file 41. The bathing file 41 stores, for example, bathing environment information 31, biometric information 32, bather's physical information 33, intermediate variables 34, and processing details 35. The intermediate variables 34 store feature data extracted from the bathing environment information 31, biometric information 32, and bather's physical information 33. The processing details 35 store the processing details of the bathing navigation process. The processing details 35 store, for example, the predicted probability calculated by the prediction model and the content of the recommendations made to bather M. The bathing file 41 is used, for example, to build and update the prediction file. The bathing file 41 will be described later.

[0040] The user interface 13 includes hardware that displays a screen for informing the user of information and hardware that accepts user operations. In this embodiment, the user interface 13 combines a display 13a capable of displaying information, an indicator lamp 13b that shows the operating status of the bathing navigation device 2, and a stop button 13c for stopping the operation of the bathing navigation device 2. The user interface 13 may also be a touch panel equipped with display and input acceptance functions.

[0041] The communication IF 14 includes hardware for communicating with external devices such as a wearable terminal 5, a smartphone 6, and a hot water heating system 8. The communication method of the communication IF 14 may be wireless LAN communication such as Wi-Fi (registered trademark) or short-range wireless communication such as Bluetooth (registered trademark). The communication IF 14 also includes hardware for communicating with the sensor device 23.

[0042] The audio output unit 15 includes hardware for outputting sounds such as buzzer sounds and voice guidance. The bathroom temperature sensor 16 is a sensor that measures the temperature of the bathroom 7 (bathroom temperature). The timing unit 17 measures time.

[0043] <Wearable devices> In this embodiment, the wearable terminal 5 is a wristwatch-type communication terminal device. The wearable terminal 5 may also be a pendant-type, ring-type, or glasses-type communication terminal device, or a communication terminal device worn on the chest. The wearable terminal 5 may also be worn on the upper arm or ankle. The wearable terminal 5 can transmit the heart rate of the bather M, measured by the heart rate sensor 51, to the bathing navigation device 2 using the communication unit 52.

[0044] <Smartphone> The smartphone 6 has a touch panel 61 and a communication unit 62. The smartphone 6 stores a bathing management application program (hereinafter referred to as the "bathing management app") 63. The bathing management app 63 has a function to display a screen provided by the bathing navigation program 30 on the touch panel 61 when communication is established between the smartphone 6 and the bathing navigation device 2. For example, the bathing management app 63 is provided with a screen for inputting bather's physical information 33 by the bathing navigation program 30 and displays it on the touch panel 61. The bathing navigation program 30 can acquire bather's physical information 33 by receiving the bather's physical information 33 via the touch panel 61 and storing it in the memory 12. The bathing management app 63 may, for example, access the bathing file 41 and display the bathing log of bather M on the touch panel 61. The smartphone 6 may also be a tablet device.

[0045] <Building a predictive model> This section explains how to construct a predictive model. The predictive model will utilize a binary machine learning model, such as binary logistic regression analysis or a decision tree. The data used to construct the predictive model can be obtained from laboratory experiments or field surveys.

[0046] The training data will consist of binary values ​​for core body temperature change (e.g., core body temperature change of 0.5°C not reached / reached) or binary values ​​for subjective sweating sensation ("not sweating" / "sweating (more than just being sweaty)").

[0047] The feature data used is from a time prior to the prediction time when the prediction model is executed. For example, if the prediction model is to be executed 5 minutes after the start of bathing, the data used is from 1 minute before the prediction execution time (5 minutes), i.e., 4 minutes after the start of bathing. This is because changes in core body temperature are influenced by past bathing environments. By using historical data, it becomes possible to predict changes in core body temperature or subjective sweating sensation in real time.

[0048] The feature data includes data based on bathing environment information 31, data based on biometric information 32, and data based on bather physical information 33.

[0049] The data based on the bathing environment information 31 uses indicators that can be easily measured over time using sensors, etc., and indicators that can be easily obtained through questionnaires, etc., and further uses data generated from these indicators. Examples of data based on the bathing environment information 31 include water temperature, bathroom temperature, bathing time, water level at bathing (or percentage of body surface area submerged in water), posture (bent-knee sitting, legs stretched out sitting, cross-legged, etc.), amount of water used to fill the bathtub, number of baths (how many times it has been bathed), which person is bathing (cohabitants, etc.), bathroom size, bathtub size, bathroom construction, presence or absence of bathroom windows, floor on which the bathroom is located, housing insulation rating, housing construction, housing structure, housing type, floor of the housing (in the case of apartment buildings, multi-unit housing), presence or absence of windows, season, calendar information (year / month / day / day of the week / time), exhaled CO2 concentration, whether the bathroom heater is running, humidity, airflow, and environmental evaluation index (discomfort index, WBGT). For example, the water temperature and bathroom temperature one minute prior to the prediction execution time also fall under the category of data based on the bathing environment information 31.

[0050] The data based on biometric information 32 uses indicators that can be easily measured over time using wearable sensors, etc., to represent the state of the bather M, and the data generated from those indicators. The measured data includes, for example, intra-ear temperature, sublingual temperature, axillary temperature, heart rate, skin blood flow, local sweat volume, SpO2, pressure related to body movement, acceleration, and gyroscope. The amount of change from a given point, the rate of change from a given point, the difference sequence, statistics per unit time (e.g., average, standard deviation, standard error, maximum, minimum, median, coefficient of variation, etc.) and heart rate variability index generated from this time-series data are considered data based on biometric information 32.

[0051] The data based on the bather's physical information 33 includes indicators that affect the thermal effect during bathing, and data generated from those indicators, as the physical condition of bather M. Examples of data based on the bather's physical information 33 include age, sex, height, weight, body mass index (BMI) (calculated from height and weight using a known formula), body surface area (calculated from height and weight using a known formula), body fat percentage, skeletal muscle mass, bone mass, body fat mass, body age (e.g., measured values ​​using a body composition analyzer), and exercise frequency (questionnaire).

[0052] <Update to the prediction model> Generally, in building predictive models, the more data available, the broader the scope of application and the better the model can be built. However, the body's thermophysiological response during bathing, while affected by changes in lifestyle, can be considered unchanging as long as the body's structure remains the same. Therefore, data acquired for past research and development can be used to build predictive models. Furthermore, if new data is acquired through laboratory experiments or field surveys, the scope of application of feature data expands, and the prediction accuracy of the predictive model can be refined. Therefore, in this approach, bathing logs are accumulated, and a predictive model is built and updated based on these accumulated bathing logs.

[0053] As the amount of data used to build a predictive model increases, the feature data that contributes to the prediction may change. Therefore, updating a predictive model includes replacing the feature data.

[0054] <Examples of Predictive Models> The inventors constructed the predictive model shown in Equation 1 using machine learning with supervised binary logistic regression analysis. In Equation 1, p represents the predicted probability, a represents the coefficient, x represents the feature data, b represents a constant, and e represents the base of the natural logarithm (Napier's number).

[0055]

number

[0056] The data used to build the predictive model consisted of bathing data obtained from subject experiments where diverse physical information and bathing environment information were varied. As biological information 32, time-series data of tympanic membrane temperature and time-series data of heart rate were obtained.

[0057] The training data used was time-series data of tympanic membrane temperature acquired as biological information 32. The amount of change in tympanic membrane temperature starting from the start of bathing was calculated, and this was classified into two categories: (1) when the temperature rose to 0.4°C or higher, and (0) when it did not rise to 0.4°C or higher. The reason for using 0.4°C here is to allow the system to prepare for dispensing the bath until it reaches 0.5°C, and to ensure that the bath is dispensed when it reaches 0.5°C. The reason for using the change in tympanic membrane temperature as training data is that tympanic membrane temperature is said to reflect brain temperature, which is responsible for thermoregulation.

[0058] Based on previous research findings on hot springs, the feature data x used is shown in Figure 3. In Figure 3, "Importance" indicates the degree of influence on the change in tympanic membrane temperature to 0.4°C or higher. In this form, "Importance" is shown by the odds ratio. The "*" listed with the odds ratio indicates the statistical significance of the odds ratio. "*" indicates significance with a probability of less than 5%. "**" indicates significance with a probability of less than 1%. "***" indicates significance with a probability of less than 0.1%. As shown in Figure 3, the feature data x uses data that is significant with a probability of less than 5% for tympanic membrane temperature to 0.4°C or higher.

[0059] For example, as shown in Figure 3, "age group" serves as an indicator of the physical decline of bather M. Specifically, young people "under 20" have flexible cardiovascular systems, their body temperature rises quickly when bathing, and they return to normal quickly after bathing. Elderly people "60 and over" have increased cardiovascular hardening, and their body temperature rises slowly and returns slowly. Middle-aged people "30-50" are in the intermediate generation between young people and the elderly. Considering that physical decline affects core body temperature during bathing, "age group" can be classified into "under 20," "30-50," and "60 and over." In binary logistic regression analysis, it is not possible to use more than three data points as explanatory variables. Therefore, two dummy variables are generated for "age group," and these are used as feature data x. Specifically, the feature data x includes a "first age group" which is classified into two categories: "under 20" (0) and "30-50" (1), and a "second age group" which is classified into two categories: "under 20" (0) and "60 and over" (1).

[0060] "Gender" is an indicator of heat balance from the skin to the core. Men and women generally have characteristic physical differences, such as differences in body fat percentage and muscle mass. For example, because women have more body fat than men, heat is less easily transferred from the skin to the core, and their core body temperature does not rise as easily. Considering that "gender differences" thus affect the rise in core body temperature, "gender" is classified into two categories, "male (0)" and "female (1)," and is used in the feature data x.

[0061] "Height" serves as an indicator of the body surface area to which heat is transferred from the water. Specifically, taller people have a higher sitting height, which means that the ratio of the body surface area in contact with the water to the total surface area is smaller while bathing. When the proportion of body surface area submerged in water is small, core body temperature does not rise easily. Therefore, "height" itself is used as a feature data x.

[0062] Note that "age group," "gender," and "height" are examples of data based on bather physical information 33, and are easily obtainable through input by bather M.

[0063] "Heart rate 1 minute prior" is an indicator that shows the heart rate one minute before the predicted execution time. Heart rate baselines differ among bathers due to lifestyle habits, exercise habits, genetic factors, etc. Maximum heart rate can generally be calculated as 220 bpm - age (years), and as this formula shows, there is an upper limit to the number of heartbeats a person can have. Therefore, "heart rate 1 minute prior" itself is adopted as feature data x.

[0064] "Heart rate change 1 minute prior" is an index that shows the change in heart rate from the start of bathing to 1 minute before the predicted execution. Heart rate tends to increase from the time of elapsed time measurement, and "heart rate change 1 minute prior" reveals the responsiveness to the thermal effect of the bath from the start of bathing. Therefore, "heart rate change 1 minute prior" itself is adopted as feature data x.

[0065] "1-minute prior heart rate difference" is an indicator that represents the difference between the heart rate at the time of prediction and the heart rate one minute prior. For example, when the water temperature is high, the change in heart rate per unit time tends to be larger. In this way, "1-minute prior heart rate difference" reveals the responsiveness to the thermal effect of bathing. Therefore, "1-minute prior heart rate difference" itself is adopted as feature data x.

[0066] The "percentage change in heart rate from 3 minutes prior to 1 minute prior" is an indicator that shows the amount of change in heart rate from 3 minutes after the start of bathing to 1 minute before the predicted execution. The effects of body movement and changes in posture before and after bathing statistically settle down approximately 3 minutes after the start of bathing. Therefore, the percentage change in heart rate from the point when the effects of body movement and changes in posture have subsided reveals the responsiveness to the thermal effect of the bath. For this reason, the "percentage change in heart rate from 3 minutes prior to 1 minute prior" itself is adopted as the feature data x.

[0067] Note that "heart rate 1 minute ago," "change in heart rate 1 minute ago," "heart rate difference 1 minute ago," and "percentage change from 3 minutes ago" are examples of data based on biometric information 32 and can be easily calculated from time-series heart rate data measured by the wearable device 5.

[0068] Since the thermal effect of the hot water increases with the time spent in the bath, "elapsed time" is used as the feature data x. "Elapsed time (minutes)" is the elapsed time measured from the start of bathing (0 minutes into bathing).

[0069] The thermal effect of hot water is influenced by the immediate bathing environment, and increases with higher bathroom temperatures. Therefore, the "bathroom temperature one minute ago" itself is used as feature data x. Also, since the thermal effect of hot water increases with higher water temperatures, the "water temperature one minute ago" itself is used as feature data x. In this configuration, the bathroom temperature data and water temperature data from one minute ago are used as feature data x, but generally, bathing environment data is relatively stable compared to biometric information data, so it does not necessarily have to be from one minute ago.

[0070] Because differences in peripheral vascular reactivity occur depending on whether bather M is in a transitional phase of cold acclimatization, a transitional phase of heat acclimatization, or a heat acclimatization phase, "season" is classified into two binary categories, spring / summer (0) and autumn / winter (1), and adopted as feature data x.

[0071] Note that "elapsed time," "bathroom temperature 1 minute ago," "water temperature 1 minute ago," and "season" are examples of data based on the bathing environment information 31. "Elapsed time" can be easily measured by the timing unit 17. "Bathroom temperature 1 minute ago" and "water temperature 1 minute ago" can be easily measured by the bathroom temperature sensor 16 and the water temperature sensor 28. "Season" can be obtained from the day of bathing or the day of bathing.

[0072] As shown in Figure 3, the odds ratio for "feature data x" indicates how many times more influential the feature is when the influence of the comparison target on the change in tympanic membrane temperature is set to "1" in the case of qualitative data such as "age group," "gender," and "season," where there are comparison targets such as "under 20 (0)," "male (0)," and "spring / summer (0)." In the case of quantitative data such as "height," "heart rate 1 minute ago," "change in heart rate 1 minute ago," "heart rate difference 1 minute ago," "rate of change from 3 minutes ago heart rate 1 minute ago," "elapsed time," "bathroom temperature 1 minute ago," and "water temperature 1 minute ago," it indicates how many times more influential the feature is when the value increases by "1." Since the odds ratio indicates how many times more influential the feature is when the value changes by "1," features can be compared side by side. "Water temperature 1 minute ago" has the highest influence on the change in tympanic membrane temperature and is the biggest factor that increases the change in tympanic membrane temperature. "Coefficient a" is the value used in the aforementioned formula 1. The coefficient "a" is based on the influence of the feature data x on the change in tympanic membrane temperature caused by the hot water (in this form, the odds ratio).

[0073] Among the feature data x, the "age group" "60s and over" has an odds ratio less than "1" when the influence on the change in tympanic membrane temperature for "20s and under" is set to "1", thus contributing to the suppression of the increase in the change in tympanic membrane temperature. The "season group" "autumn / winter" has an odds ratio less than "1" when the influence on the change in tympanic membrane temperature for "spring / summer" is set to "1", thus contributing to the suppression of the increase in the change in tympanic membrane temperature. The "change in heart rate 1 minute prior" and "heart rate difference 1 minute prior" also have odds ratios less than "1", thus contributing to the suppression of the increase in the change in tympanic membrane temperature. These data are assigned a negative value to "coefficient a", which acts to lower the prediction probability p.

[0074] On the other hand, among the feature data x, the odds ratio for "age group" (30s-50s) is smaller than "1" when the influence on the change in tympanic temperature for "under 20s" is set to "1", the odds ratio for "female" (sex) is smaller than "1" when the influence on the change in tympanic temperature for "male" is set to "1", and the odds ratio for "autumn / winter" (season) is smaller than "1" when the influence on the change in tympanic temperature for "spring / summer" is set to "1", indicating that these factors contribute to suppressing the increase in the change in tympanic temperature. The odds ratios for "height", "heart rate 1 minute prior", "rate of change in heart rate 1 minute prior from 3 minutes prior", "elapsed time", "bathroom temperature 1 minute prior", "water temperature 1 minute prior", and "season" are larger than "1", indicating that these factors contribute to an increase in the change in tympanic temperature. These data are assigned a positive value to "coefficient a", which acts to increase the prediction probability p.

[0075] The constant b is set to -101.301.

[0076] <Predictive accuracy of the predictive model> Figure 4 shows the judgment cross-tabulation table. The estimated group of tympanic membrane temperature changes in Figure 4 is the value when the cutoff value of the prediction probability p is set to 22.6%. Region a shows the frequency of "actually measured tympanic membrane temperature change of 0.4°C or more, and predicted tympanic membrane temperature change of 0.4°C or more by the prediction model". Region b shows the frequency of "actually measured tympanic membrane temperature change of 0.4°C or more, and predicted tympanic membrane temperature change of less than 0.4°C by the prediction model". Region c shows the frequency of "actually measured tympanic membrane temperature change of less than 0.4°C, and predicted tympanic membrane temperature change of 0.4°C or more by the prediction model". Region d shows the frequency of "actually measured tympanic membrane temperature change of less than 0.4°C, and predicted tympanic membrane temperature change of less than 0.4°C by the prediction model".

[0077] In this embodiment, of the 2,846 bathing data points acquired to construct the prediction model of the embodiment, 516 belonged to area a, 59 to area b, 231 to area c, and 2,040 to area d.

[0078] A predictive model can be evaluated using three metrics: recall, precision, and accuracy. Recall is the percentage of cases where the tympanic membrane temperature (core body temperature) actually reached 0.4°C or higher, and the model also predicted that the tympanic membrane temperature would be 0.4°C or higher. In other words, recall can be calculated as (number of data points in region a) / (number of data points in region a + number of data points in region b). The recall for this model is 0.899.

[0079] Precision is the percentage of data where the actual increase in tympanic membrane temperature was less than 0.4°C, even though the predicted tympanic membrane temperature was 0.4°C or higher. In other words, precision can be calculated as (number of data in region a) / (number of data in region a + number of data in region c). The precision for this model is 0.691.

[0080] Accuracy is the percentage of the total data where the actual value matches the predicted value. Accuracy can be calculated as (number of data points in area a + number of data points in area d) / (number of data points in area a + number of data points in area b + number of data points in area c + number of data points in area d). The accuracy for this form is 0.899.

[0081] This form of prediction model has a high accuracy rate of 0.899, indicating that it can calculate the prediction probability p with high precision.

[0082] Figure 5 is a scatter plot of the data used to build the prediction model. The vertical axis of Figure 5 shows the actual value (°C) of the change in tympanic membrane temperature. The horizontal axis shows the prediction probability (%) calculated using the prediction model. sv is the target increase in core body temperature. The target increase in core body temperature sv is set as the change in core body temperature at which a bather feels "sweating". cv is the cutoff value. The cutoff value cv is the value that distinguishes between cases where the core body temperature has been determined to be equal to or greater than the target increase sv, and cases where the core body temperature has not been determined to be equal to or greater than the target increase sv, based on the prediction probability p. In this embodiment, the target increase in core body temperature sv is set to 0.4°C. The cutoff value cv is set to 22.6%. The cutoff value cv is an example of a "threshold".

[0083] The prediction model of the bathing navigation program 30 calculates the prediction probability p, rather than the core body temperature itself. Therefore, the bathing navigation program 30 requires a cutoff value cv to decide whether or not to recommend that the bather M get out of the bath based on the prediction probability p.

[0084] The predictive model is designed to achieve the objective of safe bathing. Therefore, the prediction probability p at the point where the recall is highest should be adopted as the cutoff value cv for the prediction accuracy metric. However, focusing solely on recall would significantly reduce precision, potentially leading to bathers being prompted to leave the bath before they feel the warming effect, resulting in increased complaints and dissatisfaction due to the discrepancy between their perception and the actual bathing experience, thus compromising convenience. Therefore, in this model, the prediction probability p at which recall and accuracy are equal is used as the cutoff value cv.

[0085] <Hot water supply navigation processing> Next, the aforementioned hot water navigation process will be explained with reference to Figure 6. When the bathing navigation device 2 is powered on, the CPU 11 starts and executes the bathing navigation program 30. When the CPU 11 detects that a bather M has entered the bath, it executes the hot water navigation process shown in Figure 6. In this embodiment, the CPU 11 detects the bather's entry by an increase in water pressure measured by the water pressure sensor 29. Alternatively, the CPU 11 may detect the bather M's entry from measurement data of a water level sensor that detects the water level of the hot water 72 in the bathtub 71. Alternatively, the CPU 11 may detect whether or not a bather M is in the bath using a human presence sensor.

[0086] The CPU 11 first notifies the user IF 13 of the start information (S10). For example, the CPU 11 switches the indicator lamp 13b from off to green. This allows the bather M to know that their bathing activity is being monitored by the bathing navigation device 2.

[0087] The CPU 11 starts acquiring heart rate and bathing environment information (S12). For example, the wearable terminal 5 measures the bather M's heart rate at predetermined time intervals (e.g., every second) using the heart rate sensor 51. The CPU 11 acquires the heart rate measured by the heart rate sensor 51 from the wearable terminal 5. For example, the CPU 11 acquires the water temperature measured by the water temperature sensor 28 of the sensor device 23 and the bathroom temperature measured by the bathroom temperature sensor 16 of the main unit 21. S12 is an example of "biometric information acquisition processing," "bathing environment information acquisition processing," "biometric information acquisition step," and "bathing environment information acquisition step."

[0088] Furthermore, the bathing navigation program 30 acquires bather physical information before executing the bathing navigation process. For example, before bather M enters the bathroom 7, the bathing navigation program 30 acquires bather physical information 33, such as age and height, which has been entered into the smartphone 6. This process is an example of the "bather physical information acquisition process" and the "bather physical information acquisition step".

[0089] The CPU 11 determines whether the waiting time has elapsed (S13). The CPU 11 waits until the waiting time has elapsed (S13: NO). In this embodiment, the waiting time is set to 3 minutes. That is, the CPU 11 does not make a decision based on the predicted probability p until the fluctuation in heart rate due to body movement stabilizes.

[0090] When the waiting time has elapsed (S13:YES), CPU11 executes a prediction probability calculation process to calculate the prediction probability p (S21). S21 is an example of the "prediction probability calculation process" and the "prediction probability calculation step".

[0091] The prediction probability calculation process will be explained based on Figure 7. The CPU 11 extracts feature data x based on the bathing environment information 31, biometric information 32, and bather physical information 33 (S111).

[0092] For example, CPU 11 extracts "elapsed time (minutes)", "water temperature (°C) 1 minute ago", "bathroom temperature (°C) 1 minute ago", and "season" as feature data x based on the bathing time, water temperature data, bathroom temperature data, and bathing start date included in the bathing environment information 31. Also, for example, CPU 11 extracts "heart rate 1 minute ago (bpm)", "change in heart rate 1 minute ago", "heart rate difference 1 minute ago", and "rate of change from 3 minutes ago heart rate 1 minute ago" as feature data x based on the heart rate time series data included in the biometric information 32. CPU 11 extracts "age group", "gender", and "height (cm)" as feature data x based on the bather M's bather physical information 33.

[0093] CPU 11 substitutes the feature data x obtained in S111 into the prediction model shown in Equation 1 and calculates the prediction probability p (S112). The prediction model includes not only feature data x based on bathing environment information 31 and biometric information 32, but also feature data x based on bather physical information 33. The feature data x used in the prediction model is weighted by coefficient a. A coefficient a of "-0.978" is set for "60s" and a coefficient a of "1.245" is set for "30-50s". Therefore, even if bathers in the same bathing environment are in the same age group as bathers in their 60s and older and bathers in their 30s and 50s, there will be a difference in the prediction probability p. Thus, when determining whether or not the core body temperature has risen to 0.4℃ or higher based on the prediction probability p, the difference in influence due to age will be taken into account.

[0094] The CPU 11, having calculated the predicted probability p, returns to Figure 6 and determines whether the calculated predicted probability p is greater than or equal to the cutoff value cv (S22). The cutoff value cv is set to a value where the recall and accuracy are equal. Therefore, if it is determined that the core body temperature has risen to 0.4°C or higher based on the predicted probability p, there is a high probability that the actual core body temperature is also 0.4°C or higher, and it can be expected that a determination that satisfies the bather M will be made.

[0095] If the CPU 11 determines that the predicted probability p is greater than or equal to the cutoff value cv (S22: YES), that is, if it predicts that the core body temperature has risen to 0.4°C or higher, it executes the first recommendation process (S41). The first recommendation process is a process to recommend that the bather M get out of the bath when it is predicted that the core body temperature has risen to 0.4°C or higher.

[0096] For example, as shown in Figure 8(a), the CPU 11 may display a message on the display 13a such as, "Your body is warming up. It's about time to get out of the bath. Let's get ready to get out of the bath," or it may cause the voice output unit 15 to give a voice notification. Alternatively, the CPU 11 may switch the indicator lamp 13b from green to yellow to advise the bather M to get out of the bath. This allows the bather M to objectively recognize that their body has warmed up to the core and to get out of the bath after experiencing the warming effect of the water.

[0097] Generally, when a person's core body temperature rises by about 0.5°C due to bathing, the bather M will feel sweating or experience a warming effect. Therefore, it is desirable to recommend getting out of the bath when the core body temperature reaches 0.5°C or higher. However, in actual bathing, actions such as washing the body and pouring water over oneself in the bathtub 71 promote metabolism compared to simply bathing, and it is possible that the core body temperature will rise more quickly. Therefore, in this configuration, getting out of the bath is recommended when the core body temperature reaches 0.4°C or higher. For example, elderly people have a duller sense of temperature than younger people, and even after being recommended to get out of the bath, they may not feel sweating or experience a warming effect, and may continue bathing, potentially deviating significantly from a physiologically safe timing.

[0098] Therefore, as shown in Figure 6, the CPU 11 performs the first recommendation process and then determines whether or not the warning conditions are met (S42). The warning conditions are for determining whether or not to warn bather M, whose core body temperature is predicted to be 0.4°C or higher, to stop bathing.

[0099] The higher the water temperature, the easier it is for core body temperature to rise, and the shorter the time it takes to deviate from a physiologically safe time. Therefore, in this configuration, the content of the warning conditions differs depending on the water temperature. For example, if the water temperature is below 41°C, the warning condition is that a first predetermined time (e.g., 2 minutes) has elapsed since the first recommended process was executed. On the other hand, if the water temperature is 41°C or higher, the warning condition is that a second predetermined time (e.g., 1 minute) has elapsed since the first recommended process was executed. The second predetermined time is set to be shorter than the first predetermined time.

[0100] If the CPU 11 determines that the warning conditions are not met (S42:NO), it determines whether or not hot water has been detected (S43). S43 is an example of a "detection process". For example, if the water pressure value measured by the water pressure sensor 29 does not change, the CPU 11 determines that hot water has not been detected (S43:NO). In this case, the CPU 11 returns to S42 and continues to monitor the bather M.

[0101] After executing the first recommendation process (S41), if the CPU 11 detects hot water discharge without meeting the warning conditions (S42: NO) (S43: YES), it proceeds to S62. For example, the CPU 11 detects hot water discharge when the water pressure value measured by the water pressure sensor 29 decreases.

[0102] On the other hand, if the CPU 11 determines after executing the first recommendation process (S41) that the warning conditions are met (S42: YES), it proceeds to S51 and issues a warning about dispensing hot water. S42 and S51 are examples of the "first warning process".

[0103] For example, as shown in Figure 8(b), the CPU 11 may display a message on the display 13a such as, "Your body is getting warmer. Shall we get out of the bath?" or have the voice output unit 15 provide an audio notification. Alternatively, the CPU 11 may switch the indicator lamp 13b from a yellow state to a red state to warn the bather M to get out of the bath. Furthermore, the CPU 11 may use the voice output unit 15 to output warning sounds such as voice guidance or a buzzer. As a result, the bathing navigation device 2 can warn the bather M to get out of the bath before bathing, which continues even after it has been determined that the core body temperature has risen to 0.4°C or higher, significantly deviates from a physiologically safe range. In addition, because the warning method differs from the recommendation method, it becomes easier for the bather M to distinguish the need to get out of the bath.

[0104] Returning to Figure 6, if the CPU 11 determines that the predicted probability p is not equal to or greater than the cutoff value cv (S22:NO), that is, if it predicts that the core body temperature is not 0.4°C or higher, it does not recommend that the bather M get out of the bath.

[0105] Even if the core body temperature is not above 0.4°C, prolonged bathing can cause significant blood pressure fluctuations upon exiting the bath or induce drowsiness, potentially making it unsafe. Therefore, if the CPU 11 determines that the predicted probability p is not equal to or greater than the cutoff value cv (S22: NO), it determines whether the elapsed time has exceeded the bath exit warning time (S31). The bath exit warning time is the time set as the amount of time for which bathing is sustainable within a physiologically safe range. In this embodiment, the bath exit warning time is set to 20 minutes.

[0106] If the CPU 11 determines that the elapsed time has not exceeded the hot water dispensing warning time (S31: NO), it determines whether the elapsed time has exceeded the hot water dispensing recommendation time (S32). The hot water dispensing recommendation time is the time set to recommend that bather M dispense hot water before the hot water dispensing warning time has elapsed. In this configuration, the hot water dispensing recommendation time is set to 17 minutes.

[0107] If the CPU 11 determines that the elapsed time has not exceeded the hot water supply recommendation time (S32: NO), it determines whether or not hot water supply has been detected (S34). If the CPU 11 does not detect hot water supply (S34: NO), it determines whether or not to perform the next determination (S35). The determination of whether or not to prompt hot water supply is performed periodically. In this configuration, the determination is performed at 1-minute intervals. The CPU 11 determines that it will not perform the next determination until 1 minute has elapsed from the time of prediction execution, i.e., when the prediction probability calculation process (S21) is executed (S35: NO). In this case, the CPU 11 returns to S34 and waits while monitoring the bather M until the next determination is performed.

[0108] If CPU11 determines that it has not detected any hot water supply (S34: NO) and that more than one minute has passed since the prediction execution time, it decides to perform the next determination (S35: YES), returns to S21, and calculates the next prediction probability p.

[0109] Therefore, if the bathing navigation device 2 predicts that the core body temperature is not 0.4°C or higher, and until the recommended time for leaving the bath has elapsed, it is highly likely that the bather M is bathing safely, and therefore does not issue a notification regarding leaving the bath.

[0110] In response to this, the CPU 11 determines that the predicted probability p is not equal to or greater than the cutoff value cv, that is, it predicts that the core body temperature is not 0.4°C or higher (S22: NO), but if the elapsed time exceeds the time for recommending discharging the bath (S31: NO, S32: YES), it executes the second recommendation process (S33). The second recommendation process is a process to recommend discharging the bather M when it does not predict that the core body temperature has reached 0.4°C or higher.

[0111] For example, as shown in Figure 8(c), the CPU 11 may display a message on the display 13a such as, "Ten minutes have passed since you started soaking in the hot water. It's time to get ready to get out of the water," or it may cause the voice output unit 15 to provide voice notification. Alternatively, the CPU 11 may switch the indicator lamp 13b from a green state to a yellow state to advise the bather M to get out of the water. This allows the bather M to recognize that they should get out of the water because they have been in the bath for too long, even if they are not aware of sweating or feeling the effects of the heat.

[0112] Returning to Figure 6, if CPU 11, which has performed the second recommendation process, does not detect hot water flow (S34: NO), it will wait while monitoring the hot water flow until the next determination is made, as described above (S35: NO).

[0113] On the other hand, if CPU 11 detects hot water discharge after executing the second recommendation process (S22: NO, S31: NO, S32: NO, S34: YES), it proceeds to S62. The processing from S62 onwards will be described later.

[0114] Even if the CPU 11 determines that the predicted probability p is not equal to or greater than the cutoff value cv (S22: NO), if it determines that the elapsed time has exceeded the bath dispensing warning time (20 minutes) (S31: YES), it executes a warning process (S51) and warns the bather M to dispensing water. S31 and S51 are examples of the "second warning process". Therefore, the bathing navigation device 2 is expected to make the bather M, who is not aware of sweating or not feeling the warming effect, aware that prolonged bathing may make it difficult to dispensing water safely, and encourage them to dispensing water. As the warning process in S51 has been described above, its explanation will be omitted here.

[0115] The CPU 11, which performed the warning process, then performs a second warning process to prompt the user to dispense hot water again (S53).

[0116] An example of the re-warning process procedure will be explained based on the flowchart in Figure 9. The CPU 11 resets the re-warning count n (where n is a natural number) by assigning 0 to it (S211). The CPU 11 then determines whether or not it has received a stop instruction, which is an instruction to stop the warning prompting the dispensing of hot water (S212).

[0117] For example, if the stop button 13c is operated and a stop command is received by the CPU 11, the CPU 11 stops the warning (S221). For example, the CPU 11 clears the message that was displayed on the display 13a. For example, the CPU 11 switches the indicator lamp 13b from a red illuminated state to a green illuminated state. For example, the CPU 11 stops the warning sound output from the audio output unit 15. This allows the bather M to intentionally eliminate the discomfort caused by the alarm.

[0118] The CPU 11 determines whether or not hot water has been detected (S222). If the CPU 11 detects hot water (S222: YES), it terminates the re-warning process and proceeds to S62 in Figure 6.

[0119] For example, if the CPU 11 does not detect a change in the water pressure value measured by the water pressure sensor 29, it will not detect hot water discharge (S222: NO). In this case, the CPU 11 will determine whether 10 minutes have passed since the first warning (S223). Note that the time elapsed since the first warning does not have to be 10 minutes, as long as it is a time when safe hot water discharge can be encouraged.

[0120] If CPU11 determines that less than 10 minutes have passed since the initial warning (S223:NO), it returns to S222. CPU11 waits while monitoring the bathing status of bather M and the time elapsed since the initial warning.

[0121] If the CPU 11 determines that it has not detected hot water flow (S222: NO) and that 10 minutes have passed since the initial warning (S223: YES), it executes an emergency warning process (S231). The emergency warning process issues a stronger warning than the initial warning because the bather M may be asleep or unconscious in the bathtub 71. For example, the CPU 11 outputs a loud warning sound or warning message from the voice output unit 15. In addition, the CPU 11 may, for example, cause a warning to be issued by a hot water remote control (not shown) of the hot water heating equipment 8 installed in the kitchen or living room, to inform the family of the bather M of the abnormality.

[0122] On the other hand, if the stop button 13c is not operated after the first warning, for example, the CPU 11 determines that the stop command has not been received (S212: NO). In this case, the CPU 11 determines whether or not one minute has elapsed since the previous warning without detecting hot water discharge (S213). Note that the one minute is the time that takes into account the time to stop the alarm and dispense hot water, and a different time may be used as the reference point.

[0123] For example, after issuing a warning prompting the dispenser to dispense hot water in S51 of Figure 6, the CPU 11 waits for one minute while monitoring the operation of the stop button 13c and the bathing status (S213:NO).

[0124] On the other hand, if the CPU 11 determines that one minute has passed without detecting the dispensing of hot water after issuing a warning prompting the dispenser to dispense hot water in S51 of Figure 6 (S213: YES), it determines whether the number of re-warnings n is three or less (S214). If the number of re-warnings n is three or less (S214: YES), the CPU 11 executes a re-warning process (S215). For example, as shown in Figure 8(d), the CPU 11 may display a message on the display 13a saying, "Your body is sufficiently warm. Let's get out of the bath," or it may have the voice output unit 15 give an audio notification. The CPU 11 keeps the indicator lamp 13b lit red even after the re-warning. The CPU 11 re-outputs a warning sound using the voice output unit 15.

[0125] The CPU 11, which executed the re-warning process, adds 1 to the re-warning count n (S216) and returns to S212. If the CPU 11 receives a stop instruction after the re-warning (S212:YES), it stops the re-warning (S221).

[0126] If CPU 11 does not accept a stop command after the second warning (S212: NO), and if it has not detected hot water flow for 1 minute since executing the previous second warning process (S213: YES), it will execute the second warning process in the same manner as above (S214: YES, S215, S216) and return to S212. CPU 11 will repeat the process from S213 to S216 until it has issued a maximum of 3 second warnings without accepting a stop command.

[0127] CPU 11 executes emergency warning processing (S231) if the number of re-warnings n exceeds 3 (S214: NO), that is, if the warning is repeated 4 times. The emergency warning processing has been explained above, so we will omit the explanation here.

[0128] The CPU 11, which has executed the emergency warning process, determines whether or not it has received a stop command (S232). If the CPU 11 determines that it has received a stop command in response to the operation of the stop button 13c (S232:YES), it stops the emergency warning (S233) and determines whether or not hot water has been detected (S241). The stopping of the emergency warning is the same as stopping the warning in S221, so the explanation is omitted. On the other hand, if the CPU 11 has not received a stop command (S232:NO), it continues to issue the emergency warning and determines whether or not hot water has been detected (S241).

[0129] If CPU 11 does not detect hot water flow (S241: NO), it returns to S232 and waits while monitoring the operation of the stop button 13c and the bathing status of bather M. If CPU 11 detects hot water flow (S241: YES), it terminates the re-warning process and proceeds to S62 in Figure 6.

[0130] Returning to Figure 6, CPU 11 proceeds to S62 and finishes acquiring heart rate and bathing environment information, which was started in S12 (S62).

[0131] The CPU 11 creates a bathing file 41 and saves information about bathing that is the target of the bathing navigation process to the bathing file 41 (S63). The bathing file 41 is saved in the non-volatile area of ​​memory 12. S63 is an example of "memory processing".

[0132] For example, the CPU 11 creates a bathing file 41 with a filename that includes the date and time the bathing started, so that it can distinguish between bathing files 41. The CPU 11 stores measurement data showing the water temperature measured by the water temperature sensor 28, the water pressure measured by the water pressure sensor 29, and the bathroom temperature measured by the bathroom temperature sensor 16 in the bathing environment information 31. The CPU 11 also determines the season based on the date and time the bathing started and stores the determined season in the bathing environment information 31. The CPU 11 also stores the size of the bathtub that bather M entered via smartphone 6 into the bathing navigation device 2 in the bathing environment information 31. The CPU 11 also stores heart rate time-series data showing the heart rate measured by wearable terminal 5 in biometric information 32. The CPU 11 also stores the gender, age, and height that bather M entered into the bathing navigation device 2 via smartphone 6 into bather body information 33. For example, CPU 11 stores the feature data x extracted in S111 in Figure 7 in intermediate variable 34. CPU 11 stores the predicted probability p calculated in S21 in Figure 6 in processing content 35. CPU 11 stores the recommendation content in S41 in Figure 6, the recommendation content in S33, the warning content in S51, the re-warning content in S215 in Figure 9, and the emergency warning content in S233 in processing content 35. If the bather M leaves the bath before the predicted probability p becomes equal to or greater than the cutoff value cv, and before the bathing time has elapsed since the recommended time for leaving the bath, CPU 11 stores "no notification" in processing content 35.

[0133] Subsequently, the CPU 11 notifies that the hot water navigation process has finished by turning off the indicator lamp 13b (S64), and then terminates the hot water navigation process.

[0134] <Update process> As shown in Figure 2, the bathing navigation system 1 automatically updates the predictive model of the bathing navigation program 30 using an update program 81 stored in the hot water heating equipment 8. The hot water heating equipment 8 is an example of an "external device".

[0135] For example, as shown in Figure 10, when the bathing navigation program 30 of the bathing navigation device 2 detects an update timing (A1), it sends an update request to the hot water heating equipment 8 using the communication unit 14 (A2). The update request is accompanied by a bathing file 41. The update timing may be regular, such as at the end of the month or once a year, or it may occur when a predetermined number of bathing files 41 have been accumulated in the memory 12. A2 is an example of the "transmission process".

[0136] When the hot water heating equipment 8 receives a bathing file 41 using the communication unit 83 (A2), it stores and remembers the received bathing file 41 in the bathing log database (hereinafter referred to as "bathing log DB") 82 (A3). As a result, past bathing-related information is collected in the bathing log DB 82. AS2 is an example of "reception processing". A3 is an example of "bathing history storage processing".

[0137] The hot water heating and bathing equipment 8 constructs a new prediction model based on the bathing file 41 stored in the bathing log DB 82 (B1). B1 is an example of the "construction process". The hot water heating and bathing equipment 8 transmits update information to the bathing navigation device 2 using the communication unit 83 (C1). The update information includes information such as the newly constructed prediction model and the construction date and time. C1 is an example of the "prediction model transmission process".

[0138] When the bathing navigation program 30 receives update information using the communication unit 14, it updates the existing prediction model based on that update information (C2). In other words, the existing prediction model is updated to a new prediction model. C2 is an example of an "update process". The bathing navigation program 30 reduces the memory load by deleting the bathing file 41, which has been sent to the hot water heating equipment 8 and backed up, from the memory 12 (C3).

[0139] As described in detail above, the bathing navigation program 30 of this embodiment extracts feature data x by considering not only biometric information 32 and bathing environment information 31 but also bather physical information 33, and calculates a prediction probability p by substituting the extracted feature data x into a prediction model. Therefore, the bathing navigation program 30 has higher prediction accuracy compared to predictions based on biometric information 32 and bathing environment information 31 that do not consider bather physical information 33. For example, even with the same bathing environment information 31, differences in age and gender will result in differences in the calculated prediction probability p. In other words, it is possible to calculate a prediction probability that takes into account the elderly, who are at high risk of bathing accidents. The bathing navigation program 30 recommends getting out of the bath by comparing such a prediction probability p with a cutoff value cv, so for example, whether bather M is elderly or young, or whether bather M is male or female, it can recommend getting out of the bath before deviating from a physiologically safe timing. Furthermore, this bathing navigation program 30 calculates a predicted probability p, which is the probability that the core body temperature will rise to 0.4°C or higher, using a prediction model, rather than the core body temperature itself. Therefore, the prediction result is less affected by the bather M's body temperature before entering the bath. As a result, the prediction model used to calculate the predicted probability is simpler than a prediction model that considers the heat balance from the skin to the core. This makes it possible to predict the rise in core body temperature with high accuracy without placing an excessive computational load on the device or requiring high processing power. Thus, this bathing navigation program 30 can accurately predict changes in core body temperature using a simple prediction model and recommend that the bather M exit the bath before deviating from a physiologically safe timing.

[0140] (Second Embodiment) Next, the bathing navigation program, bathing navigation method, bathing navigation device, and bathing navigation system of the second embodiment will be described with reference to the drawings. The bathing navigation device 2A of the second embodiment shown in Figure 11 predicts core body temperature without using biological information 32. This differs from the bathing navigation device 2 of the first embodiment, which predicts core body temperature using biological information 32. Here, we will mainly explain the differences from the first embodiment, and will omit explanations of the same configuration and functions as the first embodiment as appropriate.

[0141] As shown in Figure 11, the bathing navigation device 2A is connected to external devices such as a smartphone 6 and a hot water supply and heating equipment 8 in a communicative manner to form a bathing navigation system 1A. The bathing navigation device 2A does not need to be connected to a wearable terminal 5.

[0142] The bathing navigation device 2A has a bathing navigation program 30A stored in the memory 12 of the main unit 21A. The bathing navigation program 30A causes the bathing navigation device 2A to calculate a predicted probability using the bathing environment information 31 and the bather's physical information 33, and recommends getting out of the bath if the calculated predicted probability exceeds a threshold. The bathing navigation device 2A stores a bathing file 41A that does not contain biometric information 32 in the memory 12.

[0143] Next, the operation of the bathing navigation device 2A will be explained. When power is inserted, the CPU 11 of the bathing navigation device 2A reads the bathing navigation program 30A from the memory 12 and executes it.

[0144] Bather M enters the bathroom 7 without wearing the wearable device 5. When the CPU 11 detects that bather M has entered the bath, it executes the hot water navigation process shown in Figure 12. The hot water navigation process shown in Figure 12 performs the same process as the hot water navigation process in Figure 6, except for processes S1016, S1021, and S1062.

[0145] When CPU 11 starts acquiring bathing environment information 31 at S1012 in Figure 12, it executes the prediction probability calculation process (S1012, S1021). Note that CPU 11 executes the prediction probability calculation process without acquiring the heart rate of the bather M.

[0146] In the prediction probability calculation process shown in Figure 13, the CPU 11 acquires feature data xA based on bathing environment information 31 and bather physical information 33 (S1111), and substitutes the acquired feature data xA into the prediction model shown in Equation 1 to calculate the prediction probability (S112). In other words, the CPU 11 does not use heart rate in calculating the prediction probability.

[0147] Here, the bathing environment information 31 and the bather's physical information 33 used to calculate the predicted probability do not fluctuate wildly due to changes in posture caused by body movement during bathing or hydrostatic pressure acting on the body. Therefore, the bathing navigation program 30A may execute the predicted probability calculation process S1021 immediately after processing S1012, or it may wait for a certain period of time after processing S1012 before executing the predicted probability calculation process S1021. This waiting time is to reduce the processing load on the CPU 11, and can be shorter than the waiting time shown in S13 of Figure 6, which is for waiting for heart rate fluctuations to stabilize.

[0148] An example of feature data xA is shown in Figure 14. The feature data xA shown in Figure 14 includes, for example, binarized data for bather body information 33, specifically for ages 30-50, 60 and over, and gender. The feature data xA also includes, for example, height and weight as bather body information 33. Height and weight are used as values ​​to one decimal place. The bathing navigation device 2A, for example, works in conjunction with a smartphone 6 to receive input of bather body information 33 in advance and stores it in memory 12.

[0149] Feature data xA includes, for example, bathing environment information 31, the time elapsed since the start of bathing (hereinafter referred to as bathing time), the temperature of the bathroom 7 (hereinafter referred to as bathroom temperature), the temperature of the water in the bathtub 71 (hereinafter referred to as water temperature), and the season. Unlike biometric information 32 such as heart rate, bathing environment information 31 does not fluctuate wildly due to the body movements of the bather M. Therefore, bathing environment information 31 can be used in feature data xA without correction.

[0150] Note that the timing and frequency of acquisition of the season, bathroom temperature, and water temperature may differ from that of this configuration. For example, since the season does not change during bathing, it may be acquired immediately after the start of the water dispensing navigation process, or only once a day or once a week. The bathroom temperature and water temperature may be acquired only once after bathing is detected, but since they may change during bathing due to actions such as pouring water over the bather, adding water or cold water by the bather M, and heat dissipation from the surface of the water, it is desirable to acquire them multiple times after bathing is detected. The bathroom temperature and water temperature may be acquired from the bathroom temperature sensor 16 and the water temperature sensor 28, or they may be temporarily stored in memory 12 and read from memory 12 when S1012 is executed.

[0151] In this configuration, the constant b is set to -105.949.

[0152] Figure 15 is a scatter plot of the data used to build the prediction model. The vertical axis of Figure 15 shows the actual value (°C) of the change in tympanic membrane temperature. The horizontal axis shows the prediction probability (%) calculated using the prediction model. sv is the target increase in core body temperature. The target increase in core body temperature sv is set as the change in core body temperature at which a bather feels "sweating". cv is the cutoff value. The cutoff value cv is the value that distinguishes between cases where the core body temperature is judged to have exceeded the target increase sv and cases where it is judged not to have exceeded the target increase sv, based on the prediction probability p. In this configuration, the target increase in core body temperature sv is set to 0.4°C. The cutoff value cv is set to 17.8%. The cutoff value cv is an example of a "threshold". The cutoff value cv represents a state where false alarms and non-false alarms are balanced.

[0153] If we define the cutoff value cv as the prediction probability p at which the recall and accuracy are equal, then in the scatter plot of Figure 15, the prediction probability cv is set to 17.8%. When the cutoff value cv is 17.8%, the recall is 89.3%, which means that it is possible to make a bathing recommendation that is acceptable to the bather M.

[0154] <Verification> This study investigates the impact of the presence or absence of biometric information on the bathing monitoring function. Data obtained from 12 types of bathing actions, as shown in Figure 16, was used for the investigation. Each of the 12 bathing actions was assigned a bathing identification number. Identification numbers 1-6 represent the bathing actions of the first bather, while identification numbers 7-12 represent the bathing actions of the second bather. As shown in Figure 16, the first and second bathers were different individuals with different genders, ages, weights, and heights. Also, as shown in Figure 16, both the first and second bathers performed the bathing action six times each. For each bathing action, the season, bathing time, bathroom temperature, and water temperature were acquired as bathing environment information. The season was acquired once when bathing was detected. Bathroom temperature and water temperature were acquired every 0.5 minutes. The season, bathing time, bathroom temperature, and water temperature acquired for each bathing action are shown for bathing identification numbers 1-12. The bathroom temperature and water temperature shown in Figure 16 represent the average values ​​within the bathing time. Since all bathing activities were performed on different days during the winter, the bathing time, bathroom temperature, and water temperature differ. For bathing activities with bathing identification numbers 1 to 12, the first and second bathers wore wearable devices 5 to acquire heart rate time-series data.

[0155] Figure 17A shows the time-series predicted probability calculated by inputting the bather's physical information, bathing environment information, and the first bather's heart rate time-series data into the bathing navigation device 2 of the first embodiment for each bathing action with bathing identification numbers 1 to 6 shown in Figure 16. In other words, Figure 17A shows the time-series data of the predicted probability using biometric information 32. The cutoff value cv shown in Figure 17A is set to 22.8%.

[0156] Heart rate fluctuates wildly due to changes in posture caused by body movement during bathing and hydrostatic pressure on the body. Therefore, as shown in Figure 17A, the predicted probability using heart rate time series data is affected by the wild fluctuations in the heart rate time series data and fluctuates. Heart rate changes significantly immediately after bathing due to changes in posture caused by bathing movements and hydrostatic pressure on the body. To minimize the influence of heart rate fluctuations on the predicted probability, the bathing navigation device 2 starts calculating the predicted probability 4 minutes after detecting bathing, for example, as shown in the predicted probability time series data P11~P16 in Figure 17A.

[0157] As shown in the predicted probability time series data P12-P16 in Figure 17A, when bathing operations are performed for bathing identification numbers 2-6 shown in Figure 16, the bathing navigation device 2 issues an alarm recommending exiting the bath when the predicted probability exceeds the cutoff value cv at the time it starts monitoring the bathing. At the time of the first alarm, it is highly likely that the first bather does not subjectively feel any sweating. Therefore, the bathing navigation device 2 may issue an alarm before the bather feels the warming effect of the bath.

[0158] Furthermore, for example, during the bathing operation with bathing identification number 2, as shown in the predicted probability time series data P12 in Figure 17A, the bathing navigation device 2 stops issuing alarms approximately 1 minute after the initial alarm (bathing time approximately 5 minutes) when the predicted probability falls below the cutoff value cv. Approximately 1 minute after the alarm stops (bathing time approximately 6 minutes), the bathing navigation device 2 issues another alarm as the predicted probability exceeds the cutoff value cv again. Approximately 0.5 minutes after the re-issuance (bathing time approximately 6.5 minutes), the bathing navigation device 2 issues yet another alarm as the predicted probability falls below the cutoff value cv again. 1.5 minutes after the alarm stops again (bathing time 8 minutes), the bathing navigation device 2 issues yet another alarm as the predicted probability exceeds the cutoff value cv. Thus, the bathing navigation device 2, which calculates the predicted probability using heart rate time series data, may issue alarms frequently. As shown in the predicted probability time series data P13 and P16, the bathing actions indicated by bathing identification numbers 3 and 6 also trigger frequent alarms. Frequent alarms can lead to a loss of trust among bathers in the alarms that prompt them to exit the bath at a physiologically safe time. As a result, bathers may stop continuously using the bathing monitoring function of the bathing navigation device 2.

[0159] Figure 17B shows the time-series predicted probabilities calculated by inputting the bather's physical information and bathing environment information of the first bather into the bathing navigation device 2A of the second embodiment for each bathing action with bathing identification numbers 1 to 6 shown in Figure 16. In other words, Figure 17B shows the time-series data of predicted probabilities without using biometric information. The cutoff value cv shown in Figure 17B is set to 17.8%.

[0160] As shown in the predicted probability time series data P21-P26 in Figure 17B, the predicted probability without using heart rate time series data is not affected by changes in posture due to body movement during bathing or fluctuations in heart rate due to hydrostatic pressure on the body, and therefore fluctuates less than the predicted probability time series data P11-P16 shown in Figure 17A. Since bathing environment information (season, bathing time, bathroom temperature, and water temperature) and bather physical information (gender, age, weight, and height) are not affected by the bather's body movement, the bathing navigation device 2A can calculate the predicted probability and monitor bathing immediately after detecting bathing. In this verification, the predicted probability is calculated and bathing is monitored from one minute after detecting bathing. This is because few bathers leave the bath within one minute of starting bathing, and the probability of physiological safety is extremely high.

[0161] As shown in the predicted probability time series data P21-P26 in Figure 17B, the predicted probability calculated without using heart rate time series data is not affected by changes in posture due to body movement during bathing or hydrostatic pressure on the body, and therefore does not fluctuate as much as the predicted probability shown in Figure 17A, which is calculated using heart rate time series data. As shown in the predicted probability time series data P21-P26, during the bathing actions of bathing identification numbers 1-6 shown in Figure 16, the bathing navigation device 2A does not emit an alarm because the predicted probability does not exceed the cutoff value cv when it starts monitoring the bathing. In other words, the bathing navigation device 2A does not emit an alarm before the first bather subjectively feels the sensation of sweating.

[0162] The predicted probability time series data P21-P26 in Figure 17B shows that the predicted probability gradually increases after bath monitoring begins, and after exceeding the cutoff value cv, it never falls below the cutoff value cv. Therefore, the bathing navigation device 2A does not frequently emit alarms. When the bathing navigation device 2A emits an alarm, the first bather is subjectively feeling sweating. In other words, the bathing navigation device 2A emits an alarm when the first bather is experiencing the warming effect of the bath, prompting them to get out of the bath. Thus, the bathing navigation device 2A improves the bather's trust in the bathing monitoring function, and increases the likelihood that the bather will continue to use the bathing monitoring function.

[0163] Furthermore, the bathing navigation device 2A can monitor bathing even if the bather does not wear the wearable terminal 5 while bathing. Therefore, bathers do not have to go through the trouble of wearing the wearable terminal 5, making it easy for them to continuously use the bathing monitoring function of the bathing navigation device 2A.

[0164] Figure 18A shows the time-series predicted probability calculated by inputting the bathing navigation device 2 of the first embodiment, along with the bather's physical information, bathing environment information, and the second bather's heart rate time-series data acquired during the bathing action, for each bathing action with bathing identification numbers 7 to 12 shown in Figure 16. In other words, Figure 18A shows the time-series data of the predicted probability using biometric information. The cutoff value cv shown in Figure 18A is set to 22.8%.

[0165] The bathing navigation device 2 monitors bathing from 4 minutes after the user starts bathing. As shown in the predicted probability time series data P31, P32, and P35 in Figure 18A, the bathing navigation device 2 can calculate the predicted probability and monitor bathing for bathing activities with bathing durations of 4 minutes or more, as shown by bathing identification numbers 7, 8, and 11 in Figure 16. However, it does not calculate the predicted probability and cannot monitor bathing activities with bathing durations of 4 minutes or less, as shown by bathing identification numbers 9, 10, and 12. Therefore, the bathing navigation device 2 cannot monitor bathing for short durations.

[0166] Figure 18B shows the time-series predicted probabilities calculated by inputting the bather's physical information and bathing environment information of the second bather into the bathing navigation device 2A of the second embodiment for each bathing action with bathing identification numbers 7 to 12 shown in Figure 16. In other words, Figure 18B shows the time-series data of predicted probabilities without using biometric information. The cutoff value cv shown in Figure 18B is set to 17.8%.

[0167] The bathing navigation device 2A monitors bathing from one minute after bathing begins. As shown in Figure 16, all bathing actions with bathing identification numbers 7 to 12 involve a bathing time of one minute or more. Therefore, as shown in the predicted probability time series data P41 to P46 in Figure 18B, the bathing navigation device 2A can calculate the predicted probability for all bathing actions with bathing identification numbers 7 to 12 and monitor bathing. In other words, the bathing navigation device 2A can monitor even short bathing times.

[0168] The predicted probability time series data shown in Figures 17A, 17B, 18A, and 18B above were calculated assuming that the bathroom temperature and water temperature change due to the bather's actions such as heating, adding hot water, or adding cold water. However, even when the bather's actions such as heating, adding hot water, or adding cold water are occurring, the bathing navigation device 2A of the second embodiment does not use biometric information 32 that fluctuates wildly due to changes in posture caused by body movement during bathing or hydrostatic pressure on the body, so it can calculate predicted probabilities that reflect only the changes in bathroom temperature and water temperature.

[0169] As explained in detail above, the bathing navigation program 30A of the second embodiment extracts feature data xA by considering not only the bathing environment information 31 but also the bather's physical information 33, and calculates the prediction probability p by substituting the extracted feature data xA into the prediction model. Therefore, the bathing navigation program 30 has higher prediction accuracy compared to predictions based on the bathing environment information 31 without considering the bather's physical information 33. For example, even with the same bathing environment information 31, differences in age and gender will result in differences in the calculated prediction probability p. In other words, it is possible to calculate a prediction probability that takes into account the elderly, who are at high risk of bathing accidents. The bathing navigation program 30A recommends getting out of the bath by comparing such a prediction probability p with a cutoff value cv, so for example, whether the bather M is elderly or young, or whether the bather M is male or female, it can recommend getting out of the bath before deviating from a physiologically safe timing. Furthermore, this bathing navigation program 30A calculates the predicted probability p of a core body temperature change rate of 0.4°C or higher using a prediction model, rather than using the core body temperature itself. Therefore, the prediction result is less affected by the bather's (M) body temperature before entering the bath. As a result, the prediction model used to calculate the predicted probability is simpler than a prediction model constructed considering the heat balance from the skin to the core. This allows for highly accurate prediction of the rise in core body temperature without imposing an excessive computational load on the device or requiring high computational processing power. Thus, this bathing navigation program 30A accurately predicts changes in core body temperature using a simple prediction model and can advise the bather (M) to exit the bath before deviating from a physiologically safe timing. Moreover, since the bathing navigation program 30A calculates the predicted probability without using biometric information 32 that fluctuates wildly due to changes in posture caused by body movement during bathing or hydrostatic pressure on the body, it can advise the bather to exit the bath immediately after entering the bath and before deviating from a physiologically safe timing, enabling monitoring even during short bathing sessions.

[0170] This embodiment is merely illustrative and does not limit the present invention in any way. Therefore, the present invention can naturally be improved and modified in various ways without departing from its essence.

[0171] For example, the bathing navigation device 2 may be configured using a hot water supply and heating system 8. For example, if the bathroom remote control of the hot water supply and heating system 8 has a communication function that communicates with an external device such as a wearable terminal 5 or a smartphone 6, the bathing navigation program 30 may be stored in the control device of the hot water supply and heating system 8. In this case, the bathing navigation program 30 may use the water level sensor, water temperature sensor, and bathroom temperature sensor provided in the hot water supply and heating system 8 instead of the water pressure sensor 29, water temperature sensor 28, and bathroom temperature sensor 16. Also, the bathing navigation program 30 may use the display, indicator lamp, operation buttons, speaker, and clock of the bathroom remote control instead of the display 13a, indicator lamp 13b, stop button 13c, voice output unit 15, and timing unit 17. Furthermore, the control device of the hot water supply and heating system 8 may be equipped with an update program 81 and a bathing log DB 82.

[0172] For example, in the above embodiment, the bathing navigation program 30 displays a management screen on the smartphone 6 to accept the setting or modification of the bather's physical information 33. However, the setting or modification of the bather's physical information 33 may also be accepted using the user interface 13 of the bathing navigation device 2 or the bathroom remote control of the hot water heating equipment 8. Such a bathing navigation program 30 can acquire the bather's physical information 33 without using the smartphone 6, so the bather's physical information 33 can be easily set or modified when the bather M changes or before bathing. The bathing navigation program 30 may also accept the bather's physical information after the bather M has entered the bathroom 7. For example, the bathing navigation program 30 may access a device that stores the bather's physical information 33 and acquire the bather's physical information 33 before calculating the predicted probability p using a prediction model.

[0173] For example, the cutoff value cv does not have to be a value that equals the recall rate and the accuracy rate. However, by setting the cutoff value cv to a value that equals the recall rate and the accuracy rate, the timing of the recommendation to leave the bath coincides with the timing when the bather M becomes aware of sweating or feels the warming effect, making it possible for the bather who is recommended to leave the bath to receive the warming effect from the water and then leave.

[0174] For example, steps S42, S51, and S53 in Figure 6 are optional. However, the bathing navigation program 30 predicts that the core body temperature has risen above the target amount and recommends getting out of the bath. After a predetermined time has elapsed, it issues a warning to the bather M to get out of the bath, thereby preventing the bather M, whose core body temperature has risen above the target amount, from continuing to bathe at a time that is significantly outside of a physiologically safe period.

[0175] For example, the predetermined time set in the warning condition S42 in Figure 6 may be uniform regardless of the water temperature. However, if the bathing navigation program 30 sets the predetermined time in the warning condition to differ according to the water temperature, the timing of the warning to bather M to dispense the water will differ according to the water temperature. This allows the program to issue a warning to bather M when their core body temperature rises above the target amount at an appropriate timing in accordance with the change in core body temperature.

[0176] For example, steps S31, S51, and S53 in Figure 6 are optional. However, if the bathing navigation program 30 is configured to issue a warning to exit the bath if, for example, the bather M is an elderly person whose core body temperature does not rise easily, and the bathing time exceeds the exit warning time before the predicted probability p exceeds the cutoff value cv, then it is possible to encourage the bather M to exit the bath before deviating from a physiologically safe timing, while adapting to the bather M's physical function.

[0177] For example, steps S32 and S33 in Figure 6 are optional. However, the bathing navigation program 30 can make bather M aware that their bathing time is getting long by advising them to exit the bath before the bathing time exceeds the exit warning time, thereby encouraging bather M to exit the bath safely.

[0178] For example, the bathing navigation program 30 does not need to update its prediction model. However, the bathing navigation program 30 stores bathing-related information in a bathing file 41 in memory 12, and by updating the prediction model using the stored bathing file 41, the prediction model can be refined and prediction accuracy can be improved.

[0179] For example, the bathing navigation program 30 may send an update request in response to user operations and update the predictive model. The bathing navigation program 30 may also automatically upload the bathing file 41 to an external server to reduce the memory load on the memory 12.

[0180] For example, if the memory 12 of the bathing navigation device 2 has sufficient memory capacity, the update program 81 may be stored in the memory 12, or the bathing file 41 may be accumulated and saved in the memory 12. In this case, the bathing navigation device 2 can automatically update its prediction model without relying on an external device.

[0181] For example, the update program 81 and the bath log DB 82 may be stored on an external server in the cloud. In this case, the bath navigation device 2 may be directly connected to the server, or it may be connected to the server via a relay device such as a hot water heating system 8 or a smartphone 6. Multiple bath navigation devices 2 are connected to the server. With this configuration, the server can collect an unspecified number of bath files 41 and build a predictive model based on them, thereby enabling the construction of a more sophisticated predictive model than in the above embodiment. The bath navigation program 30 may update the predictive model in response to a request from the server, or the bath navigation program 30 may request the server to send the predictive model, download the predictive model, and update it.

[0182] For example, the bathing navigation program 30 may store the bathing file 41 in external memory such as an SD card connected to the bathing navigation device 2. External memory and servers are examples of "memory accessible by the bathing navigation device." In this case, the user may, for example, export the bathing file 41 stored in the external memory to an information processing device such as a personal computer or mobile terminal, and store it in the information processing device. The information processing device may have the update program 81 stored in itself, or it may be connected to a server equipped with the update program 81. The information processing device passes the bathing file 41 to the update program 81 to build a prediction model, and stores the built prediction model in external memory. When the external memory containing the prediction model is connected to the bathing navigation device 2, the bathing navigation program 30 may retrieve the prediction model from the external memory and update the prediction model.

[0183] For example, the user may use an information processing device to download a new prediction model from the server and store it in external memory, and the bathing navigation program 30 may automatically update the existing prediction model using the new prediction model when the external memory is connected to the bathing navigation device 2.

[0184] For example, the bathing navigation program 30 may, when an external memory containing the update program 81 is connected to the bathing navigation device 2 in a communicative manner, have the CPU 11 execute the update program 81, build a prediction model using the bathing file 41 stored in memory 12, and update the prediction model.

[0185] Furthermore, in any flowchart disclosed in the embodiments, the execution order of any multiple processes in any multiple steps can be arbitrarily changed or executed in parallel, as long as no inconsistencies arise in the processing content.

[0186] Furthermore, the processes disclosed in the embodiments may be executed by a single CPU, multiple CPUs, hardware such as an ASIC, or a combination thereof. Also, the processes disclosed in the embodiments can be implemented in various forms, such as a recording medium or method that stores a program for executing the process. [Explanation of symbols]

[0187] 1. Bathing Navigation System 2. Bathing Navigation Device 5. Wearable devices 11 CPU 12 memory 13 User Interface 14 Communications Department 28. Water temperature sensor 30 Bathing Navigation Program

Claims

1. A bathing navigation program that can be executed by a bathing navigation device and provides information to a bather immersed in bathwater, The aforementioned bathing navigation device, A process for acquiring bathing environment information, which is information about the bathing environment of the aforementioned bather, A process for acquiring bather body information, which is information about the body of the bather, Based on the bathing environment information obtained in the bathing environment information acquisition process and the bather's physical information obtained in the bather's physical information acquisition process, feature data significant to the rise in core body temperature is extracted, and the extracted feature data is substituted into a prediction model for calculating the prediction probability, which is the probability that the bather's core body temperature will exceed the target rise amount, and a prediction probability calculation process is performed to calculate the prediction probability. If the predicted probability calculated in the prediction probability calculation process is equal to or greater than a threshold, the bather is advised to get out of the bath; if the predicted probability calculated in the prediction probability calculation process is not equal to or greater than the threshold, the bather is not advised to get out of the bath; To execute A bathing navigation program is structured in such a way.

2. In the bathing navigation program described in claim 1, The threshold is set to a value where the recall and accuracy are equal. A bathing navigation program is structured in such a way.

3. In the bathing navigation program described in claim 1, The aforementioned bathing navigation device, A detection process for detecting the hot water flow of the aforementioned bather, After recommending the bather to drain the bathwater in the first recommendation process, if it is determined that the warning conditions are met, a warning to drain the bathwater is issued to the bather; if it is determined that the warning conditions are not met, no warning to drain the bathwater is issued to the bather. Make it run, The aforementioned warning condition is that a predetermined time has elapsed after the execution of the first recommendation process while the detection process has not detected the hot water discharge. A bathing navigation program is structured in such a way.

4. In the bathing navigation program described in claim 3, The aforementioned warning conditions are determined by the water temperature sensor of the bathing navigation device, and the predetermined time varies depending on the temperature of the bathwater into which the bather enters. A bathing navigation program is structured in such a way.

5. In the bathing navigation program described in claim 1, The aforementioned bathing navigation device, If the predicted probability calculated in the aforementioned prediction probability calculation process is not equal to or greater than the threshold, and the bather's bathing time exceeds the bathing exit warning time set before the timing deviates from a physiologically safe timing, a second warning process is executed to issue a warning to the bather. A bathing navigation program is structured in such a way.

6. In the bathing navigation program described in claim 5, The aforementioned bathing navigation device, If the bathing time exceeds the recommended time for discharging the hot water, which is set prior to the warning time for discharging the hot water, a second recommendation process is executed to recommend that the bather discharge the hot water. A bathing navigation program is structured in such a way.

7. In the bathing navigation program described in claim 1, The aforementioned bathing navigation device, The bathing navigation device includes a memory processing step that stores bathing-related information in the device's memory, An update process to update the prediction model using the bathing-related information stored in the memory, To execute A bathing navigation program is structured in such a way.

8. A bathing navigation method that provides information to bathers while they are soaking in the bathwater, A bathing environment information acquisition step involves acquiring bathing environment information, which is information about the bathing environment of the aforementioned bather. A bather body information acquisition step involves acquiring bather body information, which is information about the body of the bather. A computer extracts feature data significant to the rise in core body temperature based on the bathing environment information acquired in the bathing environment information acquisition step and the bather's physical information acquired in the bather's physical information acquisition step, and substitutes the extracted feature data into a prediction model for calculating the prediction probability, which is the probability that the bather's core body temperature will exceed the target rise amount, and calculates the prediction probability in a prediction probability calculation step. If the predicted probability calculated in the aforementioned prediction probability calculation step is equal to or greater than a threshold, the bather is advised to exit the bath; if the predicted probability is not equal to or greater than the threshold, the bather is not advised to exit the bath; To do A bathing navigation method configured in such a way.

9. A bathing navigation device capable of running a bathing navigation program that provides information to bathers immersed in bathwater, Controller and Communications Department and, User interface and It has, The aforementioned controller A process for acquiring bathing environment information, which is information about the bathing environment of the aforementioned bather, A process for acquiring bather body information, which is information about the body of the bather, Based on the bathing environment information obtained in the bathing environment information acquisition process and the bather's physical information obtained in the bather's physical information acquisition process, feature data significant to the rise in core body temperature is extracted, and the extracted feature data is substituted into a prediction model for calculating the prediction probability, which is the probability that the bather's core body temperature will exceed the target rise amount, and a prediction probability calculation process is performed to calculate the prediction probability. If the predicted probability calculated in the aforementioned prediction probability calculation process is equal to or greater than a threshold, the bather is advised to exit the bath; if the predicted probability is not equal to or greater than the threshold, the bather is not advised to exit the bath. Execute A bathing navigation device configured in such a way.

10. In a bathing navigation system in which a bathing navigation device described in claim 9 and an external device are communicated with each other, The controller of the bathing navigation device is A memory process that stores bathing-related information about the bather's bathing each time they bathe, A transmission process that transmits the bathing-related information to the external device, An update process to update the existing prediction model with a new prediction model constructed by the external device, Execute, The bathing-related information includes, at a minimum, the bathing environment information, the bather's physical information, the feature data used to calculate the prediction probability, the prediction probability calculated by the prediction probability calculation process, and the content of the notification given to the bather based on the prediction probability. The external device is, A receiving process for receiving the bathing-related information transmitted from the bathing navigation device, A bathing history storage process that stores and stores the bathing-related information received in the aforementioned receiving process, A construction process to construct the new predictive model based on the bathing-related information stored in the bathing history storage process, A prediction model transmission process that transmits the new prediction model constructed in the construction process to the bathing navigation device, Execute A bathing navigation system configured in such a way.

Images