Electrocardiogram-based exercise recommendation method and computer program recorded on a recording medium for executing the method

Abstract

The present invention proposes a method for recommending exercises suited to a user's real-time health condition based on an electrocardiogram (ECG). The method includes the steps of acquiring an electrocardiogram (ECG) signal measured from a user, analyzing the ECG signal using pre-trained artificial intelligence (AI) to identify the user's cardiac pump status, and determining a type of exercise to be recommended to the user in accordance with the identified cardiac pump status. Thus, a user can receive recommendations for exercises optimized for their current health condition simply by measuring their ECG.

Description

[Technical Field]

[0001] The present invention relates to smart healthcare, and more particularly to a method for recommending exercises suitable for a user's real-time health condition based on an electrocardiogram (ECG), and a computer program recorded on a recording medium for executing the method. [Background technology]

[0002] An electrocardiogram (ECG) is a graphical record of the visual changes in electrical activity generated by the heart muscles. Briefly explaining the cardiac conduction system, the sinoatrial node (SA node) periodically generates electrical signals to induce cardiac contractions. The atrioventricular node (AV node) transmits the electrical signals generated by the SA node through the atria, after slightly delaying them. The electrical signals transmitted from the AV node are then distributed throughout the ventricles via the His bundle and Purkinje fibers.

[0003] The electrocardiogram (ECG) waveforms that can be seen during the cardiac impulse conduction process are as follows: The P wave is seen when the electrical signal generated by the sinoatrial node polarizes the atria, causing the valvular myocardium to contract, then depolarizes and relaxes again. The PQ wave is seen when the atrioventricular node delays the electrical signal to prevent the ventricles from immediately responding when the atria contract. The QRS complex is seen when the delayed electrical signal is transmitted by the AV node (Q), causing the ventricles to immediately polarize (R) and then immediately depolarize (S). The ST wave is seen during the resting period when the ventricles contract and blood moves around the body, preventing the electrical signal from stimulating the heart. The T wave is seen when the ventricles weakly polarize and then depolarize again, causing the ventricles and valvular myocardium to simultaneously relax. In some cases, a U wave due to repolarization of the intraventricular septum can also be seen.

[0004] The order of appearance of waveforms (P waves, PQ waves, QRS waves, ST waves, T waves, and U waves) that can be confirmed by such an electrocardiogram (ECG), as well as their respective shapes, sizes, and intervals, can vary in various forms depending on the health state. In other words, by analyzing the order of appearance, shapes, sizes, and intervals of waveforms included in an electrocardiogram (ECG), it is possible to infer the health state and even the various causes that cause such health state.

[0005] Meanwhile, an artificial neural network (ANN) is a detailed methodology of machine learning, and refers to an algorithm that has the form of a network in which neurons, the basic operational units corresponding to human nerve cells, are linked together. Representative artificial neural networks (ANN) include multilayer perceptrons (MLPs), recurrent neural networks (RNNs), and convolutional neural networks (CNNs).

[0006] Specifically, a multilayer perceptron (MLP) is an artificial neural network (ANN) with multiple hidden layers between the input and output layers. A recurrent neural network (RNN) is an artificial neural network (ANN) with a recurrent connection structure in which the output of neurons is fed back to the input layer. A convolutional neural network (CNN) is an artificial neural network (ANN) that mimics the structure of the human visual nerve, extracts feature maps from multiple convolutional layers, and then extracts only important parts from the feature maps by reducing the dimension of the matrix through subsampling.

[0007] On the other hand, exercise refers to various physical activities performed for the purpose of maintaining or improving physical strength, performance, or health. Such exercise can be classified into aerobic exercise, anaerobic exercise, and flexibility exercise depending on the effect on the human body, and can also be classified into muscle strengthening exercise, bone strengthening exercise, balance exercise, and multi-physical activity depending on the body part to be trained.

[0008] Exercise capacity refers to the degree of ability to exercise. Generally, exercise capacity is evaluated based on the age, gender, and presence or absence of late-onset diseases of a subject. For this reason, conventional services that recommend or suggest exercises only recommend or suggest exercises in order from exercises that have a low impact on the user's body to exercises that have a high impact on the body, or simply recommend or suggest exercises that correspond to age, gender, late-onset diseases, etc.

[0009] However, cardiac performance, which is the factor that has the greatest impact on exercise capacity, can change from moment to moment depending on previously performed activities, metabolism, stress, etc. For example, if a person drank too much the previous day, their exercise capacity may be significantly reduced due to a decrease in myocardial contractility and an increase in blood clots caused by dehydration. In addition, it is practically difficult for non-medical professionals to undergo echocardiography to determine whether the exercise they are doing is appropriate for their health condition before starting exercise in their daily lives.

[0010] Therefore, there is a need for a means that can recommend exercises suitable for a user's real-time health condition without requiring the user to perform burdensome actions. Summary of the Invention [Problem to be solved by the invention]

[0011] An object of the present invention is to propose a method that can recommend exercises suitable for a user's real-time health condition based on an electrocardiogram (ECG).

[0012] Another object of the present invention is to propose a computer program recorded on a recording medium for executing a method capable of recommending exercises suitable for a user's real-time health condition based on an electrocardiogram (ECG).

[0013] The technical problems of the present invention are not limited to those mentioned above, and other technical problems not mentioned will be clearly understood by those skilled in the art from the following description. [Means for solving the problem]

[0014] To achieve the above technical objectives, the present invention proposes a method for recommending exercises suitable for a user's real-time health condition based on an electrocardiogram (ECG). The method includes the steps of acquiring an electrocardiogram (ECG) signal measured from a user, analyzing the ECG signal using pre-trained artificial intelligence (AI) to identify the user's cardiac pump status, and determining a type of exercise to be recommended to the user in accordance with the identified cardiac pump status.

[0015] The method may further include, after determining the type of exercise, transmitting information including the determined type of exercise to a user equipment (UE) pre-configured corresponding to the user.

[0016] More specifically, the identifying step may include a step of dividing the electrocardiogram (ECG) signal into a plurality of partial signals according to chronological order, a step of inputting the divided partial signals into the artificial intelligence (AI) respectively and obtaining a plurality of probability values from the artificial intelligence (AI), and a step of identifying the cardiac pump status of the user based on the obtained plurality of probability values.

[0017] In this case, the AI may include a first artificial neural network (ANN) related to preload, which corresponds to the end diastolic volume of the heart, a second artificial neural network (ANN) related to afterload, which corresponds to the resistance the heart experiences while contracting, and a third artificial neural network (ANN) related to myocardial contractility.

[0018] According to one embodiment, the first artificial neural network may be trained based on the displacements of P waves, PQ waves, and QRS waves included in each of the divided partial signals, the second artificial neural network may be trained based on the displacements of QRS waves, ST waves, and T waves included in each of the divided partial signals, and the third artificial neural network may be trained based on the displacements of all waveforms included in each of the divided partial signals.

[0019] The step of identifying the heart pump state may include identifying a heart rate based on the number of waveforms included in the electrocardiogram (ECG) signal per unit time, determining the preload state value based on a plurality of probability values obtained from the first artificial neural network (ANN), determining the afterload state value based on a plurality of probability values obtained from the second artificial neural network (ANN), and determining the myocardial contractility state value based on a plurality of probability values obtained from the third artificial neural network (ANN), and identifying the heart pump state based on the heart rate, the preload state value, the afterload state value, and the myocardial contractility state value.

[0020] Meanwhile, the step of identifying the cardiac pump status may involve analyzing the electrocardiogram (ECG) signal using the artificial intelligence (AI) to identify the user's maximal volume of oxygen uptake.

[0021] The step of determining the type of exercise may include determining an exercise type that matches the user's cardiac pumping state from a pre-defined exercise dictionary, and determining an exercise intensity that matches the determined exercise type and the maximum oxygen uptake.

[0022] The step of determining the type of exercise may include calculating the user's basal metabolism based on the user's gender, height, and weight, which are input in advance corresponding to the user, calculating the user's activity metabolism by applying the determined type of exercise and exercise intensity to the calculated basal metabolism, and determining the duration of the exercise based on the calculated activity metabolism.

[0023] In order to achieve the above technical objectives, the present invention proposes a computer program recorded on a recording medium for executing a method for recommending exercises suitable for a user's real-time health condition based on an electrocardiogram (ECG). The computer program can be coupled to a computing device including a memory, a transceiver, an input / output device, and a processor for processing instructions resident in the memory. The computer program is recorded on a recording medium and causes the processor to execute the following steps: acquiring an electrocardiogram (ECG) signal measured from a user via the transceiver or the input / output device; analyzing the ECG signal using pre-trained artificial intelligence (AI) to identify the user's cardiac pump status; and determining a type of exercise to be recommended to the user corresponding to the identified cardiac pump status.

[0024] Further details of the embodiments are included in the detailed description and accompanying drawings. [Effects of the Invention]

[0025] According to an embodiment of the present invention, a user can simply measure their electrocardiogram (ECG) to receive recommendations for exercises optimized for the user's current health condition.

[0026] The effects of the present invention are not limited to those mentioned above, and other effects not mentioned will be apparent to those skilled in the art to which the present invention pertains from the claims. [Brief explanation of the drawings]

[0027] [Figure 1] 1 is an exemplary diagram illustrating an exercise recommendation unit according to an embodiment of the present invention; [Figure 2] 1 is an exemplary diagram illustrating an exercise recommendation system according to an embodiment of the present invention; [Figure 3] 2 is a logical block diagram of an exercise recommendation server according to an embodiment of the present invention; [Figure 4] 1 is an exemplary diagram illustrating an artificial neural network (ANN) according to an embodiment of the present invention. FIG. [Figure 5] 1 is an exemplary diagram illustrating a process for segmenting an electrocardiogram (ECG) signal according to some embodiments of the present invention. [Figure 6] 1 is an exemplary diagram illustrating a process for segmenting an electrocardiogram (ECG) signal according to some embodiments of the present invention. [Figure 7] 10A and 10B are exemplary diagrams illustrating a process of determining a state value according to some embodiments of the present invention. [Figure 8] 1 is an exemplary diagram illustrating a process of identifying a heart pump state according to an embodiment of the present invention; [Figure 9] 2 is a hardware configuration diagram of an exercise recommendation server according to an embodiment of the present invention; [Figure 10] 1 is a flowchart illustrating an exercise recommendation method according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0028] It should be noted that the technical terms used herein are merely used to describe specific embodiments and are not intended to limit the present invention. Furthermore, unless otherwise defined herein, technical terms used herein should be interpreted in a way that is commonly understood by a person of ordinary skill in the art to which the present invention pertains, and should not be interpreted in an overly comprehensive or overly narrow sense. Furthermore, if a technical term used herein is incorrect and cannot accurately express the concept of the present invention, it should be understood by substituting a technical term that can be correctly understood by a person skilled in the art. Furthermore, general terms used herein should be interpreted according to their predefined meanings or the context, and should not be interpreted in an overly narrow sense.

[0029] Furthermore, as used herein, singular expressions include plural expressions unless the context clearly dictates otherwise. In this application, terms such as "comprise" or "have" should not be interpreted as necessarily including all of the components or steps described in the specification, but should be interpreted as not including some of the components or steps, or as possibly including additional components or steps.

[0030] Furthermore, terms including ordinal numbers such as "first" and "second" used in this specification may be used to describe various components, but the components should not be limited to these terms. These terms are used only to distinguish one component from another. For example, a first component may be referred to as a second component, and similarly, a second component may be referred to as a first component, without departing from the scope of the present invention.

[0031] When a component is said to be "coupled" or "connected" to another component, it may be directly coupled or connected to the other component, but there may be other components in between. On the other hand, when a component is said to be "directly coupled" or "directly connected" to another component, it should be understood that there are no other components in between.

[0032] Hereinafter, a preferred embodiment of the present invention will be described in detail with reference to the accompanying drawings. Identical or similar elements will be given the same reference numerals regardless of the drawing numbers, and redundant description thereof will be omitted. Furthermore, in the description of the present invention, if it is determined that a detailed description of related known technology may obscure the gist of the present invention, such a detailed description will be omitted. It should be noted that the accompanying drawings are merely provided to facilitate understanding of the concept of the present invention, and the concept of the present invention should not be construed as being limited by the accompanying drawings. The concept of the present invention should be construed as extending to all modifications, equivalents, or alternatives other than those shown in the accompanying drawings.

[0033] As described above, conventional services that recommend or suggest exercises only recommend or suggest exercises in order from least to most impactful on the user's body, or simply recommend or suggest exercises that correspond to age, gender, late-onset diseases, etc. However, cardiac capacity, which is the factor that has the greatest impact on athletic performance, can change from moment to moment depending on previously performed activities, metabolism, stress, etc.

[0034] In order to overcome such limitations, the present invention proposes a means for recommending exercises suitable for a user's real-time health condition.

[0035] FIG. 1 is an exemplary diagram illustrating an exercise recommendation means according to an embodiment of the present invention.

[0036] As shown in FIG. 1, the exercise recommendation means according to one embodiment of the present invention measures a user's electrocardiogram (ECG) 10 using an electrocardiogram reader (ECG reader) 100, analyzes the measured electrocardiogram (ECG) 10 using artificial intelligence (AI), identifies the cardiac pump status, etc. according to the user's real-time health condition, and instantly recommends to the user an exercise 20 corresponding to the identified cardiac pump status, etc.

[0037] Therefore, according to one embodiment of the present invention, a user can receive recommendations for exercises optimized for the user's current health condition simply by measuring an electrocardiogram (ECG) 10 .

[0038] Hereinafter, an apparatus and method for realizing the above-mentioned features will be described in detail.

[0039] FIG. 2 is an exemplary diagram illustrating an exercise recommendation system according to an embodiment of the present invention.

[0040] As shown in FIG. 2, an exercise recommendation system according to an embodiment of the present invention may include one or more electrocardiographs 100a, 100b, ..., 100n; 100, one or more user devices 200a, 200b, ..., 200n; 200, and an exercise recommendation server 300.

[0041] The components of the exercise recommendation system according to one embodiment of the present invention merely indicate functionally separated components, and therefore, two or more components may be integrated and implemented in an actual physical environment, or a single component may be implemented separately from each other in an actual physical environment.

[0042] To explain each component, the electrocardiograph 100 is a device capable of measuring and recording a user's electrocardiogram (ECG) 10 .

[0043] Specifically, the electrocardiograph 100 induces electric potential changes due to electrical activity in the myocardium through electrodes in contact with the user's body, amplifies the induced electric potential changes, and records them as a waveform.

[0044] The electrocardiogram 100 according to an embodiment of the present invention may generate an electrocardiogram (ECG) using any one of standard limb leads, unipolar limb leads, and precordial leads, but the lead type of the electrocardiogram 10 of the electrocardiogram 100 is not limited thereto. The electrocardiogram 100 may include electrodes for 1 lead, 6 leads, or 12 leads, but the number of leads of the electrocardiogram 100 is not limited thereto. The electrocardiogram 100 may be of any one of a watch type 100a, a portable type 100b, and a Holter type 100n, but the type of the electrocardiogram 100 is not limited thereto.

[0045] The electrocardiogram (ECG) signal (i.e., the recorded waveform signal) measured by the electrocardiograph 100 can be directly transmitted to the exercise recommendation server 300 by the network function of the electrocardiograph 100 itself, or can be input to the exercise recommendation server 300 via a third means. For example, the third means can be any one of removable media, a user device 200, and a data input action by a user, but is not limited thereto.

[0046] Next, the user device 200 can output information about the exercise determined by the exercise recommendation server 300 .

[0047] Specifically, the user device 200 may receive information about the exercise 20 from the exercise recommendation server 300. Then, the user device 200 may output the received information about the exercise 20. The information about the exercise 20 that the user device 200 receives from the exercise recommendation server 300 may include, but is not limited to, one or more of the type of exercise, the intensity of the exercise, and the duration of the exercise.

[0048] The user device 200 according to one embodiment of the present invention is not limited to a user equipment (UE) defined by 3GPP (3rd Generation Partnership Project) or a mobile station (MS) defined by IEEE (Institute of Electrical and Electronics Engineers), but may be any device that can transmit and receive data to and from the exercise recommendation server 300 and perform calculations based on the transmitted and received data.

[0049] For example, the user device 200 may be any one of, but is not limited to, a fixed computing device such as a desktop PC 200c, a workstation, or a server, or a mobile computing device such as a smartphone 200a, a laptop PC 200b, a tablet PC, a phablet, a portable multimedia player (PMP), a personal digital assistant (PDA), or an e-book reader.

[0050] As a next configuration, the exercise recommendation server 300 can identify the heart pump status, etc. according to the user's real-time health condition based on the electrocardiogram (ECG) 10 measured by the electrocardiograph 100, and transmit information about the exercise 20 corresponding to the identified heart pump status, etc. to the user device 200.

[0051] Specific components and operations of the exercise recommendation server 300 will be described later with reference to FIGS.

[0052] The one or more electrocardiographs 100, one or more user devices 200, and exercise recommendation server 300 constituting the exercise recommendation system described above may transmit and receive data via a network that directly connects the devices to each other, which may be a combination of one or more of a secure line, a shared wired communication network, or a mobile communication network.

[0053] For example, the shared wired communication network may include, but is not limited to, Ethernet, x Digital Subscriber Line (xDSL), Hybrid Fiber Coax (HFC), and Fiber To The Home (FTTH). Also, the mobile communication network may include, but is not limited to, Code Division Multiple Access (CDMA), Wideband CDMA (WCDMA), High Speed Packet Access (HSPA), Long Term Evolution (LTE), and 5th generation mobile telecommunications.

[0054] The components of the exercise recommendation server 300 having the above-mentioned features will now be described in detail.

[0055] Figure 3 is a logical block diagram of an exercise recommendation server according to an embodiment of the present invention. Figure 4 is an exemplary diagram illustrating an artificial neural network (ANN) according to an embodiment of the present invention. Figures 5 and 6 are exemplary diagrams illustrating a process of segmenting an electrocardiogram (ECG) signal according to some embodiments of the present invention. Figure 7 is an exemplary diagram illustrating a process of determining a status value according to some embodiments of the present invention. And Figure 8 is an exemplary diagram illustrating a process of identifying a heart pump status according to an embodiment of the present invention.

[0056] As shown in FIG. 3, the exercise recommendation server 300 according to one embodiment of the present invention may include a communication unit 305, an input / output unit 310, an artificial intelligence learning unit 315, an electrocardiogram release unit 320, a health condition analysis unit 325, and an exercise determination unit 330.

[0057] The components of the exercise recommendation server 300 according to one embodiment of the present invention merely represent functionally separated components, and therefore, two or more components may be implemented in an integrated manner in an actual physical environment, or one component may be implemented in a separated manner in an actual physical environment.

[0058] To explain each component, the communication unit 305 can send and receive data to and from the electrocardiograph 100 and the user device 200.

[0059] Specifically, the communication unit 305 may receive an electrocardiogram (ECG) 10 signal from the electrocardiograph 100. The communication unit 305 may also transmit information about the exercise 20 to the user device 200. In this case, the information about the exercise 20 may include, but is not limited to, one or more of the type of exercise, the intensity of the exercise, and the duration of the exercise.

[0060] The next component, the input / output unit 310, can receive commands from an administrator through a user interface (UI) or output calculation results.

[0061] Specifically, the input / output unit 310 may receive an input of a window size, which is a criterion for removing signals from the electrocardiogram (ECG) 10. The input / output unit 310 may receive an input of a validated range, which is a criterion for removing outliers from probability values acquired from an artificial intelligence (AI). The input / output unit 310 may receive an input of an exercise dictionary in which cardiac pump status, oxygen intake, and activity metabolic rate are matched with exercise type, exercise intensity, and exercise duration.

[0062] Meanwhile, the input / output unit 310 can receive an activation function to be applied to the GMF layer of the artificial neural network (ANN), an activation function to be applied to the MLP layer, a weight, and a bias.

[0063] The input / output unit 310 can also output a signal of the user's electrocardiogram (ECG) 10. The input / output unit 310 can also output information about exercises 20 that should be recommended to the user.

[0064] As a next configuration, the artificial intelligence learning unit 315 can train an artificial intelligence (AI) for identifying a cardiac pump state corresponding to the health condition of the user based on the signal of the electrocardiogram (ECG) 10. The artificial intelligence learning unit 315 can also train an artificial intelligence (AI) for identifying a maximal volume of oxygen uptake (VO2 Max) corresponding to the health condition of the user based on the signal of the electrocardiogram (ECG) 10.

[0065] Here, the cardiac pump status is a status value indicating the ability of the heart to circulate blood. Such cardiac pump status includes heart rate, preload, afterload, and myocardial contractility. And the maximum oxygen uptake is the maximum amount of oxygen that the body can consume per unit time.

[0066] Specifically, the artificial intelligence (AI) to be learned by the artificial intelligence learning unit 315 can be broadly divided into four types of artificial neural networks (ANN).

[0067] The first artificial neural network (ANN) is related to preload, which corresponds to the end diastolic volume of the heart. The second artificial neural network (ANN) is related to afterload, which corresponds to the resistance the heart experiences while contracting. The third artificial neural network (ANN) is related to myocardial contractility. Finally, the fourth artificial neural network (ANN) is related to maximum oxygen uptake.

[0068] According to one embodiment, the artificial intelligence learning unit 315 can train a first artificial neural network (ANN) based on displacements of P waves, PQ waves, and RQS waves included in each of a plurality of partial signals divided from an electrocardiogram (ECG) signal. The artificial intelligence learning unit 315 can train a second artificial neural network (ANN) based on displacements of QRS waves, ST waves, and T waves included in each of a plurality of partial signals divided from an electrocardiogram (ECG) signal. The artificial intelligence learning unit 315 can train a third artificial neural network (ANN) and a fourth artificial neural network (ANN) based on displacements of all waveforms included in each of a plurality of partial signals divided from an electrocardiogram (ECG) signal.

[0069] Each of the first artificial neural network (ANN), second artificial neural network (ANN), third artificial neural network (ANN), and fourth artificial neural network (ANN) to be learned by the artificial intelligence learning unit 315 may include an input layer, a hidden layer, and an output layer.

[0070] In particular, as shown in FIG. 4, the hidden layers of the first to fourth artificial neural networks (ANN) according to one embodiment of the present invention can be configured in a form that combines Generalized Matrix Factorization (GMF) and Multi-Layer Perceptron (MLP).

[0071] In other words, the hidden layer of the first artificial neural network (ANN) may include a GMF for learning a linear relationship between the features of the electrocardiogram (ECG) signal (i.e., the displacement features for the P, PQ, and RQS waves) and the state of preload, and an MLP for learning a nonlinear relationship between the features of the electrocardiogram (ECG) signal and the state of preload. The hidden layer of the second artificial neural network (ANN) may include a GMF for learning a linear relationship between the features of the electrocardiogram (ECG) signal (i.e., the displacement features for the QRS, ST, and T waves) and the state of afterload, and an MLP for learning a nonlinear relationship between the features of the electrocardiogram (ECG) signal and the state of afterload. The hidden layer of the third artificial neural network (ANN) may include a GMF for learning a linear relationship between the features of the electrocardiogram (ECG) signal (i.e., the displacement features for all waveforms) and the state of myocardial contractility, and an MLP for learning a nonlinear relationship between the features of the electrocardiogram (ECG) signal and the state of myocardial contractility. The hidden layer of the fourth artificial neural network (ANN) can include a GMF for learning a linearity relation between the electrocardiogram (ECG) signal features (i.e., the displacement features for all waveforms) and the maximum oxygen uptake, and an MLP for learning a non-linearity relation between the electrocardiogram (ECG) signal features and the maximum oxygen uptake.

[0072] More specifically, the AI learning unit 315 may extract one or more electrocardiogram (ECG) signal features according to a predetermined feature extraction standard for P wave, PQ wave, and QRS wave displacements. The AI learning unit 315 may one-hot encode the extracted ECG signal features and the preload state to represent them as vectors, and embed the vector for the ECG signal features and the vector for the preload state to generate dense vectors with reduced dimensions. The AI learning unit 315 may input the Hadamard product of the dense vector for the ECG signal features and the dense vector for the preload state to the GMF layer of the hidden layer, and input the concatenation of the vector for the ECG signal features and the vector for the preload state to the MLP layer of the hidden layer, thereby training a first artificial neural network (ANN).

[0073] The artificial intelligence learning unit 315 may extract one or more electrocardiogram (ECG) signal features according to a predetermined feature extraction standard for QRS wave, ST wave, and T wave displacements. The artificial intelligence learning unit 315 may one-hot encode the extracted ECG signal features and the preload state to represent them as vectors, and may embed the vector for the ECG signal features and the vector for the preload state to generate dense vectors with reduced dimensions. The artificial intelligence learning unit 315 may input the Hadamard product of the dense vector for the ECG signal features and the dense vector for the preload state to the GMF layer of the hidden layer, and input a concatenated value of the vector for the ECG signal features and the vector for the preload state to the MLP layer of the hidden layer to train a second artificial neural network (ANN).

[0074] The artificial intelligence learning unit 315 may extract one or more electrocardiogram (ECG) signal features according to a predetermined feature extraction standard for all waveform variations of the ECG signal. The artificial intelligence learning unit 315 may one-hot encode the extracted ECG signal features and the preload state to represent them as vectors, and may embed the vector for the ECG signal features and the vector for the preload state to generate dense vectors with reduced dimensions. The artificial intelligence learning unit 315 may input the Hadamard product of the dense vector for the ECG signal features and the dense vector for the preload state to the GMF layer of the hidden layer, and input a concatenated value of the vector for the ECG signal features and the vector for the preload state to the MLP layer of the hidden layer, thereby training a third artificial neural network (ANN) and a fourth artificial neural network (ANN).

[0075] For this purpose, in each artificial neural network (ANN) according to an embodiment of the present invention, the GMF layer can be realized as follows:

[0076]

number

[0077] In this case, p u is a dense vector of electrocardiogram (ECG) signal features, q i is a dense vector of cardiac pump state (maximal oxygen uptake in the case of the fourth artificial neural network), p u○ q i is the Hadamard product between the dense vector of electrocardiogram (ECG) signal features and the dense vector of heart pump state (maximal oxygen uptake in the case of the fourth artificial neural network), a is the activation function of the artificial neural network (ANN), h is the edge weight of the hidden layer, and h T can be a vector product of edge weights.

[0078] A dense vector is a vector whose dimension is reduced by embedding a vector represented by one-hot encoding. In other words, a dense vector is a vector whose dimension is reduced by removing zero values from a sparse vector represented by one-hot encoding.

[0079] Also, in an artificial neural network (ANN), the MLP layer can be realized by the following Equations 2 and 3.

[0080]

number

[0081]

number

[0082] In this case, p u is a dense vector of electrocardiogram (ECG) signal features, q i is a dense vector of the heart pump state (maximum oxygen uptake in the case of the fourth artificial neural network), z is the concatenation of the dense vector of the electrocardiogram (ECG) signal features and the dense vector of the heart pump state (maximum oxygen uptake in the case of the fourth artificial neural network), a is the activation function of the artificial neural network (ANN), W is the weight of each layer that makes up the MLP, and W T is the vector product of the weights, b is the bias of each layer that makes up the MLP, L is the number of layers that make up the hidden layer, and δ can be the sum of a sequence.

[0083] Each artificial neural network (ANN) can also be configured in the form of a recurrent neural network having a recurrent connection structure in which the probability value output to the output layer is added to a vector for the heart pump state and then input back to the input layer.

[0084] In this way, the artificial neural network (ANN) of the artificial intelligence learning unit 315 has a form that combines GMF and MLP, and can simultaneously machine-learn not only the linear relationship that exists between the characteristics of the electrocardiogram (ECG) signal and the cardiac pump state (maximum oxygen uptake in the case of the fourth artificial neural network), but also the nonlinear relationship.

[0085] Referring to FIG. 3, the next configuration will be described. The electrocardiogram canceller 320 can cancel the signal of the electrocardiogram (ECG) 10 measured by the electrocardiograph 100.

[0086] Specifically, the ECG release unit 320 can acquire an ECG 10 signal measured on the user. That is, the ECG release unit 320 can directly receive the ECG 10 signal from the ECG monitor 100 via the communication unit 305 or can directly receive the ECG 10 signal via the input / output unit 310.

[0087] The electrocardiogram canceller 320 can divide the electrocardiogram (ECG) 10 signal into a plurality of segmentation signals according to the time series measured from the user.

[0088] 5, the electrocardiogram canceller 320 can divide the electrocardiogram (ECG) 10 signal into a plurality of partial signals S1, S2, S3, ... by discretely cutting the electrocardiogram (ECG) 10 signal at regular amplitudes without overlapping regions. In this case, the amplitude for discretely cutting the electrocardiogram (ECG) 10 signal may be a fixed value preset by the input / output unit 310, but is not limited thereto and may also be a randomized value.

[0089] 6, the electrocardiogram canceller 320 can divide the ECG 10 signal into a plurality of partial signals S1, S2, S3, ... each having a size corresponding to the window by cutting the ECG 10 signal while sliding the window along the time axis. In this case, the size of the window for cutting the ECG 10 signal can be a constant value preset by the input / output unit 310.

[0090] Meanwhile, the signal of the electrocardiogram (ECG) 10 may contain noise generated due to the user's behavior during the measurement process, etc. Therefore, the ECG canceller 320 can remove noise contained in the divided partial signals S1, S2, S3, ...

[0091] Specifically, the ECG canceller 320 can identify displacements of P waves, PQ waves, QRS waves, ST waves, and T waves included in each of the divided partial signals S1, S2, S3, .... The ECG canceller 320 can generate a normal distribution of the displacements of the identified P waves, PQ waves, QRS waves, ST waves, and T waves. Then, the ECG canceller 320 can selectively remove only partial signals having displacements of P waves, PQ waves, QRS waves, ST waves, or T waves included in a predetermined noise range from the generated normal distribution.

[0092] 3, the health condition analysis unit 325 can analyze the electrocardiogram (ECG) 10 signal using artificial intelligence (AI) to identify the heart pump state corresponding to the user's health condition. Also, the health condition analysis unit 325 can analyze the electrocardiogram (ECG) 10 signal using artificial intelligence (AI) to further identify the maximum oxygen uptake corresponding to the user's health condition.

[0093] Preferentially, the health condition analysis unit 325 can input the multiple partial signals S1, S2, S3, ... divided by the electrocardiogram release unit 320 into an artificial intelligence (AI) respectively, and obtain multiple probability values P1, P2, P3, ..., Pn from the artificial intelligence (AI).

[0094] Since the artificial intelligence (AI) according to one embodiment of the present invention includes a first artificial neural network (ANN) to a third artificial neural network (ANN), the health condition analysis unit 325 can input a plurality of partial signals S1, S2, S3, ... to the first artificial neural network (ANN) to the third artificial neural network (ANN), respectively.

[0095] Specifically, the health condition analysis unit 325 may extract one or more features from the displacements of P waves, PQ waves, and QRS waves included in each of the divided partial signals S1, S2, S3, etc. according to a predetermined feature extraction standard. The ECG cancellation unit 320 may one-hot encode the extracted one or more features to generate a vector of electrocardiogram (ECG) signal features. The ECG cancellation unit 320 may embed the generated ECG signal feature vector to generate a dense vector with reduced dimensions. The ECG cancellation unit 320 may input the generated dense vector to a GMF layer and an MLP layer of a first artificial neural network (ANN).

[0096] The health condition analysis unit 325 may extract one or more features based on a preset feature extraction criterion from the displacements of the QRS wave, ST wave, and T wave included in each of the divided partial signals S1, S2, S3, etc. The ECG cancellation unit 320 may perform one-hot encoding on the extracted one or more features to generate a vector of electrocardiogram (ECG) signal features. The ECG cancellation unit 320 may embed the generated ECG signal feature vector to generate a dense vector with reduced dimensions. The ECG cancellation unit 320 may input the generated dense vector to a GMF layer and an MLP layer of a second artificial neural network (ANN).

[0097] The health condition analysis unit 325 may extract one or more features from all waveform displacements included in each of the divided partial signals S1, S2, S3, etc. according to a predetermined feature extraction criterion. The ECG cancellation unit 320 may one-hot encode the extracted one or more features to generate a vector of electrocardiogram (ECG) signal features. The ECG cancellation unit 320 may embed the generated ECG signal feature vector to generate a dense vector with reduced dimensions. The ECG cancellation unit 320 may input the generated dense vector to a GMF layer and an MLP layer of a third artificial neural network (ANN) and a fourth artificial neural network (ANN).

[0098] The health condition analysis unit 325 can then obtain a plurality of probability values P1, P2, P3, ..., Pn corresponding to a plurality of partial signals S1, S2, S3, ... from each of the first artificial neural network (ANN) to the third artificial neural network (ANN).

[0099] Next, the health condition analysis unit 325 can identify the user's cardiac pump status based on the plurality of probability values P1, P2, P3, ..., Pn obtained from the first artificial neural network (ANN) to the third artificial neural network (ANN), and can identify the user's maximum oxygen uptake based on the plurality of probability values P1, P2, P3, ..., Pn obtained from the fourth artificial neural network (ANN).

[0100] Specifically, the health condition analysis unit 325 can identify the user's heart rate based on the number of waveforms included in an electrocardiogram (ECG) signal per unit time. The health condition analysis unit 325 can determine a preload state value based on a plurality of probability values P1, P2, P3, ..., Pn obtained from a first artificial neural network (ANN). The health condition analysis unit 325 can determine an afterload state value based on a plurality of probability values P1, P2, P3, ..., Pn obtained from a second artificial neural network (ANN). The health condition analysis unit 325 can determine a myocardial contractility state value based on a plurality of probability values P1, P2, P3, ..., Pn obtained from a third artificial neural network (ANN). The health condition analysis unit 325 can then identify the heart pump state based on the identified user's heart rate, preload state value, afterload state value, and myocardial contractility state value.

[0101] Additionally, the health condition analysis unit 325 can identify the user's maximum oxygen uptake based on a plurality of probability values P1, P2, P3, . . . , Pn obtained from a fourth artificial neural network (ANN).

[0102] 7, the health condition analysis unit 325 may remove outliers from the plurality of probability values P1, P2, P3, ..., Pn obtained from each artificial neural network (ANN) based on a predetermined validated range. The health condition analysis unit 325 may determine the status value or maximum oxygen uptake of the cardiac pump from an average value Pavr of the plurality of probability values P1, P2, ..., Pm from which the outliers have been removed or a median value Pmed of the plurality of probability values P1, P2, ..., Pm from which the outliers have been removed.

[0103] That is, the health condition analysis unit 325 may remove outliers from the plurality of probability values P1, P2, P3, ..., Pn obtained from the first artificial neural network (ANN) and determine an average or median value of the probability values P1, P2, P3, ..., Pn from which the outliers have been removed as a preload state value. The health condition analysis unit 325 may remove outliers from the plurality of probability values P1, P2, P3, ..., Pn obtained from the second artificial neural network (ANN) and determine an average or median value of the probability values P1, P2, P3, ..., Pn from which the outliers have been removed as a postload state value. The health condition analysis unit 325 may remove outliers from the plurality of probability values P1, P2, P3, ..., Pn obtained from the third artificial neural network (ANN) and determine an average or median value of the probability values P1, P2, P3, ..., Pn from which the outliers have been removed as a state value of the myocardial contractility.The health condition analysis unit 325 may remove outliers from the plurality of probability values P1, P2, P3, ..., Pn obtained from the first artificial neural network (ANN) and determine an average or median value of the probability values P1, P2, P3, ..., Pn from which the outliers have been removed as a state value of the myocardial contractility.

[0104] 3, the exercise determination unit 330 may determine an exercise to be recommended to the user based on the cardiac pump status and maximum oxygen intake identified by the health condition analysis unit 325. The exercise determined by the exercise determination unit 330 may include one or more of the type of exercise, the intensity of exercise, and the duration of exercise.

[0105] 8, the exercise determination unit 330 may first determine the type of exercise to be recommended to the user in accordance with the heart pump status identified by the health condition analysis unit 325. To this end, the exercise determination unit 330 may determine an exercise type that matches the heart pump status of the user from a pre-defined exercise dictionary. Then, the exercise determination unit 330 may determine the maximum oxygen uptake identified by the health condition analysis unit 325 and the exercise intensity that matches the previously determined exercise type.

[0106] Additionally, the exercise determination unit 330 may calculate the user's basal metabolism based on the user's gender, height, and weight, which are input in advance. The exercise determination unit 330 may calculate the user's activity metabolism by applying the previously determined exercise type and exercise intensity to the calculated basal metabolism. The exercise determination unit 330 may then determine the exercise duration based on the calculated activity metabolism.

[0107] According to an embodiment of the present invention, the exercise determiner 330 may identify a preset location of the user device 200 corresponding to the user. To this end, the exercise determiner 330 may receive a location value of the user device 200 directly from the user device 200 or may receive a location value from the user. The exercise determiner 330 may search for an altitude and weather conditions corresponding to the identified location of the user device 200. Then, the exercise determiner 330 may apply a weight to the maximum oxygen uptake identified by the health status analyzer 325 according to the searched altitude and weather conditions.

[0108] According to another embodiment of the present invention, the exercise determination unit 330 may determine the type of exercise by reflecting at least one of a time weight corresponding to the time point at which the electrocardiogram (ECG) 10 signal is measured and a human weight preset corresponding to the user.

[0109] In this case, the time weighting value may be a value set corresponding to the time and season at which the signal of the electrocardiogram (ECG) 10 is measured, and the personal weighting value may be a value set corresponding to the heart pump status identified based on the signal of the electrocardiogram (ECG) 10 measured for people registered as friends of the user.

[0110] The exercise determination unit 330 may transmit information about the determined exercise to a user equipment (UE) 200 that is preset corresponding to the user. In this case, the information about the exercise may include one or more of the type of exercise, the intensity of the exercise, and the duration of the exercise.

[0111] Hereinafter, hardware for implementing the logical components of the exercise recommendation server 300 having the above-mentioned features will be described in more detail.

[0112] FIG. 9 is a hardware configuration diagram of an exercise recommendation server according to an embodiment of the present invention.

[0113] As shown in FIG. 9, the exercise recommendation server 300 according to one embodiment of the present invention may include a processor 350, a memory 355, a transceiver 360, an input / output device 365, a data bus 370, and a storage 375.

[0114] Specifically, the processor 350 can implement the operations and functions of the exercise recommendation server 300 based on commands from software 380 a that implements an exercise recommendation method and resides in the memory 355 .

[0115] The memory 355 may be loaded with software 380b that implements the exercise recommendation method and is stored in the storage 375.

[0116] The transceiver 360 may transmit data to and receive data from one or more of the electrocardiograph 100 and the user equipment (UE) 200 .

[0117] The input / output device 365 can receive input of signals required for the operation of the exercise recommendation server 300 or output calculation results to the outside according to instructions from the processor 350 .

[0118] The data bus 370 is connected to the processor 350, the memory 355, the transceiver 360, the input / output device 365, and the storage 375, respectively, and can serve as a communication path for transmitting signals between the respective components.

[0119] The storage 375 may store an application programming interface (API), library files, resource files, etc., required for executing software 380a implementing an exercise recommendation method according to various embodiments of the present invention. The storage 375 may store software 380b implementing an exercise recommendation method according to various embodiments of the present invention. The storage 375 may also include a database 385 for storing artificial intelligence (AI) learning data, an exercise dictionary, various setting values, etc.

[0120] According to one embodiment of the present invention, the software 380a, 380b for implementing the exercise recommendation method, which is resident in the memory 355 or stored in the storage 375, may be a computer program recorded on a recording medium to cause the processor 350 to execute the steps of acquiring an electrocardiogram (ECG) 10 signal measured on a user via the transceiver 360 or the input / output device 365, analyzing the electrocardiogram (ECG) 10 signal using pre-trained artificial intelligence (AI) by the processor 350 to identify the user's cardiac pump status, and determining the type of exercise to be recommended to the user in accordance with the identified cardiac pump status by the processor 350.

[0121] More specifically, the processor 350 may include, but is not limited to, one or more of a central processing unit (CPU), an application-specific integrated circuit (ASIC), a chipset, and a logic circuit.

[0122] The memory 355 may include, but is not limited to, one or more of a read-only memory (ROM), a random access memory (RAM), a flash memory, and a memory card.

[0123] The input / output device 365 may include, but is not limited to, one or more of input devices such as a button, a switch, a keyboard, a mouse, a joystick, and a touch screen, and output devices such as a liquid crystal display (LCD), a light emitting diode (LED), an organic light emitting diode (OLED), an active matrix organic light emitting diode (AMOLED), a printer, and a plotter.

[0124] When the embodiments described herein are implemented using software, the methods described above may be implemented as modules (processes, functions, etc.) that perform the respective functions described above. Each module may reside in memory 355 and be executed by processor 350. Memory 355 may be internal or external to processor 350 and may be coupled to processor 350 via various known means.

[0125] 9 may be implemented by various means (e.g., hardware, firmware, software, or a combination thereof). When implemented by hardware, an embodiment of the present invention may be implemented by one or more Application Specific Integrated Circuits (ASICs), Digital Signal Processors (DSPs), Digital Signal Processing Devices (DSPDs), Programmable Logic Devices (PLDs), Field Programmable Gate Arrays (FPGAs), processors, controllers, microcontrollers, microprocessors, etc.

[0126] Furthermore, when embodied in firmware or software, an embodiment of the present invention may be embodied in the form of modules, procedures, functions, etc. that perform the above-described functions or operations and recorded on a recording medium that can be read by various computer means. Here, the recording medium may include program instructions, data files, data structures, etc., alone or in combination.

[0127] The program instructions recorded on the recording medium may be specially designed and constructed for the present invention, or may be of the type well known and available to those of ordinary skill in the computer software art. For example, recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as compact disk read-only memories (CD-ROMs) and digital video disks (DVDs), magneto-optical media such as floptical disks, and hardware devices specially configured to store and execute program instructions, such as read-only memories (ROMs), random access memories (RAMs), flash memories, etc.

[0128] Examples of program instructions include not only machine code, such as produced by a compiler, but also high-level language code that is executable by a computer via an interpreter, etc. Such hardware devices may be configured to act as one or more software modules to perform the operations of the present invention, and vice versa.

[0129] The operation of the exercise recommendation server 300 as described above will now be described in detail.

[0130] FIG. 10 is a flowchart illustrating an exercise recommendation method according to an embodiment of the present invention.

[0131] As shown in FIG. 10, an exercise recommendation server 300 according to an embodiment of the present invention can acquire an electrocardiogram (ECG) 10 signal measured from a user (S100).

[0132] Specifically, the exercise recommendation server 300 can receive an electrocardiogram (ECG) 10 signal directly from the electrocardiograph 100, or can receive a signal directly from a user or other person.

[0133] Next, the exercise recommendation server 300 can divide the signal of the electrocardiogram (ECG) 10 into a plurality of partial signals S1, S2, S3, . . . according to the time series priorities measured from the user (S200).

[0134] According to one embodiment, the exercise recommendation server 300 can divide the electrocardiogram (ECG) 10 signal into a plurality of partial signals S1, S2, S3, ... by cutting the electrocardiogram (ECG) 10 signal discretely at regular sizes without overlapping regions. According to another embodiment, the exercise recommendation server 300 can divide the electrocardiogram (ECG) 10 signal into a plurality of partial signals S1, S2, S3, ... having sizes corresponding to the windows by cutting the electrocardiogram (ECG) 10 signal while sliding a window along the time axis.

[0135] Meanwhile, the exercise recommendation server 300 can also remove noise contained in the divided partial signals S1, S2, S3, . . .

[0136] Next, the exercise recommendation server 300 can analyze the plurality of partial signals S1, S2, S3, ... using artificial intelligence (AI) to obtain a plurality of probability values P1, P2, P3, ..., Pn (S300).

[0137] Specifically, the exercise recommendation server 300 may extract one or more features from the displacements of P waves, PQ waves, and QRS waves included in each of the divided partial signals S1, S2, S3, etc. according to a preset feature extraction standard. The exercise recommendation server 300 may one-hot encode the extracted one or more features to generate a vector of electrocardiogram (ECG) signal features. The exercise recommendation server 300 may embed the generated vector of electrocardiogram (ECG) signal features to generate a dense vector with reduced dimensions. The exercise recommendation server 300 may input the generated dense vector to a GMF layer and an MLP layer of a first artificial neural network (ANN).

[0138] The exercise recommendation server 300 may extract one or more features from all waveform variations included in each of the divided partial signals S1, S2, S3, etc. according to a preset feature extraction standard. The exercise recommendation server 300 may one-hot encode the extracted one or more features to generate a vector of electrocardiogram (ECG) signal features. The exercise recommendation server 300 may embed the generated vector of electrocardiogram (ECG) signal features to generate a dense vector with reduced dimensions. The exercise recommendation server 300 may input the generated dense vector to the GMF layer and MLP layer of a third artificial neural network (ANN) and a fourth artificial neural network (ANN).

[0139] The exercise recommendation server 300 can obtain a plurality of probability values P1, P2, P3, ..., Pn corresponding to a plurality of partial signals S1, S2, S3, ... from each of the first artificial neural network (ANN) to the fourth artificial neural network (ANN).

[0140] Then, the exercise recommendation server 300 can identify the user's cardiac pump status based on the plurality of probability values P1, P2, P3, ..., Pn obtained from the first artificial neural network (ANN) to the third artificial neural network (ANN), and identify the user's maximum oxygen uptake based on the plurality of probability values P1, P2, P3, ..., Pn obtained from the fourth artificial neural network (ANN) (S400).

[0141] Specifically, the exercise recommendation server 300 may remove outliers from the plurality of probability values P1, P2, P3, ..., Pn obtained from each artificial neural network (ANN) based on a preset valid range. The exercise recommendation server 300 may determine the status value or maximum oxygen uptake of the cardiac pump using an average value Pavr of the plurality of probability values P1, P2, ..., Pm from which the outliers have been removed or a median value Pmed of the plurality of probability values P1, P2, ..., Pm from which the outliers have been removed.

[0142] Then, the exercise recommendation server 300 may determine an exercise to be recommended to the user based on the identified cardiac pump status and maximum oxygen uptake (S500). In this case, the determined exercise may include one or more of the type of exercise, the intensity of exercise, and the duration of exercise.

[0143] Specifically, the exercise recommendation server 300 may determine an exercise type that matches the user's cardiac pumping state from a preset exercise dictionary. The exercise recommendation server 300 may determine a maximum oxygen uptake and an exercise intensity that matches the previously determined exercise type. Additionally, the exercise recommendation server 300 may calculate the user's activity metabolic rate by applying the previously determined exercise type and exercise intensity to the user's basal metabolic rate, which is calculated based on the user's gender, height, and weight, which are previously input corresponding to the user, and may determine the duration of exercise based on the calculated activity metabolic rate.

[0144] Finally, the exercise recommendation server 300 may transmit information about the determined exercise to a user equipment (UE) 200 pre-configured in correspondence with the user (S600).

[0145] As described above, the present specification and drawings have disclosed preferred embodiments of the present invention. However, it will be obvious to those skilled in the art to which the present invention pertains that other modifications based on the technical concept of the present invention are possible in addition to the disclosed embodiments. Furthermore, although specific terms are used in the present specification and drawings, these are used in general terms merely to facilitate the description of the present invention and to facilitate understanding of the invention, and are not intended to limit the scope of the present invention. Therefore, the above detailed description should not be construed as limiting in all respects, but should be considered as illustrative. The scope of the present invention should be determined by reasonable interpretation of the appended claims, and all modifications within the equivalent scope of the present invention are included within the scope of the present invention.

Claims

1. acquiring an electrocardiogram (ECG) signal measured from a user; analyzing the electrocardiogram (ECG) signal using pre-trained artificial intelligence (AI) to identify the user's cardiac pump status; determining a type of exercise to be recommended to the user in response to the identified cardiac pump status.

2. The identifying step comprises: dividing the electrocardiogram (ECG) signal into a plurality of sub-signals according to their chronological order; inputting the divided partial signals into the artificial intelligence (AI), respectively, and obtaining a plurality of probability values from the artificial intelligence (AI); and identifying a cardiac pump status of the user based on the obtained plurality of probability values.

3. 3. The exercise recommendation method of claim 2, wherein the artificial intelligence (AI) comprises a first artificial neural network (ANN) related to a preload corresponding to an end diastolic volume of the heart, a second artificial neural network (ANN) related to an afterload corresponding to a resistance the heart experiences while contracting, and a third artificial neural network (ANN) related to a myocardial contractility of the heart.

4. 4. The exercise recommendation method of claim 3, wherein the identifying the heart pump state includes identifying a heart rate based on a number of waveforms included in the electrocardiogram (ECG) signal per unit time, determining the state value of preload based on a plurality of probability values obtained from the first artificial neural network (ANN), determining the state value of afterload based on a plurality of probability values obtained from the second artificial neural network (ANN), and determining a state value of myocardial contractility based on a plurality of probability values obtained from the third artificial neural network (ANN), and identifying the heart pump state based on the heart rate, the state value of preload, the state value of afterload, and the state value of myocardial contractility.

5. the first artificial neural network is trained based on displacements of P waves, P-Q waves, and QRS waves included in each of the divided partial signals; the second artificial neural network is trained based on displacements of a QRS wave, an ST wave, and a T wave included in each of the divided partial signals; The exercise recommendation method according to claim 3, wherein the third artificial neural network is trained based on the displacements of all waveforms included in each of the divided partial signals.

6. 2. The exercise recommendation method of claim 1, wherein the identifying the cardiac pump status includes analyzing the electrocardiogram (ECG) signal using the artificial intelligence (AI) to identify the user's maximum oxygen uptake.

7. 7. The exercise recommendation method of claim 6, wherein the determining the type of exercise includes determining an exercise type that matches the user's cardiac pumping state from a preset exercise dictionary, and determining an exercise intensity that matches the determined exercise type and the maximum oxygen uptake.

8. 10. The exercise recommendation method of claim 7, wherein the determining the type of exercise includes calculating a basal metabolic rate of the user based on the user's gender, height, and weight previously input corresponding to the user, calculating an activity metabolic rate of the user by applying the determined type of exercise and intensity of exercise to the calculated basal metabolic rate, and determining a duration of the exercise based on the calculated activity metabolic rate.

9. 2. The exercise recommendation method of claim 1, further comprising, after determining the type of exercise, transmitting information including the determined type of exercise to a user equipment (UE) pre-configured corresponding to the user.

10. A memory; a transceiver; an input / output device; a processor for processing instructions resident in the memory; acquiring, by the processor, an electrocardiogram (ECG) signal measured from a user via the transceiver or input / output device; analyzing the electrocardiogram (ECG) signal using pre-trained artificial intelligence (AI) by the processor to identify the user's heart pump status; and determining, by the processor, a type of exercise to be recommended to the user in response to the identified cardiac pump status.

Images