Sleep evaluation device, program, sleep evaluation method and learning device

Abstract

To provide a technology that creates an opportunity for patients to seek medical attention early by identifying symptoms that may be circadian rhythm sleep / wake disorder.SOLUTION: A sleep evaluation device comprises: an input unit that inputs operation history information, which is information indicating operation history, and circadian rhythms labeled in the operation history information, from an external device; a machine learning unit that performs supervised learning of a circadian rhythm classification model, which is a model for classifying circadian rhythms, based on teacher data consisting of the input operation history information of a first user and the circadian rhythms labeled in the operation history information; a classification unit that classifies the circadian rhythm of a second user based on the input operation history information of the second user and the circadian rhythm classification model; and a notification unit that notifies the second user of the circadian rhythm classified by the classification unit.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a sleep evaluation device, a program, a sleep evaluation method, and a learning device. [Background technology]

[0002] People suffer from sleep disorders. The International Classification of Sleep Disorders (ICSD-3) classifies sleep disorders into seven categories: insomnia, sleep-related breathing disorders, central hypersomnia, circadian rhythm sleep-wake disorders, parasomnias, sleep-related movement disorders, and other sleep disorders. The circadian rhythm sleep-wake disorder category, one of the sleep disorders, is divided into six categories: delayed sleep-wake phase disorder, advanced sleep-wake phase disorder, irregular sleep-wake rhythm disorder, non-24-hour sleep-wake rhythm disorder, shift work disorder, and jet lag disorder. Patients with circadian rhythm sleep-wake disorders may take more than 10 years to receive an accurate diagnosis after first seeking medical help for their sleep disorder, or may be misdiagnosed multiple times over the years (see Non-Patent Document 1). Diagnostic criteria for circadian rhythm sleep-wake disorders include sleep diaries and actigraphy. For example, Patent Document 1 listed below discloses a method for estimating the circadian rhythm of a living body based on the change over time in the length of anagen hair, which is determined by identifying anagen hair from each of a plurality of images. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2016-52414 [Non-patent literature]

[0004] [Non-Patent Document 1] Peter Mansbach, James SP Fadden, Lynn McGovern. Registry and survey of circadian rhythm sleep-wake disorder patients. Sleep Med X. Dec 2023. [Non-patent document 2] Mitsuyuki Nakao, Yoichi Takahashi, Akihiro Karashima, Ikuko Motoike, Norihiro Katayama, Kengo Kinoshita, and Kentaro Inui. Large-scale global analysis of rest-activity rhythms based on tweet frequency series. 45th Annual Meeting of the Japanese Society of Sleep Research and 30th Annual Meeting of the Japanese Society for Chronobiology. September 2023. Summary of the Invention [Problem to be solved by the invention]

[0005] However, sleep diaries and actigraphy require long-term recording, ranging from several weeks to several months, depending on the type of circadian rhythm sleep-wake disorder and individual differences, in order to establish diagnostic criteria. Recording a sleep diary is cumbersome, and long-term use requires significant effort. Actigraphy is currently not covered by insurance, is costly, and typically only lasts about two weeks. For example, the technology disclosed in Patent Document 1 requires multiple acquisition of skin images, making it impractical for long-term use. Therefore, there is a need for a measurement device that is user-friendly and can replace sleep diaries and actigraphy.

[0006] Furthermore, awareness of circadian rhythm sleep-wake disorders is very low, and many people who experience symptoms do not know the name of the disease. For example, Patent Document 2, mentioned above, suggests that, although there is a restriction of X (formerly Twitter (registered trademark)), approximately 1% of approximately 6.85 million user IDs may have a non-24-hour sleep-wake rhythm disorder (abbreviated as non24), a type of circadian rhythm sleep-wake disorder. However, it is believed that very few people are aware that they have a non-24-hour sleep-wake rhythm disorder (abbreviated as non24). The problem that this invention aims to solve is to classify circadian rhythms based on long-term daily operation history of home appliances and applications, instead of sleep diaries or actigraphy measurements, and to identify symptoms that may indicate a circadian rhythm sleep-wake disorder.

[0007] In view of the above circumstances, the present invention aims to provide a technology that can prompt patients to seek medical attention early by identifying symptoms that may be circadian rhythm sleep-wake disorders. [Means for solving the problem]

[0008] One aspect of the present invention is a sleep evaluation device that includes an input unit that inputs, from an external device, operation history information, which is information indicating an operation history, and teacher data consisting of circadian rhythms labeled with the operation history information; a machine learning unit that performs supervised learning of a circadian rhythm classification model, which is a model that classifies circadian rhythms, based on the teacher data consisting of the input operation history information of a first user and the circadian rhythms labeled with the operation history information; a classification unit that classifies the circadian rhythm of a second user based on the input operation history information of the second user and the circadian rhythm classification model; and a notification unit that notifies the second user of the circadian rhythm classified by the classification unit.

[0009] One aspect of the present invention is a program for causing a computer to function as the sleep evaluation device described above.

[0010] One aspect of the present invention is a sleep evaluation method comprising: an input step of inputting, from an external device, teacher data consisting of operation history information, which is time-series data indicating an operation history, and circadian rhythms labeled with the operation history information; a machine learning step of performing supervised learning of a circadian rhythm classification model, which is a model for classifying circadian rhythms, based on the teacher data consisting of the input operation history information of a first user and the circadian rhythms labeled with the operation history information; a classification step of classifying the circadian rhythm of a second user based on the input operation history information of the second user and the circadian rhythm classification model; and a notification step of notifying the second user of the circadian rhythm classified by the classification step.

[0011] One aspect of the present invention is a learning device that includes a control unit that learns a circadian rhythm classification model, which is a model that classifies circadian rhythms, based on training data consisting of operation history information, which is time-series data indicating operation history input from an external device, and circadian rhythms labeled with the operation history information. [Effects of the Invention]

[0012] This invention makes it possible to identify symptoms that may be circadian rhythm sleep-wake disorders from the daily operation history of home appliances and applications, thereby creating an opportunity for patients to seek medical attention early on. [Brief explanation of the drawings]

[0013] [Figure 1] 1 is a conceptual diagram of the present invention. [Figure 2] FIG. 1 is an explanatory diagram illustrating a sleep evaluation device according to a first embodiment. [Figure 3] FIG. 4 is a diagram showing a first example of an image that satisfies the image conditions in the first embodiment. [Figure 4] FIG. 6 is a diagram showing a second example of an image that satisfies the image conditions in the first embodiment. [Figure 5] FIG. 10 is a diagram showing a third example of an image that satisfies the image conditions in the first embodiment. [Figure 6] FIG. 10 is a diagram showing a fourth example of an image that satisfies the image conditions in the first embodiment. [Figure 7] FIG. 10 is a diagram showing a fifth example of an image that satisfies the image conditions in the first embodiment. [Figure 8] FIG. 10 is a diagram showing a sixth example of an image that satisfies the image conditions in the first embodiment. [Figure 9] FIG. 1 is a diagram showing an example of the hardware configuration of a sleep evaluation device according to a first embodiment. [Figure 10] FIG. 2 is a diagram showing an example of the configuration of a control unit included in the sleep evaluation device of the first embodiment. [Figure 11] 3 is a flowchart showing an example of the flow of processing executed by the sleep evaluation device of the first embodiment. [Figure 12] FIG. 10 is a diagram showing an example of the hardware configuration of an information processing device (an example of a learning device) according to a second embodiment. [Figure 13] 10 is a flowchart showing an example of the flow of processing executed by an information processing device according to the second embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0014] FIG. 1 shows a conceptual diagram of the present invention. In FIG. 1, the input unit and the notification unit are not shown. The machine learning unit 102 is a functional unit, the details of which will be described later. Therefore, in the example of FIG. 1, the machine learning unit 102 performs supervised learning of a circadian rhythm classification model, which is a model for classifying circadian rhythms, based on training data consisting of input operation history information of a first user and circadian rhythms labeled in the operation history information. In the example of FIG. 1, data D101 is an example of training data consisting of operation history information of a first user and circadian rhythms labeled in the operation history information.

[0015] The classification unit 103 is a functional unit, the details of which will be described later. Therefore, in the example of Fig. 1, the classification unit 103 classifies the circadian rhythm of the second user based on the input operation history information of the second user and a circadian rhythm classification model. In the example of Fig. 1, information D102 is an example of operation history information of the second user. In the example of Fig. 1, model M101 is an example of a circadian rhythm classification model.

[0016] The training data may be data that accurately represents the sleep-wake rhythm. The circadian rhythms that are labeled in the operation history information are manually assigned. When labeling, other information, such as the presence or absence of words such as "night shift" in the post content, may also be used as a reference. It is desirable that the person labeling the circadian rhythms has specialized knowledge of circadian rhythms.

[0017] (First embodiment) 2 is an explanatory diagram illustrating the sleep evaluation device 1 of the first embodiment. The sleep evaluation device 1 includes a control unit 11 including a processor 91, such as a CPU (Central Processing Unit), GPU (Graphics Processing Unit), or NPU (Neural Network Processing Unit), and a memory 92, which are connected via a bus.

[0018] The control unit 11 executes an input acquisition process, a machine learning process, a classification process, and a notification control process. The input acquisition process is a process in which information indicating an operation history is input from an external device. The external device is a device different from the sleep evaluation device. In other words, the input acquisition process is a process in which the control unit 11 acquires information indicating an operation history input to the sleep evaluation device from an external device that is a device different from the sleep evaluation device.

[0019] The operation history is a history of date and time information when a specific operation target is operated. An operation is, for example, turning the power of the operation target on or off. An operation is, for example, keyboard input. An operation is, for example, sending data from an application. However, the content of the operation (for example, on or off) does not matter, and only the date and time information of the operation is required. The operation target is, for example, a specific device such as a smartphone, PC, or smart glasses.

[0020] The operation target may be, for example, a home appliance such as a refrigerator, microwave, air conditioner, television, speaker, or electric fan. The operation target may be, for example, a remote control. The operation target may be, for example, a switch that turns on and off a light in a room. The operation target may be, for example, an application available on a device such as a smartphone, PC, or smart glasses. Such an application may be, for example, a chat app (or messenger app) or an electronic bulletin board. This operation history can serve as a substitute for a sleep diary or actigraphy measurements.

[0021] The machine learning process is a process of performing supervised learning of a circadian rhythm classification model using training data consisting of operation history information and circadian rhythms labeled in the operation history information. The circadian rhythm classification model is a model that classifies the circadian rhythms of the classification target based on training data consisting of input operation history information of the circadian rhythms of the first user that are to be classified and the circadian rhythms labeled in the operation history information. When the operation history information is data representing date and time, for example, points indicating the date and time of the operation may be plotted on a white image consisting of a predetermined range (96 x 72) of pixels, and the plotted image may be used as the operation history information.

[0022] During learning, a circadian rhythm classification model is created by repeatedly using a large amount (preferably around 100,000 or more to achieve high classification accuracy) of training data consisting of operation history information of the first user to be classified into circadian rhythms and circadian rhythms labeled with the operation history information.

[0023] As will be described later, circadian rhythms include classifications such as 24-hour sleep-wake, non-24-hour sleep-wake disorder (Straight), non-24-hour sleep-wake disorder (Jump), mixed disorder (24-hour sleep-wake + non-24-hour sleep-wake rhythm disorder), shift work, and irregular sleep.

[0024] The classification process is a process of classifying the circadian rhythm of the classification target based on the operation history information of the circadian rhythm classification target and the circadian rhythm classification model. Therefore, the classification process is a process of classifying the circadian rhythm of the second user based on the input operation history information of the second user and the circadian rhythm classification model, for example.

[0025] The operation history information of the nth user (n is 1 or 2) is information indicating the history of operations of the nth user, who is the nth user, on a predetermined operation target. The operation target (e.g., a smartphone) of the operation history information of the first user used in the machine learning process and the operation target (e.g., a messenger app) of the operation history information of the second user performing the classification process may be the same or different.

[0026] The notification control process is a process of controlling the operation of a predetermined notification device (hereinafter referred to as an "alarm") capable of making notifications, and causing the circadian rhythm classified by the classification process to be notified to a classification target in the classification process. Therefore, the notification control process is a process of controlling the operation of the alarm, and causing the circadian rhythm classified by the classification process to be notified to a second user. The alarm may be provided in, for example, the interface unit 12 described below.

[0027] <Examples of Effects of the Control Unit 11> The control unit 11 executes the classification process. A circadian rhythm classification model is used in the classification process. The circadian rhythm classification model is a model that classifies the type of circadian rhythm of the user. If a disorder is discovered, knowing what the disorder is will enable the user to become aware of the cause of their complaint. Therefore, by pointing out symptoms that may be circadian rhythm sleep-wake disorders, the sleep evaluation device can create an opportunity for the user to visit a medical institution early.

[0028] Furthermore, the circadian rhythm classification model classifies circadian rhythms based on operation history information. Operation history information is information indicating operation history, and operation history is a history of operations, so it can be said that the circadian rhythm classification model classifies circadian rhythms based on operation history. While it is possible that an operation object is operated while half asleep, the probability is extremely low, and if an operation object is operated, it can be said that this indicates that the user (the object of circadian rhythm classification) is awake. Therefore, operation history information can be said to be information indicating the timing when the user is awake. Operation history information can also be said to be information indicating the possibility of the user being asleep. In this way, operation history information can be said to be information related to sleep.

[0029] In the classification process, classification is performed using a trained circadian rhythm classification model obtained by the machine learning process executed in the previous stage.

[0030] <About operation history information> The operation history information will be further explained. The operation history information may be represented, for example, by time-series data indicating the operation history (hereinafter referred to as "operation history data"). The operation history information may also be represented by an image indicating the operation history (hereinafter referred to as "history image"). Image data of such an image may be generated based on the operation history data, for example.

[0031] The historical image may be, for example, an image that satisfies an image condition, which is that the coordinate on one of the vertical and horizontal axes indicates the date, the coordinate on the other axis indicates the time, and the time interval from the time indicated at one end of the other axis to the time indicated at the other end is n times 24 hours (n is an integer equal to or greater than 1).

[0032] <Examples of images that meet the image conditions> Examples of images that satisfy the image conditions will be described with reference to FIGS. 3 to 8 are diagrams showing examples of circadian rhythms of images that satisfy the image conditions in the first embodiment.

[0033] FIG. 3 shows images G301, G302, and G303. Image G301 shows an example of a first user's operation history information with a 24-hour sleep-wake circadian rhythm. Image G303 is another example of image G301. Image G302 shows an example of one year's operation history information to confirm differences from other circadian rhythms.

[0034] Figure 4 shows images G401, G402, and G403. Image G401 is operation history information for a first user, showing an example of a circadian rhythm with a non-24-hour sleep-wake rhythm disorder (Straight). Image G403 is another example of image G401. Image G402 shows an example of operation history information for one year to confirm differences from other circadian rhythms.

[0035] Figure 5 shows images G501, G502, and G503. Image G501 is operation history information for a first user, showing an example of a circadian rhythm that is a non-24-hour sleep-wake rhythm disorder (Jump). Image G503 is another example of image G501. Image G502 shows an example of operation history information for one year to confirm differences from other circadian rhythms.

[0036] FIG. 6 shows images G601, G602, and G603. Image G601 is operation history information for a first user, showing an example of a mixed circadian rhythm disorder (24-hour sleep-wake + non-24-hour sleep-wake rhythm disorder). Image G603 is another example of image G601. Image G602 shows an example of operation history information for one year to confirm differences from other circadian rhythms.

[0037] FIG. 7 shows images G701, G702, and G703. Image G701 shows operation history information for a first user, illustrating an example of a circadian rhythm involving shift work. Image G703 is another example of image G701. Image G702 shows an example of operation history information for one year, for confirming differences from other circadian rhythms.

[0038] FIG. 8 shows images G801, G802, and G803. Image G801 shows an example of operation history information for a first user with an irregular circadian rhythm. Image G803 is another example of image G801. Image G802 shows an example of operation history information for one year to confirm differences from other circadian rhythms.

[0039] In each of images G301, G401, G501, G601, G701, and G801, the x-axis represents each day from day 1 to day 72, and the y-axis, from one end of the y-axis to the other, represents the 48-hour period from midnight on each day indicated by the x-axis coordinate to just before midnight on the day after the next. If an operation is performed in 30-minute increments, a dot is plotted. In other words, if an operation is performed even once within the same 30-minute period, it is plotted as "operation performed," and if no operation is performed at all, it is plotted as "no operation" and not plotted. Symbols, shapes, etc., other than dots, may also be plotted.

[0040] Therefore, images G301, G401, G501, G601, G701, and G801 are all examples where the image condition n = 2. In addition, in the examples of images G301, G401, G501, G601, G701, and G801, the leftmost coordinate on the y-axis indicates midnight on each date indicated by the x-axis coordinate, and the rightmost coordinate on the y-axis indicates just before midnight on the day after that on each date indicated by the x-axis coordinate.

[0041] The 72-day period is the number of days considered necessary to avoid mistaking it for other circadian rhythms. If the period is shorter than 72 days, the possibility of mistaking it for other circadian rhythms increases. However, data does not need to exist for every day of the 72 days; for example, even if there is no data for several days due to travel, machine learning processing and classification processing can still be performed. To properly perform machine learning processing and classification processing, it is desirable to have two or more operation history records per day.

[0042] Images G301, G401, G501, G601, G701, and G801 indicate that an operation was performed on a specific operation target on the date and time indicated by the plotted dots. While these images show examples plotted with black dots, the dots may be any color other than the background color. Therefore, images G301, G401, G501, G601, G701, and G801 each represent information indicating an operation history over 72 days + 1 day. As described above, information indicating an operation is performed also indicates that the circadian rhythm estimation target is awake. Therefore, images G301, G401, G501, G601, G701, and G801, which show an operation history spanning multiple days, also represent information indicating the sleep-wake rhythm of the circadian rhythm estimation target.

[0043] For example, image G301 shows the operation history information of the first user, which indicates a 24-hour sleep-wake circadian rhythm. 24-hour sleep-wake refers to a situation in which both bedtime and wake-up times are roughly consistent. The majority of the 24-hour sleep-wake cycle is a normal circadian rhythm. However, if the wake-up time is late (for example, around noon), this may indicate a delayed sleep-wake phase disorder. Also, if the wake-up time is early (for example, around 2 a.m.), this may indicate an advanced sleep-wake phase disorder.

[0044] For example, image G401 shows the operation history information of the first user, indicating the possibility of a circadian rhythm sleep-wake disorder in which the circadian rhythm is a non-24-hour sleep-wake rhythm disorder (Straight). A non-24-hour sleep-wake rhythm disorder (Straight) is a circadian rhythm sleep-wake disorder in which the sleep-wake rhythm is delayed by about one hour each day. Straight means that the delay interval is constant and appears linear in the image.

[0045] For example, image G501 shows the operation history information of the first user, indicating the possibility of a circadian rhythm sleep-wake disorder called a non-24-hour sleep-wake rhythm disorder (Jump). A non-24-hour sleep-wake rhythm disorder (Jump) is a circadian rhythm sleep-wake disorder in which the sleep-wake rhythm is delayed by about one hour each day. A Jump is a delay in time that is not constant and appears as a curve in the image.

[0046] For example, image G601 shows the operation history information of the first user, indicating the possibility of a circadian rhythm sleep-wake disorder in which the circadian rhythm is a mixed disorder (24-hour sleep-wake + non-24-hour sleep-wake rhythm disorder). A mixed disorder (24-hour sleep-wake + non-24-hour sleep-wake rhythm disorder) is a circadian rhythm sleep-wake disorder in which the circadian rhythm is a mixture of a 24-hour sleep-wake rhythm disorder and a non-24-hour sleep-wake rhythm disorder.

[0047] For example, image G701 shows the operation history information of the first user, and indicates that the circadian rhythm is shift work. Shift work here refers to working day shifts, night shifts, night shifts, and other shifts over a certain period of time. Shift work is an artificial circadian rhythm, but if you experience insomnia or excessive sleepiness, you may have shift work disorder, a circadian rhythm sleep-wake disorder.

[0048] For example, image G801 shows the operation history information of the first user, indicating that the circadian rhythm may be irregular sleep. Irregular sleep refers to when the bedtime and wake-up times are not consistent almost every day, or when the bedtime and wake-up times cannot be estimated. If the sleep is classified as irregular and the user also experiences insomnia or excessive sleepiness, there is a possibility that the user has an irregular sleep-wake rhythm disorder, a type of circadian rhythm sleep-wake disorder.

[0049] <Diagram showing an example of the hardware configuration of the sleep evaluation device 1> 9 is a diagram showing an example of the hardware configuration of the sleep evaluation device 1 of the first embodiment. The sleep evaluation device 1 includes a control unit 11 and executes a program. By executing the program, the sleep evaluation device 1 functions as a device including the control unit 11, an interface unit 12, and a storage unit 13.

[0050] More specifically, the processor 91 reads out a program stored in the storage unit 13 and stores the read out program in the memory 92. The processor 91 executes the program stored in the memory 92, causing the sleep evaluation device 1 to function as a device including the control unit 11, the interface unit 12, and the storage unit 13.

[0051] The control unit 11 controls the operation of the various functional units included in the sleep evaluation device 1. The control unit 11 executes, for example, the input acquisition process, machine learning process, classification process, and notification control process described above. The control unit 11 records, for example, various pieces of information generated by the execution of various processes in the memory unit 13. The control unit 11 acquires, for example, information stored in the memory unit 13. Acquiring information recorded in the memory unit 13 is a process of reading information stored in the memory unit 13 from the memory unit 13.

[0052] The interface unit 12 includes an input unit 121 and an alarm 122. The input unit 121 receives, for example, at least operation history information from an external device. In other words, the input unit 121 receives at least operation history information input from the external device. Therefore, the input unit 121 is a communication interface that is communicatively connected via wire or wireless to at least an external device that is a source of the operation history information. Such an external device may be any device that is a source of the operation history information. For example, the external device may be the operation target itself, a specific terminal such as a smartphone, a personal computer, or smart glasses, a home appliance such as a refrigerator, a microwave, an air conditioner, a television, or an electric fan, a remote control, or a switch that turns on and off lights in a room. Furthermore, such an external device may be, for example, a server that stores the execution history of a program such as an interactive app or an electronic bulletin board that is the operation target.

[0053] The input unit 121 may also be communicatively connected to, for example, a device that transmits circadian rhythms labeled with operation history information. In such a case, the input unit 121 acquires the operation history information and the circadian rhythms labeled therewith by communicating with an external device that transmits the operation history information and the circadian rhythms labeled therewith. The device that transmits such data may be, for example, a server or a terminal such as a personal computer. Note that the training data refers to data in which the operation history information is paired with the circadian rhythm classification labeled with the operation history information.

[0054] The input unit 121 may be a communication interface communicably connected to the transmission source device. In such a case, the input unit 121 acquires pairs of operation history information and circadian rhythms through wired or wireless communication with the transmission source device. The transmission source device may be, for example, a server or a terminal such as a personal computer operated by the creator of the training data.

[0055] When the input unit 121 acquires operation history information in which a labeled circadian rhythm exists, the circadian rhythm classification model is trained using the circadian rhythm and the operation history information. When the input unit 121 acquires operation history information in which a labeled circadian rhythm does not exist, classification processing is performed instead of training.

[0056] The alarm 122 (an example of an alarm unit) is configured to include a display device such as a CRT (Cathode Ray Tube) display, a liquid crystal display, an organic EL (Electro-Luminescence) display, etc. The alarm 122 may be configured to include a sound output device such as a speaker instead of such a display device, or may be configured to include both a display device and a sound output device.

[0057] The alarm 122 may output, for example, information input to the input unit 121 by an image or sound. The alarm 122 may also output, for example, the result of processing by the control unit 11 by an image or sound. Such an alarm 122 is an example of an alarm that is controlled by the control unit 11 and notifies the second user of the circadian rhythm classified by the classification process.

[0058] The storage unit 13 is configured using a non-transitory computer-readable recording medium such as a magnetic hard disk drive or semiconductor storage device. The storage unit 13 stores various information related to the sleep evaluation device 1. The storage unit 13 stores information input via the interface unit 12, for example. The storage unit 13 stores various information generated by the operation of the control unit 11, for example. Note that the storage unit 13 does not necessarily have to be included in the sleep evaluation device 1, and may exist, for example, on the cloud as long as the control unit 11 can communicate with it.

[0059] <Diagram showing an example of the configuration of the control unit 11 included in the sleep evaluation device 1> 10 is a diagram showing an example of the configuration of the control unit 11 included in the sleep evaluation device 1 of the first embodiment. The control unit 11 includes an input acquisition unit 101, a machine learning unit 102, a classification unit 103, and a notification control unit 104. The input acquisition unit 101 executes input acquisition processing. The machine learning unit 102 executes machine learning processing. The classification unit 103 executes classification processing. The notification control unit 104 executes notification control processing.

[0060] 11 is a flowchart showing an example of the flow of processing executed by the sleep evaluation device 1 of the first embodiment. Operation history information and labeled circadian rhythms are input to the input unit 121 (step S101). Here, the operation history information input includes operation history information of a first user and operation history information of a second user. Note that step S101 is an example of an input step in which operation history information indicating operation history is input from an external device.

[0061] When teacher data consisting of operation history information and circadian rhythms labeled therefor is input to the input unit 121, the machine learning unit 102 executes machine learning processing based on the teacher data consisting of the input operation history information of the first user and the circadian rhythms labeled therefor (step S102). Note that in step S101, the circadian rhythm paired with the operation history information of the first user does not necessarily have to be input from the input unit 121; for example, it may be one that has been stored in advance in a predetermined storage device such as the storage unit 13, or it may be acquired from a predetermined external device.

[0062] That is, the machine learning unit 102 performs supervised learning of the circadian rhythm classification model based on training data consisting of the input operation history information of the first user and the circadian rhythms labeled therefor. Note that step S102 is an example of a machine learning process that performs supervised learning of the circadian rhythm classification model, which is a model that classifies circadian rhythms, based on training data consisting of the input operation history information of the first user and the circadian rhythms labeled therefor.

[0063] If the input unit 121 has operation history information but no circadian rhythm is labeled therewith, the classification unit 103 executes classification processing based on the operation history information of the second user (step S103). That is, the classification unit 103 classifies the circadian rhythm of the second user based on the input operation history information of the second user and the circadian rhythm classification model. Note that step S103 is an example of a classification process that classifies the circadian rhythm of the second user based on the input operation history information of the second user and the circadian rhythm classification model.

[0064] Next, the notifier 122 notifies the second user of the circadian rhythm classified by the classification unit 103 (step S104). Note that step S104 is an example of a notifying step of notifying the second user of the circadian rhythm classified by the classification step.

[0065] Note that either step S102 or step S103 is performed depending on whether or not a circadian rhythm is labeled in the operation history information. Also, step S104 may be performed after step S103 is performed.

[0066] The sleep evaluation device 1 configured in this manner executes classification processing, and therefore, as described in the above-mentioned <Examples of Effects Produced by the Control Unit 11>, the sleep evaluation device 1 can identify symptoms that may be circadian rhythm sleep-wake disorders, thereby creating an opportunity for the patient to visit a medical institution early.

[0067] (Second embodiment) 12 is a diagram showing an example of the hardware configuration of an information processing device 2 (an example of a learning device) in the second embodiment. The information processing device 2 includes a control unit 21 including a processor 93 such as a CPU, GPU, or NPU, and a memory 94, which are connected via a bus, and executes a program. By executing the program, the information processing device 2 functions as a device including the control unit 21, an interface unit 22, and a storage unit 23.

[0068] More specifically, the processor 93 reads out a program stored in the storage unit 23 and stores the read out program in the memory 94. The processor 93 executes the program stored in the memory 94, whereby the information processing device 2 functions as a device including the control unit 21, the interface unit 22, and the storage unit 23.

[0069] The control unit 21 controls the operation of various functional units included in the information processing device 2. The control unit 21 executes, for example, machine learning processing. The definition of the machine learning processing is the same as that of the machine learning processing in the first embodiment. The definition of the operation history information is the same as that in the first embodiment. The definition of the circadian rhythm is also the same as that in the first embodiment. Therefore, the definition of the training data used to train the circadian rhythm classification model is also the same as that in the first embodiment. Also, when the operation history information is represented by an image, the image may be an image that satisfies the image conditions described in the first embodiment, similar to the first embodiment.

[0070] The control unit 21 acquires, for example, information stored in the storage unit 23. Acquiring the information recorded in the storage unit 23 is a process of reading out the information stored in the storage unit 23 from the storage unit 23.

[0071] The interface unit 22 includes a communication interface for connecting the information processing device 2 to an external device. The interface unit 22 communicates with the external device via a wired or wireless connection.

[0072] The external device is, for example, a device that transmits signals or information used for learning. The interface unit 22 acquires the signals or information used for learning by communicating with the device that transmits the signals or information used for learning. The signals or information used for learning are, for example, teacher data. In other words, the signals or information used for learning are, for example, a set of operation history information and a circadian rhythm labeled therewith.

[0073] The external device may be, for example, a classification device. The classification device is a device that classifies circadian rhythms (i.e., performs classification processing) using the trained circadian rhythm classification model obtained by the information processing device 2. In such a case, the classification device can execute the trained circadian rhythm classification model obtained by the information processing device 2 through communication via the interface unit 22. The classification device may output the results of the classification processing to a predetermined output destination.

[0074] The interface unit 22 may be configured to include input devices such as a mouse, a keyboard, a touch panel, etc. The interface unit 22 may be configured as an interface that connects these input devices to the information processing device 2. In this way, the input devices of the interface unit 22 accept input of various information to the information processing device 2 via wired or wireless connections. Note that signals or information do not necessarily have to be input to the communication interface of the interface unit 22, but may also be input to the input devices of the interface unit 22.

[0075] The interface unit 22 outputs, for example, various types of information. The interface unit 22 includes, for example, a display device such as a CRT display, a liquid crystal display, or an organic EL display, and a speaker. The interface unit 22 may be configured as an interface that connects these display devices or speakers to the information processing device 2. Therefore, the interface unit 22 may output, for example, information input to an input device of the interface unit 22 as an image or sound.

[0076] The storage unit 23 is configured using a computer-readable storage medium device such as a magnetic hard disk drive or a semiconductor storage device. The storage unit 23 stores various information related to the information processing device 2. The storage unit 23 stores information input via the interface unit 22, for example. The storage unit 23 stores various information generated by the operation of the control unit 21, for example. Note that the storage unit 23 does not necessarily have to be provided in the information processing device 2, and may exist, for example, on a cloud as long as the control unit 21 can communicate with it.

[0077] 13 is a flowchart showing an example of the flow of processing executed by the information processing device 2 in the second embodiment. The control unit 21 executes machine learning processing (step S201).

[0078] The control unit 21 may perform classification processing. The classification processing performed by the control unit 21 is similar to the classification processing in the first embodiment. In such a case, the control unit 21 may control the operation of a predetermined output destination, such as the interface unit 22, to output the results of the classification processing.

[0079] Therefore, the interface unit 22 of the information processing device 2 may further include, for example, an alarm 122. In such a case, the control unit 21 may control the operation of the alarm 122 to notify the result of the classification process.

[0080] <Examples of Effects of Information Processing Device 2> The information processing device 2 includes a control unit 21 that executes a learning process. The circadian rhythm classification model, which is a model learned in the learning process, is a model that classifies the circadian rhythm of a circadian rhythm classification target (i.e., a user). The learning process improves the accuracy of circadian rhythm classification. If users can know with higher accuracy what their circadian rhythms are, they are more likely to review their lifestyle habits and visit a medical institution. As a result, the possibility of improving circadian rhythm sleep-wake disorders increases. Therefore, such an information processing device 2 can increase the possibility of improving circadian rhythm sleep-wake disorders. In other words, such an information processing device 2 can solve the problem of providing a technology that increases the possibility of improving circadian rhythm sleep-wake disorders.

[0081] The information processing device 2 of this modified example configured as described above includes a control unit 21 that executes a learning process. Using the trained model obtained by executing the learning process, it becomes possible to identify symptoms that may be circadian rhythm sleep-wake disorders. Using the trained circadian rhythm classification model to identify symptoms that may be related to circadian rhythms can create an opportunity for the patient to visit a medical institution. Furthermore, it is possible to eliminate the effort and time required for machine learning.

[0082] Furthermore, the modified information processing device 2 configured in this manner is equipped with a control unit 21 that executes learning processing, and therefore can provide technology that increases the possibility of improving circadian rhythm sleep-wake disorders, as described in <Examples of effects achieved by the information processing device 2>.

[0083] The sleep evaluation device 1 may be implemented using multiple information processing devices connected to each other via a network so that they can communicate with each other. In this case, the processes executed by the sleep evaluation device 1 may be distributed among the multiple information processing devices.

[0084] The information processing device 2 may be implemented using a plurality of information processing devices communicably connected via a network. In this case, each process executed by the information processing device 2 may be distributed and executed by the plurality of information processing devices.

[0085] All or part of the functions of the sleep evaluation device 1 and the information processing device 2 may be realized using hardware such as an ASIC (Application Specific Integrated Circuit), a PLD (Programmable Logic Device), or an FPGA (Field Programmable Gate Array). The program may be recorded on a computer-readable recording medium. Examples of computer-readable recording media include portable media such as flexible disks, magneto-optical disks, ROMs, and CD-ROMs, and storage devices such as hard disks built into computer systems. The program may be transmitted via a telecommunications line.

[0086] The first and second embodiments of the present invention have been described above in detail with reference to the drawings, but the specific configurations are not limited to the first and second embodiments, and also include designs within the scope of the present invention that do not deviate from the gist of the present invention. [Explanation of symbols]

[0087] REFERENCE SIGNS LIST 1... sleep evaluation device, 11... control unit, 12... interface unit, 13... memory unit, 121... input unit, 122... alarm, 101... input acquisition unit, 102... machine learning unit, 103... classification unit, 104... alarm control unit, 2... information processing device, 21... control unit, 22... interface unit, 23... memory unit, 91... processor, 92... memory, 93... processor, 94... memory

Claims

1. an input unit that inputs operation history information, which is information indicating an operation history, and a circadian rhythm labeled with the operation history information from an external device; a machine learning unit that performs supervised learning of a circadian rhythm classification model, which is a model that classifies circadian rhythms, based on training data consisting of the input operation history information of the first user and circadian rhythms labeled with the operation history information; a classification unit that classifies the circadian rhythm of the second user based on the input operation history information of the second user and the circadian rhythm classification model; a notification unit that notifies the second user of the circadian rhythm classified by the classification unit; A sleep evaluation device comprising:

2. When the operation history information is data representing a date and time, the machine learning unit plots points indicating the dates and times when the operations were performed on a white background image consisting of pixels in a predetermined range, and uses the plotted white background image as the operation history information. The sleep evaluation device according to claim 1 .

3. In the white background image, the coordinates of one of the vertical and horizontal axes indicate the date, and the coordinates of the other axis indicate the time, The time interval from the time indicated at one end of the other axis to the time indicated at the other end is n times 24 hours (n is an integer of 1 or more). The sleep evaluation device according to claim 2 .

4. A computer as the sleep evaluation device according to any one of claims 1 to 3, a machine learning unit that performs supervised learning of a circadian rhythm classification model, which is a model that classifies circadian rhythms, based on training data consisting of the input operation history information of the first user and circadian rhythms labeled with the operation history information; A program for realizing the functions of a classification unit that classifies the circadian rhythm of the second user based on the input operation history information of the second user and the circadian rhythm classification model.

5. an input step of inputting operation history information, which is time-series data indicating an operation history, and a circadian rhythm labeled with the operation history information from an external device; a machine learning process for performing supervised learning of a circadian rhythm classification model, which is a model for classifying circadian rhythms, based on training data consisting of the input operation history information of the first user and circadian rhythms labeled with the operation history information; a classification step of classifying the circadian rhythm of the second user based on the input operation history information of the second user and the circadian rhythm classification model; a notification step of notifying the second user of the circadian rhythm classified by the classification step; A sleep evaluation method comprising:

6. a control unit that learns a circadian rhythm classification model that is a model for classifying the circadian rhythm of a user who operates the device, based on training data that includes operation history information that is time-series data indicating an operation history output from an external device and circadian rhythms labeled with the operation history information; A learning device comprising:

Images