Exercise heart rate estimation system and exercise heart rate estimation method

Abstract

Provided is an exercise heart rate estimation system comprising: an acquisition unit that acquires biological pulse wave information and activity information in temporal correspondence with each other; a first pulse rate acquisition unit that acquires a first pulse rate on the basis of the pulse wave information; an activity information analysis unit that acquires a feature intensity of a frequency component of the activity information and a feature frequency corresponding to the feature intensity; a second pulse rate acquisition unit that acquires a second pulse rate on the basis of remaining components excluding the feature frequency and frequency components in the vicinity thereof from among the frequency components of the pulse wave information; and a heart rate estimation unit that estimates an exercise heart rate on the basis of the first pulse rate, the second pulse rate, the feature intensity of the frequency component of the activity information, and the feature frequency.

Description

Exercise Heart Rate Estimation System and Exercise Heart Rate Estimation Method

[0001] [Related Application] This application claims priority from Japanese Patent Application No. 2025-006740, titled "Exercise Heart Rate Estimation System and Exercise Heart Rate Estimation Method," filed on January 17, 2025, the disclosure of which is incorporated herein by reference in its entirety. The present invention relates to an exercise heart rate estimation system and an exercise heart rate estimation method.

[0002] As background art in this technical field, Japanese Patent No. 6919959 (Patent Document 1) can be cited. This publication states that "in an information processing system in which a user terminal device connected to a measurement device and a management server are connected via a network, the management server 100 includes a biological data generation unit 132 and a user support data generation unit 133. The biological data generation unit 132 executes a predetermined calculation on the measurement data and generates biological data including at least either maximum oxygen uptake information or heart rate information. The user support data generation unit 133 compares at least either the maximum oxygen uptake information with reference maximum oxygen uptake information or the heart rate information with target heart rate information, and generates user support data regarding changes in the user's exercise load or rest amount according to the result of the comparison." (See the abstract).

[0003] Japanese Patent No. 6919959

[0004] Reiss, Attila, et al. "Deep PPG: Large-scale heart rate estimation with convolutional neural networks.(ディープPPG:畳み込みニューラルネットワークによる大規模心拍数推定)" Sensors 19.14 (2019): 3079

[0005] As shown in Patent Document 1, with the spread of wearable devices such as smart rings and smartwatches, it has become relatively easy to acquire biometric information, such as pulse rate. Furthermore, in order to efficiently obtain the effects of exercise in health management or exercise therapy, aerobic exercise is recommended, and there is a need to acquire heart rate during exercise more easily and accurately. However, the less burdensome a wearable device is on the body, the more likely it is that the biometric information it can acquire will contain artifacts caused by physical movement. This tendency can be a particular challenge when it comes to acquiring heart rate, for example.

[0006] This technology provides a new mechanism for estimating heart rate during exercise.

[0007] To solve the above problems, for example, the configuration described in the claims is adopted. The present application includes multiple means for solving the above problems, but to give one example, it is characterized by providing an exercise heart rate estimation system comprising: an acquisition unit that acquires pulse wave information and activity information of a living organism in a temporal correspondence; a first pulse rate acquisition unit that acquires a first pulse rate based on the pulse wave information; an activity information analysis unit that acquires the characteristic intensity of the frequency components of the activity information and the characteristic frequency corresponding to said characteristic intensity; a second pulse rate acquisition unit that acquires a second pulse rate based on the remaining components of the frequency components of the pulse wave information, excluding the characteristic frequency and the frequency components in its vicinity; and a heart rate estimation unit that estimates the exercise heart rate based on the first pulse rate, the second pulse rate, the characteristic intensity of the frequency components of the activity information, and the characteristic frequency.

[0008] While "heart rate" measured from an electrocardiogram and "pulse rate" measured from pulse wave information are strictly distinguished, in a normal state without arrhythmias, heart rate and pulse rate coincide. Therefore, in this technology, "heart rate" can be understood as a concept that includes "pulse rate."

[0009] According to the present invention, a new mechanism for estimating heart rate during exercise can be provided. Other problems, configurations, and effects will be clarified by the following description of embodiments.

[0010] Figure 1 shows an example configuration of an exercise heart rate estimation system 100 according to one embodiment. Figure 2 shows an example of the hardware configuration of a wearable device 101. Figure 3 shows an example of the hardware configuration of a user terminal 102. Figure 4 shows an example of the hardware configuration of a management server 103. Figure 5 shows an example of health management information according to one embodiment. Figure 6 shows an example of other health management information according to one embodiment. Figure 7 is a flowchart of exercise heart rate estimation according to one embodiment. Figure 8 is a graph showing the relationship between a second pulse rate for reference and a medical heart rate obtained based on a medical device. Figure 9 is a graph illustrating the relationship between the estimated exercise heart rate according to one embodiment and a medical heart rate obtained based on a medical device. Figures 10(A) to (C) are schematic graphs to explain an example of calculating a first heart rate for reference.

[0011] The following description will be based on an exercise heart rate estimation system and exercise heart rate estimation method according to one embodiment, with reference to the drawings as appropriate. In the drawings, components having the same function may not be denoted by reference numerals or described again.

[0012] [Exercise Heart Rate Estimation System and Exercise Heart Rate Estimation Method] Figure 1 shows an example configuration of an exercise heart rate estimation system 100 according to one embodiment. The exercise heart rate estimation system 100 and exercise heart rate estimation method according to this technology are typically a system and method for estimating the exercise heart rate by monitoring the user's biometric information using a wearable device 101. The exercise heart rate estimation system 100 includes one or more wearable devices 101, one or more user terminals 102, and one or more management servers 103. The exercise heart rate estimation system 100 may also include one or more chargers 104.

[0013] In one embodiment, the wearable device 101, user terminal 102, management server 103, and charger 104 are configured to send and receive information from each other via a network, for example. Furthermore, the wearable device 101, user terminal 102, and charger 104 are configured to connect via short-range wireless communication such as Bluetooth®. These wearable devices 101, user terminal 102, management server 103, and charger 104 may be configured to send and receive information from each other via a network, for example, but only among themselves. Also, any two or more of the functional elements of the wearable device 101, user terminal 102, management server 103, and charger 104 may be integrated, or any of them may be separated into two or more units.

[0014] The wearable device 101 is a terminal that a user (an example of a living organism) wears on their own body. The user terminal 102 is a terminal used by the user of the wearable device 101 or the exercise heart rate estimation system 100. The management server 103 is a terminal used by the administrator who manages the exercise heart rate estimation system 100.

[0015] Each terminal and the management server 103 in the exercise heart rate estimation system 100 may be a mobile device such as a smartphone, tablet, mobile phone, or personal digital assistant (PDA), or a wearable device such as glasses (including goggles), a wristwatch, or clothing. The management server and terminals may also be a stationary or portable computer, or a server located in the cloud or on a network. From a functional standpoint, they may also be VR (Virtual Reality) terminals, AR (Augmented Reality) terminals, or MR (Mixed Reality) terminals. Alternatively, the management server and terminals may be a combination of multiple such terminals. For example, a combination of one smartphone and one wearable device can logically function as a single terminal. Each terminal may also be an information processing terminal other than those mentioned above.

[0016] Each terminal and management server 103 of the exercise heart rate estimation system 100 may optionally be equipped with a processor that executes an operating system, applications, and programs; a main memory such as RAM (Random Access Memory); an auxiliary storage device such as an IC card, hard disk drive, SSD (Solid State Drive), and flash memory; a communication control unit such as a network card, wireless communication module, or mobile communication module; an input device such as a touch panel, keyboard, mouse, voice input device, motion controller, or motion detection device using image capture from a camera; an output device such as a monitor, display, printer, audio output device, or oscillator; and a timing device. The input device may also be equipped with sensors such as GPS, gyro sensors, and acceleration sensors. The output device may be a device or terminal that transmits information for output to an external monitor, display, printer, or other device.

[0017] The main memory stores various programs and applications (software modules), and the processor executes these programs and applications to realize each functional element of the overall system. These modules may be implemented by one or more programs or applications. Furthermore, each module may be implemented by an independent program or application, or as a subprogram or function within a single integrated program or application. These modules may also be implemented in hardware (hardware modules) by integrating circuits or employing a microcomputer.

[0018] Furthermore, each module may be implemented by a single processor or by multiple processors. Also, each module may be installed in a single terminal (including a management server) or divided among two or more terminals (including management servers) interconnected via a network. Furthermore, each module may be installed in each, or one or more, of the two or more terminals (including management servers) interconnected via a network. Consideration is also given to cases where some modules are implemented in different countries from others.

[0019] In this specification, each module is described as the entity (subject) that performs the processing; however, in reality, processing is carried out by the processor executing various programs and applications.

[0020] The auxiliary storage device stores various databases (DBs). A "database" is a collection of data organized and gathered to accommodate arbitrary data operations (e.g., extraction, addition, deletion, overwriting, etc.) from a processor or external computer. The auxiliary storage device is a functional element (storage unit) that stores one or more data collections. The implementation method of a database is not limited; for example, it may be a database management system, spreadsheet software, or text files such as XML or JSON. The database may be independently provided and connectable to a processor, etc. The database is intended, but not limited to, storing some or all of the information constituting the data collection in JSON format files. The database may be configured to store various types of information as a relational database or a non-relational database.

[0021] [Management Server] The management server 103 is an additional element in this technology. Figure 4 shows an example of the hardware configuration of the management server 103. The management server 103 is composed of a computer server located on the cloud, for example. The management server 103 includes a main memory 401 and an auxiliary memory 402. The management server 103 also includes the processor 403 as described above, an input device 404, an output device 405, and a communication control unit 406.

[0022] The main memory 401 stores programs and applications such as the execution module 411, health management module 412, health information evaluation module 413, pre-processing module 414, and post-processing module 415. The processor 403 executes these programs and applications to realize each functional element of the management server 103.

[0023] The execution module 411 controls the basic operation of the management server 103 for providing services by the exercise heart rate estimation system 100. For example, the execution module 411 works in conjunction with the execution module 311 of the user terminal 102 to control the basic operation for causing the user terminal 102 to perform the exercise heart rate estimation service. The execution module 411 can also support the operation of each module, as described later, and the coordination between each module. The execution module 411 may also be configured to work in conjunction with the execution module 211 of the wearable device 101, for example.

[0024] The health management module 412 acquires the user's biometric detection information. The health management module 412 is an example of an acquisition unit in this technology. The health management module 412 also provides health management information to the user terminal 102. For example, the health management module 412 acquires the user's biometric detection information using the wearable device 101, and outputs the results of the user's health evaluation, which is evaluated by the health information evaluation module 413 based on the acquired biometric detection information of the user, to the user terminal 102.

[0025] The health information evaluation module 413 evaluates the user's health status. The health information evaluation module 413 estimates and evaluates the user's health status based on various biometric detection information of the user acquired by the wearable device 101, for example. In this embodiment, the health information evaluation module 413 estimates the user's heart rate during exercise, for example. The health information evaluation module 413 estimates the user's heart rate during exercise using a heart rate estimation model, for example. The health information evaluation module 413 is an example of a heart rate estimation unit in this technology. Hereinafter, the exercise heart rate estimated by the health information evaluation module 413 may simply be referred to as the "estimated heart rate."

[0026] The heart rate estimation model M is a machine learning model trained to take as input the feature intensity of the frequency components of the learning activity information of a living organism, the feature frequency corresponding to said feature intensity, a third pulse rate obtained based on the learning pulse wave information of the living organism, and a fourth pulse rate obtained based on the remaining components of the frequency components of the learning pulse wave information excluding the feature frequency and its vicinity, and to output the medical heart rate, which is the heart rate of the living organism obtained in correspondence with the learning activity information and the learning pulse wave information.

[0027] The analysis algorithm in machine learning is not particularly limited and may include, for example, supervised learning, unsupervised learning, reinforcement learning, deep learning, etc. In this embodiment, a supervised learning method is used as the analysis algorithm. The supervised learning method may be, for example, a regression method, a classification method, or a combination of these. The supervised learning method is not limited to this, but for example it may be a classification method, and more specifically, at least one of the following can be used: gradient descent, ensemble learning (boosting), or decision tree. In this embodiment, as the learning algorithm for the heart rate estimation model M, for example, a gradient boosting decision tree (GBDT) method, which combines these three methods, is used. Examples of GBDTs include histogram-based and pre-sorted, and for example, a histogram-based gradient boosting decision tree (HGBDT) method may be used. By employing a histogram-based method (e.g., LightGBM) as the algorithm for creating decision trees, efficient estimation can be achieved.

[0028] The training data, which includes the feature intensity of the frequency components of learning activity information, the feature frequency corresponding to said feature intensity, a third pulse rate obtained based on the learning pulse wave information of a living organism, and a fourth pulse rate obtained based on the remaining components of the frequency components of the learning pulse wave information, excluding the feature frequency and its vicinity, will be described later.

[0029] The preprocessing module 414 is an element that processes (preprocesses) various biometric detection information of the user prior to inputting it to the health information evaluation module 413. For example, the preprocessing module 414 preprocesses the various biometric detection information of the user, which is the input, so that the health information evaluation module 413 can perform a more appropriate evaluation (including estimation). In this embodiment, the preprocessing module 414 expresses the heart rate in the frequency domain by performing time-frequency analysis on the pulse wave information of the body. Furthermore, in order to remove artifacts from this, the preprocessing module 414 prepares activity information expressed in the frequency domain.

[0030] The preprocessing module 414 acquires a first pulse rate based on, for example, the pulse wave information of a living organism. The preprocessing module 414 also acquires, for example, the characteristic intensity of the frequency components of the activity information and the characteristic frequency corresponding to that characteristic intensity, based on the activity information of the living organism. The preprocessing module 414 further acquires a second pulse rate based on the remaining components of the pulse wave information, excluding the characteristic frequency and its vicinity. The preprocessing module 414 acquires the first pulse rate, characteristic intensity, characteristic frequency, and second pulse rate as input information for, for example, the health information evaluation module 413. In other words, the preprocessing module 414 is an example of a first pulse rate acquisition unit, an activity information analysis unit, and a second pulse rate acquisition unit in this technology.

[0031] The post-processing module 415 is an element that performs additional processing (post-processing) on ​​the output of the health information evaluation module 413. For example, the post-processing module 415 performs post-processing on the estimated heart rate output of the health information evaluation module 413 so that the user can better understand the output. For example, the post-processing module 415 obtains the evaluation result of the exercise heart rate based on the estimated heart rate.

[0032] The auxiliary storage device 402 stores various types of information necessary for realizing the above-mentioned functions of the management server 103. For example, the auxiliary storage device 402 stores user information 410, health management information 420, heart rate estimation model information 430, etc. The implementation method of each type of information stored in the auxiliary storage device 402 is not limited to this embodiment, and each type of information may be implemented in a distributed manner across multiple database servers or cloud applications.

[0033] User information 410 is, for example, information about a user who uses the exercise heart rate estimation system 100. User information 410 is typically information about a user who uses the wearable device 101. User information 410 is not limited to this, but includes, for example, user identification information, information about the user's attributes such as gender and age, and information from a health questionnaire about the user. Each piece of information is stored, for example, linked to user identification information.

[0034] Figures 5 and 6 show examples of health management information 420 according to one embodiment. Health management information 420 is information acquired, calculated, or estimated by the exercise heart rate estimation system 100. The figures illustrate, for example, information of the health management information 420 that is stored in an auxiliary storage device. Health management information 420 may include, for example, a first pulse rate HR1 based on pulse wave information as shown in (A), a feature intensity I of the frequency components of activity information and a feature frequency F corresponding to the feature intensity as shown in (B), and a second pulse rate HR2 based on the remaining components after excluding the feature frequency of the pulse wave information and its vicinity frequency components as shown in (C). Health management information 420 may also include, for example, an estimated heart rate HRex (exercise heart rate) as shown in (D). Each of these pieces of information is associated with, for example, time (time) information. The contents of each piece of information will be described later. The auxiliary storage device can store information other than the information shown in the figures.

[0035] The heart rate estimation model information 430 is information necessary for using the heart rate estimation model M in the exercise heart rate estimation system 100. The heart rate estimation model information 430 may include, for example, the entire heart rate estimation model M, i.e., the model function and parameters and the program for operating it, or it may include parameters for operating a machine learning model stored on an external server or the like.

[0036] [User Terminal] Figure 3 shows an example of the hardware configuration of the user terminal 102. The user terminal 102 is composed of a terminal such as a smartphone, tablet, notebook PC, or desktop PC. The user terminal 102 includes a main memory 301 and an auxiliary memory 302. The user terminal 102 also includes a processor 303 as described above, an input device 304, an output device 305, a camera 306, and a communication control unit 307.

[0037] The main memory 301 of the user terminal 102 stores programs and applications such as the execution module 311, health management module 312, health information evaluation module 313, pre-processing module 314, and post-processing module 315. The processor 303 executes these programs and applications to realize each functional element of the user terminal 102. In other words, the main memory 301 of the user terminal 102 stores the same functional modules as the main memory 401 of the management server 103, and the user terminal 102 can perform the same processing as the management server 103.

[0038] The functions of each of these modules are the same as described above, so their explanation will be omitted. Note that the main memory 301 does not necessarily have to store all of the above-mentioned functional modules; it may store only some of them (for example, the execution module 311, the health management module 312, and the post-processing module 315).

[0039] As will be described later, the main memory 301 of the user terminal 102 stores a health management application program that manages the user's health based on sensor data acquired from the ring-shaped wearable device 101. This health management application program includes, for example, an exercise heart rate estimation program that estimates the exercise heart rate and an evaluation program for evaluating the exercise heart rate.

[0040] The auxiliary storage device 302 stores various types of information necessary for realizing the above-mentioned functions of the user terminal 102. The auxiliary storage device 302 can store, for example, user information 310, health management information 320, heart rate estimation model information 330, etc. Each of these pieces of information may be, for example, a part of the information stored in the management server 103 (for example, a first pulse rate HR1, a feature intensity I of the frequency component of the activity information, a feature frequency F corresponding to this feature intensity, a second pulse rate HR2, an estimated heart rate (exercise heart rate), etc.).

[0041] [Wearable Device] The wearable device 101 is a device that can be attached to a living body such as a human being, thereby non-invasively acquiring the biological information of the living body as digital information. The wearable device 101 according to this embodiment is a ring-shaped device (wearable ring) that is attached to a human finger (typically one finger), as shown in Figure 1. Figure 2 is an example of the hardware configuration of the wearable device 101 according to one embodiment.

[0042] The wearable device 101 includes a main memory 201 and an auxiliary memory 202. The management server 103 also includes the processor 203 as described above, a sensor module 204, a gyro sensor / accelerometer module 205, a charge management module 206, and a communication control unit 207. The sensor module 204 and the gyro sensor / accelerometer module 205 are examples of input devices.

[0043] The processor 203 is, for example, a digital signal processing device that controls the operations of the sensor module 204, the gyro sensor - accelerometer module 205, the charging management module 206, and the communication control unit 207. As the processor 203, for example, a MCU (Micro Controller Unit), a MPU (Micro Processor Unit), etc. can be provided. As the storage device, for example, a flash memory or an EEP - ROM (Electrically Erasable Programmable Read - Only Memory) can be provided. The wearable device 101 may include an output device not shown in the figure.

[0044] The wearable device 101 may or may not include a planar display device such as a display capable of displaying character information and image information, for example. The ring - type wearable device 101 of the present embodiment includes, for example, a light - emitting part each consisting of a single LED lamp, but does not include a display device capable of displaying character information and image information.

[0045] The sensor module 204 controls, for example, the operation of various sensors provided in the wearable device 101 and the sensor signals. The sensor module 204 also controls, for example, the operating conditions of the sensors. The sensor module 204 may also include digital processing functions such as a Digital Signal Controller (DSC), Digital Signal Processor (DSP), or FPGA (Field Programmable Gate Array) that process sensor data acquired by the sensors in real time. The sensor module 204 includes, for example, any combination of resistors, capacitors, coils, diodes, transistors, etc., and can optionally include circuits such as amplifiers, clocks, bandgap references, digital isolators, RC filters (e.g., low-pass filters, high-pass filters), band-pass filters, band-elimination filters, and analog-to-digital converters. Furthermore, the sensor module 204 may include analog circuit blocks such as an RMS-DC converter circuit, an average value detection circuit, an RMS detection circuit, a signal monitoring circuit, and an adaptive threshold setting circuit that process signals acquired via the sensors. Furthermore, the sensor module 204 can be configured to perform various signal processing operations using any combination of analog circuits and / or digital processing functions.

[0046] The wearable device 101 includes, for example, a pulse wave sensor S1, a temperature sensor S2, a microwave sensor S3, an electrocardiogram sensor S4, etc., and the sensor module 204 is connected to these sensors S1 to S4. These sensors S1 to S4 are, for example, examples of signal detection devices (sensors) for detecting various kinds of biological information, and are examples of input devices for the wearable device 101. Each sensor S1 to S4 may also be equipped with a digital signal controller (DSC) that processes sensor data in real time.

[0047] The pulse wave sensor S1 is a Photoplethysmogram (PPG) sensor that detects volume changes in a blood vessel (e.g., artery) of a living body. In the ring-type wearable device 101, the pulse wave sensor S1 can typically be a reflective pulse wave sensor composed of a light-emitting element that emits light of a predetermined wavelength toward the living body and a light-receiving element that detects the light reflected within the living body. The light-emitting element can be constituted by, for example, various Light-emitting diodes (LEDs), and the light-receiving element can be constituted by, for example, various Photo Diodes (PDs).

[0048] By analyzing the PPG signal acquired by the PPG sensor, various biological information can be obtained. Specifically, for example, the light-emitting element can include a green LED that generates light with a central wavelength of 500 nm or more and 600 nm or less. Since green light has a high absorption rate by hemoglobin in the blood and is less affected by ambient light such as sunlight, a relatively stable volume pulse wave can be measured. Based on the pulsation of the volume pulse wave by this green light, for example, highly reliable heart rate and heart rate variability information can be obtained.

[0049] Also, oxyhemoglobin, deoxyhemoglobin (also referred to as reduced hemoglobin), and glycated hemoglobin can each have different absorption coefficients (typically, absorption spectra) for wavelengths of red light or near-infrared light. Thus, as a preferred example, for instance, the light-emitting element can include a combination of a red LED with a central wavelength of 630 nm or more and 690 nm or less and an infrared LED with a central wavelength of 810 nm or more and 990 nm or less. Based on the difference in the absorption coefficients of these two lights for oxyhemoglobin and deoxyhemoglobin, SpO2 (Peripheral Blood Oxygen Saturation), blood oxygen concentration, and heart rate can be obtained.

[0050] As another preferred example, the light-emitting element may include, for example, a combination of LEDs that generate three or more (e.g., three or four) different lights with central wavelengths of 600 nm to 990 nm. Based on the difference in the extinction coefficients of oxyhemoglobin, deoxyhemoglobin, and glycated hemoglobin, blood hemoglobin concentration, blood glycated hemoglobin concentration, etc., can be calculated.

[0051] Furthermore, the light-emitting element may include LEDs that generate light of different wavelengths than those mentioned above. Since hemodynamic information is convolved into the shape of the pulse wave, blood pressure can be estimated by analyzing the shape of the volume pulse wave (which may be the same as or different from any of the PPG signals mentioned above). Also, by analyzing the shape of the volume pulse wave, blood viscosity can be estimated, and blood glucose levels can be detected from the blood viscosity. For blood glucose measurement, spectroscopic methods (e.g., near-infrared spectroscopy, Raman spectroscopy, infrared spectroscopy, etc.) may be used. For example, the light-emitting element may include a combination of two or more (e.g., two or three) LEDs that generate near-infrared light with a central wavelength of approximately 1200 nm to 1600 nm. This allows blood glucose levels to be calculated based on the light absorption spectrum derived from glucose.

[0052] The pulse wave sensor S1 of this embodiment is, for example, a PPG sensor capable of emitting and receiving (detecting) at least green light, and capable of calculating heart rate.

[0053] The temperature sensor S2 is an optional sensor capable of detecting the temperature (body temperature) of a living organism. Typically, the temperature sensor S2 can detect information regarding the skin temperature or core body temperature of a living organism. The temperature sensor S2 is not limited to this, but for example, a thermopile-type infrared sensor can be used.

[0054] The microwave sensor S3 is an optional sensor that can be installed and, for example, detect the frequency characteristics around the resonant frequency by radiating microwaves onto a living organism. By analyzing these frequency characteristics, information such as skin moisture content, sweat volume, and blood glucose levels can be obtained.

[0055] The electrocardiogram sensor S4 is an optional sensor capable of detecting electrical activity associated with the movement of the heart. The electrocardiogram sensor S4 can be used as one of the measurement electrodes in various lead methods. Therefore, for example, the electrocardiogram sensor S4 may be configured to obtain electrocardiogram information in cooperation with other electrocardiogram sensors S4 provided on other wearable devices 101. As an example, electrocardiogram information can be obtained using the bipolar lead method with two ring-shaped wearable devices 101 worn on the fingers of both the left and right hands. Alternatively, electrocardiogram information may be obtained by a combination of a ring-shaped wearable device 101 worn on the finger of one hand, a wristwatch-shaped wearable device 101 worn on the wrist of the other hand, a user terminal 102 (smartphone, etc.), or a medical device, etc.

[0056] The gyro sensor / accelerometer module 205 is equipped with a gyro sensor and an accelerometer. The gyro sensor / accelerometer module 205 can detect the angular velocity and acceleration of the wearable device 101 and can detect changes in the position of the wearable device 101 (displacement, tilt, movement, etc.). In addition, the gyro sensor / accelerometer module 205 can acquire information regarding the posture, activity level, activity intensity, calories burned, steps taken, and behavioral discrimination of the person wearing the wearable device 101.

[0057] Furthermore, the gyro sensor / accelerometer module 205, like the sensor module 204 described above, can optionally be equipped with digital processing functions such as a DSC, DSP, FPGA, etc., for processing the sensor data acquired by the sensor in real time, as well as various analog circuits.

[0058] The charge management module 206 is an element that manages charging in the charging system provided in the wearable device 101. Figure 2 shows an example of a wireless charging system that charges the secondary battery 208 from the charger 104 without physical contact via a wireless charging receiver 209. However, the wearable device 101 may be configured to include, in place of and / or in addition to the wireless charging receiver 209, contact electrodes (not shown) for contact charging with the charger 104.

[0059] The wireless charging system includes, for example, a charge management module 206, a secondary battery 208, and a wireless charging receiver 209 provided in a wearable device 101, and an external charger 104. The charge management module 206 manages the charging process to prevent overcurrent, overvoltage, overheating, etc., of the secondary battery 208. The secondary battery 208 is not particularly limited, but examples include lithium-ion batteries, lithium polymer batteries, and all-solid-state lithium-ion batteries.

[0060] Furthermore, various charging systems can be employed as wireless charging systems, such as electromagnetic induction, magnetic field resonance, electric field coupling, and radio wave reception. For example, a wireless charging system compatible with Near Field Communication (NFC) standards or Qi standards can be preferably adopted in this embodiment. Among these, a system compatible with NFC standards is particularly preferable because it allows for miniaturization and cost reduction.

[0061] The communication control unit 207 is configured to connect to other terminals (for example, user terminal 102) via a network. The network can be wired or wireless, and each terminal can send and receive information from each other via the network. In the case of wireless communication, communication devices conforming to standards such as Bluetooth®, Wi-Fi®, or LTE® can be used. In this embodiment, the communication control unit 207 connects to the user terminal 102 via BLE (Bluetooth Low Energy) communication.

[0062] The wearable device 101 stores programs and applications such as an execution module 211, a health management module 212, a health information evaluation module 213, a pre-processing module 214, and a post-processing module 215 in its main memory 201. The processor 203 then executes these programs and applications to realize each functional element of the wearable device 101. In other words, by storing the same functional modules in the main memory 201 of the wearable device 101 as in the main memory 401 of the management server 103, the wearable device 101 can perform the same processing as the management server 103.

[0063] The functions of each of these modules are the same as described above, and their explanation will be omitted. Note that the main memory 201 does not necessarily have to store all of the above-mentioned functional modules; it may store only some of them (for example, at least the health management module 212, the health information evaluation module 213, and the preprocessing module 214). Furthermore, each functional module may be composed of some or all analog circuits. In other words, the main memory 201 of the wearable device 101 is equipped with the acquisition unit, the first pulse rate acquisition unit, the activity information analysis unit, the second pulse rate acquisition unit, and the heart rate estimation unit in this technology.

[0064] The auxiliary storage device 202 stores various types of information necessary for realizing the above-mentioned functions of the wearable device 101. For example, the auxiliary storage device 202 can store user information, health management information, heart rate estimation model information, and so on.

[0065] The auxiliary storage device 202 of this embodiment stores, for example, a first pulse rate, the feature intensity of activity information, the feature frequency corresponding to the feature intensity, and a second pulse rate. These may be common to some of the information stored in the management server 103 (for example, user information, heart rate estimation model information, etc.). The auxiliary storage device 202 of this embodiment also stores a heart rate estimation model M and its parameters (an example of heart rate estimation model information). As will be described later, these heart rate estimation models M are realized, for example, by adjusting a histogram-based gradient boosting regression tree model with pre-set parameters.

[0066] [Method for acquiring heart rate during exercise] Next, a method for acquiring heart rate during exercise according to one embodiment will be described. The exercise heart rate estimation system 100 is configured to acquire the heart rate of a user wearing the wearable device 101 during exercise by operating the wearable device 101 in "exercise heart rate acquisition mode".

[0067] In the user terminal 102, for example, when the user selects to turn "exercise heart rate acquisition mode" to "ON" (i.e., gives an operation command), the execution module 311 of the user terminal 102 sends an instruction to the wearable device 101 to operate in exercise heart rate acquisition mode. The execution module 311 of the user terminal 102 may also send information to the management server 103 indicating that it has received an instruction from the user to turn on exercise heart rate acquisition mode.

[0068] The execution module 311 may be configured to turn on the "exercise heart rate acquisition mode" when a predetermined exercise mode (for example, running mode, swimming mode, cycling mode, etc.) is selected. Alternatively, the execution module 311 may be configured to turn on the "exercise heart rate acquisition mode" at times other than when the user is sleeping (for example, when the sleep health assessment mode is turned on).

[0069] Figure 7 is a flowchart illustrating the acquisition of heart rate during exercise according to one embodiment. The execution module 211 of the wearable device 101 operates each module in exercise heart rate acquisition mode based on instructions from, for example, the user terminal 102. In exercise heart rate acquisition mode, the health management module 212, health information evaluation module 213, and preprocessing module 214 of the wearable device 101 execute the exercise heart rate acquisition flow 700.

[0070] The exercise heart rate acquisition flow 700 includes, for example, the following steps: an activity information and pulse wave sensor information acquisition step (S710), a first pulse rate acquisition step (S720), an activity information feature intensity and feature frequency acquisition step (S730), a second pulse rate acquisition step (S740), and a heart rate estimation step (S750). Note that the above steps do not necessarily have to be performed in order; for example, steps S720 and steps S730 and S740 can be performed simultaneously. Also, step S750 can be performed sequentially as steps S710 to S740 progress.

[0071] (1) Acquisition of Biological Information The health management module 212 first acquires pulse wave information and activity information (step S710). Activity information can be any information that represents the magnitude of the user's activity, and in this embodiment, acceleration information (an example of activity level information) is acquired as activity information.

[0072] The health management module 212 acquires pulse wave sensor information from the sensor module 204, for example (step S710). The health management module 212 sends an instruction to the sensor module 204 to operate the pulse wave sensor S1 and acquire a photoelectric volume pulse wave (PPG) signal over time at a predetermined sampling rate. Here, as described above, the pulse wave sensor S1 is equipped with a green LED and is capable of detecting reflected green light from the living body. The PPG signal includes an intensity signal of the reflected green light from the living body. Although the conditions for acquiring the PPG signal are not strictly limited, for example, the sampling rate can be approximately 10 to 50 Hz (for example, approximately 25 Hz).

[0073] The health management module 212 also acquires activity information from, for example, the gyro sensor / accelerometer module 205 (step S710). The health management module 212 sends an instruction to the gyro sensor / accelerometer module 205 to operate the accelerometer and acquire acceleration information (accelerometer information) over time at a predetermined sampling rate. Although the conditions for acquiring acceleration information are not strictly limited, the sampling rate can be approximately 10 to 30 Hz (for example, approximately 12.5 Hz). The acceleration information can also be, for example, three-axis acceleration information in the x, y, and z axes.

[0074] (2) Acquisition of the first pulse rate Next, the preprocessing module 214 acquires the first pulse rate based on the acquired pulse wave information (PPG signal) (S720). Of the PPG signal, the direct current (DC) component originates from the absorption of light by venous blood and tissues and hardly changes with the beating (pulsation) of the heart. On the other hand, the alternating current (AC) component fluctuates in accordance with the periodic increase and decrease in blood (arterial blood) volume due to the beating of the heart. From this, the preprocessing module 214 can calculate the first pulse rate from the periodicity of the AC component of green light.

[0075] The method for obtaining the first pulse rate from the PPG signal is not particularly limited, and for example, one or a combination thereof of the peak detection method, zero-crossing method, spectral analysis method, template matching method, and differential method can be employed. In this embodiment, the peak detection method, which is a relatively easy method to implement, is employed. The peak detection method can obtain information corresponding to the pulse rate per minute by detecting the peak of the R wave (an upward wave with a large amplitude). The information corresponding to the pulse rate may be obtained, for example, as shown in Figure 10(A), as the number of R wave peaks per unit time of less than one minute (here, 30 seconds). Here, the peak value of the R wave can be determined to be a signal value that is higher than the previous signal value (which may be the signal value at a predetermined time) and the next signal value (which may be the signal value at a predetermined time).

[0076] Furthermore, this peak interval corresponds to the heart rate interval (IBI). Therefore, the preprocessing module 214 can calculate the average of the heart rate [bpm] per unit time window based on the IBI as the first pulse rate, for example, as shown in Figure 10(B). When obtaining the first pulse rate from the PPG signal, the preprocessing module 214 may, if necessary, remove unnecessary information using a bandpass filter, low-pass filter, high-pass filter, differential filter, SSF (Slope Sum Function) filter, moving average filter, etc., or calculate the pulse rate by detecting the bottom peak of the pulse wave.

[0077] Alternatively, the preprocessing module 214 may obtain the first pulse rate by analyzing the PPG signal with respect to frequency. As shown in Figure 10(C), the preprocessing module 214 calculates the heart rate based on the highest frequency obtained when the pulse wave information is converted to the frequency domain (e.g., power spectral density). For frequency domain analysis, for example, the PPG signal can be decomposed into frequency components using Fourier transform, Fourier series, Laplace transform, Z transform, etc., and information representing the simplified frequency domain (e.g., power spectral density, frequency spectrum, spectral density, etc.) can be obtained by optionally removing phase information. For the calculation of power spectral density, refer to the explanation of acceleration information processing described later.

[0078] The preprocessing module 214 measures a first pulse rate (beats per minute: bpm) at predetermined time windows (e.g., 30 seconds). The preprocessing module 214 stores the calculated first pulse rate in the auxiliary storage device 202, associating it with information about the time window. Although not necessarily limited to this, the first pulse rate, the maximum PSD and estimated heart rate described later, etc., can be obtained in a common time window.

[0079] On the other hand, pulse rate calculated using the peak detection method has the drawback of being prone to noise such as sensor drift, baseline fluctuations due to body movement, and amplitude modulation due to respiration. In particular, in the case of a ring-shaped wearable device 101 that is attached near the end of the arm, for example, which is swung around significantly relative to the torso during exercise, the effects of noise from body movement and respiration can greatly affect measurement accuracy.

[0080] For reference, here are some examples of resting and exercise heart rates for adults: <Resting> Normal range: 60-100 bpm Average: 70-75 bpm Athlete: 40-60 bpm <Exercise> Light exercise: 90-120 bpm Moderate exercise: 120-150 bpm Vigorous exercise: 150-180 bpm

[0081] (3) Acquisition of Feature Intensity and Feature Frequency of Activity Information Next, the preprocessing module 214 acquires the feature intensity of the frequency components of the activity information and the feature frequency corresponding to the feature intensity based on the acquired activity information (acceleration information) (step S730). For example, the preprocessing module 214 converts the acceleration information, which is a time-domain signal, into a frequency domain for each predetermined time window. Various methods listed in the PPG signal section can be used as methods for converting a time-domain signal into a frequency domain. For example, a Fourier transform such as the Fast Fourier Transform can be used for frequency analysis, and for example, power spectral density analysis may be used. This makes it possible to express the strength of the frequency components of the signal in the acceleration information (power per unit frequency) as a density. For example, the preprocessing module 214 calculates the power spectral density (PSD) for the amplitude of acceleration in the x, y, and z axis directions (sqrt(x*x + y*y + z*z)). The power spectral density can be calculated based on the power spectral density function of the following equation, for example.

[0082] In the formula, P(kΔf) is the calculated power spectrum, Δf is the frequency resolution, and W is the power spectrum. ｆThis is a correction factor based on the window type, and the correction factor can be, for example, 1.5 for a Hanning window, 1 for a rectangular window, and 3.166 for a flat-top window.

[0083] The preprocessing module 214 can, for example, use the maximum value of the calculated PSD as the feature intensity and the frequency corresponding to the maximum value of the PSD as the feature frequency for each predetermined time window. This maximum value of the PSD and frequency can be considered to represent the appearance of a peak at a particular frequency when the acceleration information contains a specific rhythm or periodic motion. The preprocessing module 214 stores the calculated maximum value of the PSD (feature intensity) and the corresponding frequency (feature frequency) in the auxiliary storage device 202, in association with the time window information.

[0084] (4) Acquisition of the second pulse rate The preprocessing module 214 also acquires a second pulse rate based on the acquired pulse wave information (PPG signal) (step S740). The second pulse rate can be the pulse rate after noise reduction during exercise.

[0085] In other words, the preprocessing module 214 converts the pulse wave information, which is a time-domain signal, into the frequency domain, similar to the activity information, for each predetermined time window. The preprocessing module 214 calculates the PSD of the pulse wave information, for example. Then, for each predetermined time window, the preprocessing module 214 removes (filters) the PSD of the pulse wave information relating to the characteristic frequency calculated in step S730 and its neighboring frequencies. The neighboring frequencies can be, for example, a few percent before and after the characteristic frequency (Fm) (for example, ±3 to 7% Fm, or ±5% Fm as an example). This makes it possible to remove PSD from the pulse wave information that is thought to contain a specific rhythm or periodic movement based on the acceleration information.

[0086] Next, the preprocessing module 214 calculates a second pulse rate using the frequency Fm2 corresponding to the largest PSD among the remaining pulse wave information PSDs from which noise has been removed. The second pulse rate [bpm] can be calculated, for example, by frequency Fm2 [1 / s] × 60 [seconds]. This second pulse rate corresponds to the heart rate calculated considering artifacts caused by exercise. The preprocessing module 214 stores the calculated second pulse rate in the auxiliary storage device 202, for example, in association with information about the time window.

[0087] (5) Estimation of heart rate during exercise Next, the health information evaluation module 213 estimates the heart rate during exercise based on the first pulse rate, the second pulse rate, feature intensity, and feature frequency (S750). Hereinafter, the estimated heart rate during exercise will simply be referred to as the "estimated heart rate". According to the inventors' findings, there is a relatively high correlation between the first pulse rate, the second pulse rate, feature intensity, and feature frequency and the estimated heart rate. Therefore, the preprocessing module 214 may calculate the estimated heart rate based on a pre-calculated correlation formula. Hereinafter, according to the inventors' findings, when the first pulse rate, the second pulse rate, feature intensity, and feature frequency are provided as explanatory variables, it has become clear that the estimated heart rate as the objective variable can be estimated simply and with high accuracy by using a heart rate estimation model M (machine learning model). Therefore, in this embodiment, the case in which the estimated heart rate is obtained using the heart rate estimation model M will be explained as an example.

[0088] In other words, the health information evaluation module 213 inputs the first pulse rate, the second pulse rate, feature intensity, and feature frequency to the heart rate estimation model M, and obtains the estimated heart rate as an output. The health information evaluation module 213 stores the calculated second pulse rate in the auxiliary storage device 202, for example, in association with information about the time window.

[0089] (6) Heart Rate Estimation Model M The heart rate estimation model M is described below. The heart rate estimation model M is a machine learning model that takes the first pulse rate, second pulse rate, feature intensity, and feature frequency obtained as described above based on the user's biometric information (pulse wave information and activity information) as input and outputs the medical heart rate, which is the heart rate calculated based on medical information.

[0090] The heart rate estimation model M is implemented in a ring-shaped wearable device 101. The auxiliary storage device 202 of the ring-shaped wearable device 101 stores, for example, a histogram-based gradient boosting regression (HGBR) tree model and parameters obtained through learning described later. The heart rate estimation model M is realized by adjusting the HGBR tree model according to the parameters. Because this heart rate estimation model M is implemented in the ring-shaped wearable device 101, it is possible to predict the heart rate during exercise in real time based on input data from which biometric information has been preprocessed.

[0091] Medical information used for learning can include, for example, electrocardiogram (ECG) data, which records the weak electrical signals generated when the heart muscle contracts. This medical information may be ECG data measured by a simple electrocardiograph (home electrocardiograph) or ECG data measured by a medical electrocardiograph. From the perspective of obtaining estimated heart rate with higher accuracy, the medical heart rate can be, for example, ECG data measured by a medical electrocardiograph.

[0092] For reference, the training data for the heart rate estimation model M implemented in this embodiment will be described. The training data was obtained from the following information simultaneously recorded from the subject: (1) Electrocardiogram (ECG) signal recorded by a medical electrocardiograph (2) PPG signal and acceleration signal recorded by the ring-type wearable device 101

[0093] The subjects simultaneously wore the electrodes of a medical electrocardiograph and a ring-shaped wearable device 101, and performed the following four types of exercise (total: 990 minutes). 59,000 biometric data samples were collected. • Desk work (102 minutes) • Walking (170 minutes) • Running (300 minutes) • Other (418 minutes)

[0094] In preparing the training data, the health management module 212 of the ring-type wearable device 101 acquired the subject's biometric information (PPG signal and acceleration signal), and the health management module 212 transmitted the acquired biometric information to a smartphone (user terminal 102) connected via low-power BLE (Bluetooth Low Energy), where it was recorded. The transmitted data (raw data) was approximately 500 bytes. In this embodiment, the subject's biometric information acquired during the machine learning stage was transmitted to an external terminal as raw data. However, as described above, the preprocessing module 214 on the ring-type wearable device 101 may perform the same processing as described above, calculating the first pulse rate, second pulse rate, feature intensity, and feature frequency at predetermined time windows, and transmitting the calculation results to the smartphone (user terminal 102).

[0095] The subject's biometric information acquired by the medical electrocardiograph was transmitted to a smartphone (user terminal 102) connected via BLE and recorded on the smartphone. For the ECG signals acquired by the medical electrocardiograph, the smartphone calculated the heart rate (medical heart rate) according to the medical electrocardiograph's processing algorithm, and recorded the calculation result on the smartphone.

[0096] Each smartphone (user terminal 102) transmitted the first pulse rate, second pulse rate, feature intensity, feature frequency, and medical heart rate to a training PC (an example of a management server 103) to prepare a training dataset.

[0097] For model training, for example, the Python library (scikit-learn) was used to train a histogram-based gradient boosting regression (HGBR) tree model using the following training data. The parameters of the heart rate estimation model M obtained through training are stored in the auxiliary storage device 302 in order to implement the heart rate estimation model M on the ring-type wearable device 101. Input: First pulse rate, second pulse rate, feature intensity, feature frequency Output (label): Medical heart rate

[0098] Figure 8 is a graph showing the relationship between a reference second heart rate and a medical heart rate obtained based on a medical device. For example, the correlation between the second heart rate calculated after considering (eliminating) artifacts identified based on activity information and the medical heart rate is extremely low at 0.14, indicating that it is difficult to determine the exercise heart rate from the biometric information detected by the ring-shaped wearable device 101.

[0099] Figure 9 is a graph showing the relationship between the estimated exercise heart rate according to one embodiment and the medical heart rate obtained based on a medical device. The correlation between the estimated exercise heart rate obtained by this technology and the medical heart rate obtained based on a medical device is 0.84, confirming that the two show a sufficiently high correlation. In other words, it has been confirmed that this technology can accurately estimate the exercise heart rate with a computational load not significantly different from conventional methods.

[0100] Non-Patent Document 1 discloses, for example, the introduction of deep learning to estimate heart rate in continuous heart rate monitoring using photoplethysmography (PPG). Furthermore, Non-Patent Document 1 discloses, for example in Table 8, the evaluation performance of heart rate estimation using the IEEE test dataset, comparing the case where a convolutional neural network (CNN) is used as the machine learning algorithm with the case where three other classical methods are used. However, the mean absolute error (MAE) of these methods disclosed in Non-Patent Document 1 is small, at approximately 9 bpm for the SpaMa algorithm, but large, at approximately 12 to 25 bpm for the other algorithms. Moreover, it is stated that classical algorithms cannot be optimized for specific sessions in daily life. Furthermore, the MAE when using a convolutional neural network (CNN) is approximately 16.5 bpm, which is significantly larger than the MAE of 10.65 bpm calculated for this technology. Furthermore, most importantly, convolutional neural networks (CNNs) are too large in size to be implemented in, for example, a small and lightweight ring-shaped wearable device 101. In this respect as well, the superiority of the exercise heart rate estimation method using this technology can be confirmed.

[0101] (7) The health management module 212 of the wearable device 101 can transmit information regarding the estimated heart rate and its time window to, for example, the user terminal 102. The health management module 312 of the user terminal 102 can store the information regarding the estimated heart rate and its time window received from the wearable device 101 in, for example, the auxiliary storage device 302.

[0102] Furthermore, the health management module 312 of the user terminal 102 can, for example, display the acquired estimated heart rate on an output device 305 such as a display, for example, in real time and / or as time-series information. This allows the user to check (for example, visually confirm) highly accurate heart rate information in real time, even during exercise.

[0103] Furthermore, the post-processing module 315 of the user terminal 102 may be configured to, for example, receive a target heart rate setting from the user to serve as a reference during exercise, evaluate whether the estimated heart rate is equal to or greater than the target heart rate, and output information indicating that the estimated heart rate is equal to or greater than the target heart rate if the estimated heart rate is equal to or greater than the target heart rate.

[0104] Furthermore, the post-processing module 315 may be configured to evaluate, for example, whether the estimated heart rate exceeds a predetermined preferred heart rate that is lower than the target heart rate, and to output information indicating that the estimated heart rate is within a preferred range if the estimated heart rate is less than the target heart rate and greater than or equal to the preferred heart rate.

[0105] The post-processing module 315 may display the above information, based on the target heart rate or preferred heart rate, in the form of text, colors, images, etc., on the display of the user terminal 102, output it as sound information, or output it as haptic information (an example of vibration information) on the user terminal 102 and / or the wearable device 101. This makes it possible to present information about the user's heart rate during exercise to the user during exercise without interfering with the user's exercise.

[0106] Furthermore, to efficiently burn fat through exercise, it is generally recommended to perform aerobic exercise at a "moderately strenuous" level. In particular, in exercise therapy for cardiac rehabilitation, it is necessary to perform "moderately strenuous" exercise within a range that does not put excessive strain on the heart. In such exercises, the exercise intensity is sometimes set appropriately based on the heart rate during exercise. This technology is beneficial because it allows for more accurate estimation of heart rate during exercise, making it easier to manage exercise intensity appropriately.

[0107] Furthermore, according to this technology, the heart rate estimation model M can be implemented, for example, within a ring-shaped, ultra-compact, and lightweight wearable device 101. The estimated heart rate can then be acquired within the ring-shaped wearable device 101. As a result, the estimated heart rate can be transmitted smoothly from the wearable device 101 to the user terminal 102, for example, via BLE. This reduces power consumption in the exercise heart rate acquisition mode (for example, including estimation and real-time transmission of exercise heart rate). This is extremely advantageous for the ring-shaped wearable device 101, which is required to be smaller and lighter, in order to extend the operating time on a single full charge. As a result, even a small wearable device 101, such as a ring, with limited battery capacity, can monitor the user's heart rate during exercise at a high frequency, for example, 25 Hz, enabling more accurate health assessment during exercise.

[0108] The health management module 412 may be configured to perform exercise-related health management concerning exercise heart rate, either in place of or in addition to the user terminal 102. Furthermore, the health management module 412 may be configured to manage information related to the user's health in addition to exercise-related health management concerning exercise heart rate. Specifically, the health management module 412, for example, works in conjunction with the health management module 312 of the user terminal 102 to acquire biometric information obtained from the user terminal 102 by the wearable device 101. The health management module 412 also analyzes the acquired biometric information using volume plethysmography and calculates at least one piece of health management information, including heart rate, heart rate variability, blood oxygen saturation, blood pressure, and blood glucose level. The health management module 412 may be configured to calculate the user's posture and displacement, activity level, calories burned, steps taken, behavioral discrimination, body temperature (skin temperature and core temperature), skin moisture content and sweating, blood glucose level, electrocardiogram, heart rate, heart rate variability, respiratory rate, and other health management information based on acquired biometric information. The health management information may include management indicators for disease prevention defined as needed, such as a vascular health index, activity level, stress level, depression level, and lifestyle-related disease risk. The health management module 412 may also perform analysis and evaluation of arbitrary health items using data obtained from various biometric signals such as displacement, body temperature, heart rate, heart rate variability, and respiratory rate. The health management module 412 outputs (stores) the calculated health management information to, for example, the health management information 420 in the auxiliary storage device 402 for management.

[0109] Furthermore, the health management module 412, for example, works in conjunction with the health management module 312 of the user terminal 102 to output (display) the calculated health management information on the display of the user terminal 102 (an example of an output device 305).

[0110] The health management module 412 may also be configured to notify the health management module if it finds any of the specified characteristics in the calculated health management information. The notification method is not particularly limited and may be, for example, displayed on the display of the user terminal 102, or sent via email or message to a designated recipient.

[0111] The embodiments relating to this technology have been specifically described above, but these are merely examples and do not limit the scope of the claims. The technology described in the claims includes various modifications and changes to the specific examples illustrated above. For example, the embodiments described above are explained in detail to make the present invention easier to understand, and are not necessarily limited to those having all the configurations described. Furthermore, it is possible to replace a part of the configuration of one embodiment with another, and it is also possible to add other configurations to the configuration of one embodiment. In addition, it is possible to add, delete, or replace other configurations for a part of the configuration of each embodiment.

[0112] This technology provides a program for executing each step of the exercise heart rate measurement method in the wearable device 101, the user terminal 102, and the management server 103. This program is stored, for example, in the main memory of the wearable device 101, the user terminal 102, and the management server 103. By executing this program, the processors of the wearable device 101, the user terminal 102, and the management server 103 can utilize the services of the exercise heart rate estimation system 100.

[0113] In this embodiment, the wearable device 101 performs the entire process from acquiring biometric information to estimating the heart rate during exercise. In other words, the user terminal 102 and the management server 103 are not essential components. However, the user terminal 102 and the management server 103 may perform some of the processing that the wearable device 101 performs. Similarly, the main components of the processing performed in the above embodiment are not limited to those shown in the embodiment, and may be performed by the wearable device 101, the user terminal 102, or the management server 103.

[0114] Furthermore, each of the above configurations, functions, processing units, processing means, etc., may be implemented in hardware, either partially or entirely, by designing them as integrated circuits, for example. Alternatively, each of the above configurations, functions, etc., may be implemented in software by having the processor interpret and execute programs that implement each function. Information such as programs, tables, and files that implement each function can be stored in memory, a recording device such as a hard disk or SSD (Solid State Drive), or a recording medium such as an IC card, SD card, or DVD (for example, a non-temporary storage medium).

[0115] Furthermore, in the diagrams showing the hardware configuration, only control lines and information lines deemed necessary for explanation are shown, and not all control lines and information lines are necessarily shown in the actual product. In reality, it can be assumed that almost all components are interconnected. Note that the above-described embodiments disclose at least the configuration described in the claims.

[0116] 100...Exercise heart rate estimation system, 101...Wearable device, 102...User terminal, 212...Health management module, 213...Health information evaluation module, 214...Pre-processing module, 311...Execution module, 312...Health management module, 313...Health information evaluation module, 314...Pre-processing module, 315...Post-processing module

Claims

1. An exercise heart rate estimation system comprising: an acquisition unit that acquires pulse wave information and activity information of a living organism in a temporally associated manner; a first pulse rate acquisition unit that acquires a first pulse rate based on the pulse wave information; an activity information analysis unit that acquires the characteristic intensity of the frequency components of the activity information and the characteristic frequency corresponding to said characteristic intensity; a second pulse rate acquisition unit that acquires a second pulse rate based on the remaining components of the frequency components of the pulse wave information, excluding the characteristic frequency and the frequency components in its vicinity; and a heart rate estimation unit that estimates the exercise heart rate based on the first pulse rate, the second pulse rate, the characteristic intensity of the frequency components of the activity information, and the characteristic frequency.

2. The exercise heart rate estimation system according to claim 1, wherein the acquisition unit acquires acceleration sensor information from an acceleration sensor as the activity information.

3. The exercise heart rate estimation system according to claim 1 or 2, wherein the activity information analysis unit obtains the maximum value of the power spectral density (PSD) as the characteristic intensity of the frequency components of the activity information, and obtains the frequency corresponding to the maximum value of the PSD as the characteristic frequency.

4. The exercise heart rate estimation system according to any one of claims 1 to 3, wherein the activity information analysis unit acquires the characteristic intensity and characteristic frequency of the frequency component of the amplitude of the activity information.

5. The exercise heart rate estimation system according to any one of claims 1 to 4, wherein the first pulse rate acquisition unit acquires the first pulse rate by one or more of the following methods: peak detection method, zero-crossing method, spectral analysis method, template matching method, and differential method.

6. The exercise heart rate estimation system according to any one of claims 1 to 5, wherein the second pulse rate acquisition unit acquires the power spectral density (PSD) of the pulse wave information and acquires the second pulse rate based on the maximum PSD among the PSDs relating to the remaining frequency components after excluding the characteristic frequency and its vicinity.

7. The exercise heart rate estimation system according to any one of claims 1 to 6, wherein the heart rate estimation unit takes as input the characteristic intensity of the frequency components of the learning activity information of the living organism, the characteristic frequency corresponding to said characteristic intensity, a third pulse rate obtained based on the learning pulse wave information of the living organism, and a fourth pulse rate obtained based on the remaining components of the frequency components of the learning pulse wave information excluding the characteristic frequency and its vicinity, and estimates the exercise heart rate using a heart rate estimation model that has been trained to output a medical heart rate based on the medical information of the living organism obtained in correspondence with the learning activity information and the learning pulse wave information.

8. The exercise heart rate estimation system according to claim 7, wherein the heart rate estimation model is trained using a histogram gradient boosting decision tree method.

9. The exercise heart rate estimation system according to claim 7, wherein the heart rate estimation model is trained to output a heart rate based on electrocardiogram information as the medical heart rate.

10. An exercise heart rate estimation system according to any one of claims 1 to 9, comprising a wearable device that can be attached to a living body, wherein the acquisition unit, the first pulse rate acquisition unit, the activity information analysis unit, the second pulse rate acquisition unit, and the heart rate estimation unit are all provided in a single wearable device.

11. The exercise heart rate estimation system according to claim 10, wherein the wearable device is a wearable ring that can be attached to any one of the fingers of a living person.

12. A method for estimating exercise heart rate, comprising: acquiring pulse wave information and activity information of a living organism in a temporal correspondence; acquiring a first pulse rate based on the pulse wave information; acquiring the characteristic intensity of the frequency components of the activity information and the characteristic frequency corresponding to said characteristic intensity; acquiring a second pulse rate based on the remaining components of the frequency components of the pulse wave information, excluding the characteristic frequency and the frequency components in its vicinity; and estimating the exercise heart rate based on the first pulse rate, the second pulse rate, the characteristic intensity of the frequency components of the activity information, and the characteristic frequency.

13. A program for causing a computer to perform each step of the method for controlling the wearable device described in claim 10.

Images