Electrocardiogram-based recipe schedule recommendation method and computer program stored in storage medium for executing same

Abstract

A method of recommending a recipe schedule suitable for a real-time health status of a user based on an electrocardiogram (ECG) is disclosed. The method may include the steps of: acquiring an electrocardiogram (ECG) signal measured with a user as a subject; analyzing the electrocardiogram (ECG) signal by using a pre-learned artificial intelligence (AI), and identifying at least one of a short nutritional ingredient and an excess nutritional ingredient corresponding to the health status of the user; and setting a recipe schedule (menu) including one or more diets capable of supplementing the identified shortage of nutritional ingredients and avoiding the excess nutritional ingredients. In this way, the user can be recommended to a recipe schedule that is optimized for the state of health at that point in time by merely measuring an electrocardiogram (ECG).

Description

Technical Field

[0001] The present invention discloses a smart health care, more specifically,

[0002] A method for recommending suitable

[0003] Method for calculating a diet plan based on a person's real-time health status and storing the diet plan in a storage medium for executing the method

[0004] A computer program on a storage medium. Background Art

[0005] An electrocardiogram (ECG) records the electrical activity of the heart muscle in a graph.

[0006] The following is a brief description of the conduction system of the heart.

[0007] The sinoatrial node (SA Node) regularly generates electrical signals to induce heart contraction.

[0008] Atrioventricular Node (AV Node) in the sinoatrial node (SA Node)

[0009] The electrical signal arrives through the atrium and is then transmitted with a slight delay.

[0010] The electrical signal from the AV node passes through the bundle of His and the Purkinje fibers.

[0011] Purkinje fibers spread throughout the ventricle.

[0012] The following are the electrocardiogram (ECG) waves that can be observed along the conduction process of the heart.

[0013] The electrical signal generated by the sinoatrial node (SA Node) polarizes the atrium and causes the myocardium of the valve to

[0014] P can be observed in the process of atrial re-depolarization after contraction, which allows the valve myocardium to relax again.

[0015] The atrioventricular node (AV Node) is formed to prevent the ventricles from immediately reacting when the atria contract.

[0016] Therefore, the PQ wave can be observed during the delay of the electrical signal.

[0017] (AV Node) The delayed signal arrives (Q) and the ventricle polarizes (R) and then depolarizes

[0018] The QRS complex can be observed during the process of ventricular contraction.

[0019] During the rest period, when the blood in the chamber moves to all parts of the body and the electrical signals do not stimulate the heart

[0020] ST waves can be observed. Moreover, after the ventricles are repolarized to a weaker degree,

[0021] The T wave (T) is observed during the simultaneous relaxation of the ventricular and valvular myocardium.

[0022] In addition, U waves due to repolarization of the intraventricular septum are also observed in some cases.

[0023] As mentioned above, the order of appearance, shape, size and interval of the multiple waveforms (P wave, PQ wave, QRS wave, ST wave, T wave and U wave) observed by the electrocardiogram (ECG) can change into various forms according to the health status. In other words, by analyzing the order of appearance, shape, size and interval of the multiple waveforms contained in the electrocardiogram (ECG), the health status can be inferred, and further, the various causes leading to the health status can be inferred.

[0024] On the other hand, artificial neural network (ANN) is one of the detailed methodologies of machine learning, which means an algorithm with a network form of multiple neurons connected to each other, which is the basic computing unit corresponding to human nerve cells. Representative artificial neural networks (ANN) include multi-layer perceptron (MLP), recurrent neural network (RNN), convolutional neural network (CNN), etc.

[0025] Specifically, a multilayer perceptron (MLP) is an artificial neural network (ANN) having multiple hidden layers between an input layer and an output layer. A recurrent neural network (RNN) is an artificial neural network (ANN) having a recursive connection structure in which the output of neurons is re-input to the input layer. Moreover, a convolutional neural network (CNN) is an artificial neural network (ANN) that extracts feature maps from multiple convolutional layers by imitating the structure of human visual nerves and extracts only important parts from the feature maps after reducing the dimensions of the matrix by subsampling.

[0026] On the other hand, a menu or diet is a plan related to the type, amount, and order of food intake by an individual during a meal. In determining the menu, not only personal preferences but also nutrients required according to the individual's health status should be considered. Therefore, existing services that recommend or suggest diets or menu plans only consider the nutrients required based on the general health status of the user based on static factors such as height, weight, and obesity.

[0027] However, the nutrients required for health status vary from time to time depending on the type and intensity of ongoing or previous activities, the type and amount of food consumed previously, etc. For example, a person who has just completed high-intensity exercise needs macronutrients such as protein and water, while a person who suddenly suffers from diarrhea or vomiting needs micronutrients such as minerals. In addition, it is not realistic to require non-medical personnel to regularly draw blood (blood drawing) in order to confirm their health status and required nutrients in their daily lives.

[0028] Therefore, there is a need to develop a tool that does not require the user to perform actions that make them feel burdened but can immediately recommend a diet plan that is suitable for the user's real-time health status. Summary of the invention

[0029] Technical issues

[0030] An object of the present invention is to provide a method that can recommend a diet plan suitable for the user's real-time health status based on an electrocardiogram (ECG).

[0031] Another object of the present invention is to provide a computer program stored in a storage medium for executing the following method, which can recommend a diet plan suitable for the user's real-time health status based on an electrocardiogram (ECG).

[0032] The technical problems of the present invention are not limited to the technical problems mentioned above, and ordinary technicians in the technical field to which the present invention belongs can clearly understand other technical problems not mentioned above from the following records.

[0033] Technical Solution

[0034] In order to solve the technical problems mentioned above, the present invention discloses a method for recommending a diet plan suitable for the real-time health status of a user based on an electrocardiogram (ECG). The method may include the following steps: an electrocardiogram signal acquisition step, acquiring an electrocardiogram (ECG) signal measured with the user as an object; an identification step, using pre-learned artificial intelligence (AI) to analyze the ECG signal, identifying at least one nutrient deficiency and excess nutrient corresponding to the health status of the user; and a diet plan setting step, setting a diet plan (menu) containing more than one diet that can supplement the identified nutrient deficiency and avoid the excess nutrient.

[0035] The method may further include the following step after the recipe schedule setting step: transmitting the information of the set recipe schedule to a user equipment (UE) preset corresponding to the user.

[0036] More specifically, the identification step may include the following steps: dividing the electrocardiogram (ECG) signal into multiple segmented signals according to the order of time, inputting the multiple segmented signals into the artificial intelligence (AI) respectively, obtaining multiple probability values ​​from the artificial intelligence (AI); and judging whether the nutrients are scarce, normal or excessive based on the multiple probability values ​​obtained.

[0037] According to an embodiment of the present invention, in the step of segmenting into a plurality of segmented signals, a grid (window) of a preset size can be moved along the time axis and the electrocardiogram (ECG) signal can be segmented into a plurality of segmented signals having a size corresponding to the grid.

[0038] On the other hand, the artificial intelligence (AI) can include a first artificial neural network (ANN) about electrolytes, a second artificial neural network (ANN) about proteins, and a third artificial neural network (ANN) about water. In this case, in the step of making the judgment, it is judged whether electrolytes are short, normal, or in excess based on a plurality of probability values ​​taken from the first artificial neural network (ANN), it is judged whether proteins are short, normal, or in excess based on a plurality of probability values ​​taken from the second artificial neural network (ANN), and it is judged whether water is short, normal, or in excess based on a plurality of probability values ​​taken from the third artificial neural network (ANN).

[0039] Moreover, the hidden layers of the first to third artificial neural networks (ANN) may include a generalized matrix factorization (GMF) for learning the linear relationship (linearity relation) between the characteristics of the electrocardiogram (ECG) signal and the nutritional content, and a multi-layer perceptron (MLP) for learning the non-linear relationship (non-linearity relation) between the characteristics of the electrocardiogram (ECG) signal and the nutritional content.

[0040] In the step of performing the judgment in the method, outliers included in the multiple probability values ​​obtained are eliminated based on a preset valid range, and one of the soft voting results based on the average value of the multiple probability values ​​after eliminating the outliers or the hard voting results based on the majority decision of the multiple probability values ​​after eliminating the outliers can be used to judge whether the nutrient is short, normal or excessive.

[0041] The recipe plan setting step may include the following steps: identifying two or more diets from a preset food dictionary that do not contain the excess nutrients or only contain a preset threshold content and include the identified missing nutrients; identifying a preset human weight corresponding to the user and a time weight corresponding to the measurement time point of the electrocardiogram (ECG) signal; and reflecting the human weight and time weight in the two or more identified diets and selecting a diet.

[0042] Moreover, in the identification step, if the first moment and the second moment have a difference within a preset minimum biotransformation time, and if the excess nutrients identified based on the electrocardiogram (ECG) signal measured with the user as the object at the first moment and the shortage nutrients identified based on the electrocardiogram (ECG) signal measured with the user as the object at the second moment are the same as each other, it can be determined that the user needs to be warned.

[0043] In order to achieve the technical problem as mentioned above, the present invention proposes a computer program stored in a storage medium for executing a method for recommending a diet plan suitable for the real-time health status of a user based on an electrocardiogram (ECG). The computer program can be combined with the following computing device, which includes a memory, a transceiver, an input / output device, and a processor for processing commands resident in the memory. Moreover, the computer program can be a computer program stored in a storage medium for executing the following steps, which are: the processor obtains an electrocardiogram (ECG) signal measured with the user as an object through the transceiver or the input / output device; the processor analyzes the electrocardiogram (ECG) signal using pre-learned artificial intelligence (AI) to identify at least one deficient nutrient and excess nutrient corresponding to the health status of the user; and the processor sets a diet plan containing more than one diet that can supplement the identified deficient nutrient and avoid the excess nutrient.

[0044] Details of other embodiments are included in the following detailed description and drawings.

[0045] Effects of the Invention

[0046] According to various embodiments of the present invention, a user can be recommended a diet plan optimized for the user's health status at that point in time simply by measuring an electrocardiogram (ECG).

[0047] The effects of the present invention are not limited to the effects mentioned above, and a person skilled in the art to which the present invention belongs can clearly understand other effects not mentioned above from the description of the scope of protection claimed in the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is an exemplary diagram for explaining a recipe plan recommendation device according to an embodiment of the present invention.

[0049] Figure 2 It is an illustrative diagram for explaining a recipe plan recommendation system according to an embodiment of the present invention.

[0050] Figure 3 It is a logical configuration diagram of a recipe plan recommendation server according to an embodiment of the present invention.

[0051] Figure 4 It is an example diagram of an artificial neural network (ANN) for illustrating an embodiment of the present invention.

[0052] Figure 5 and Figure 6 is an exemplary diagram for explaining a process of segmenting an electrocardiogram (ECG) signal according to some embodiments of the present invention.

[0053] Figure 7 and Figure 8 It is an exemplary diagram for explaining the process of determining whether each nutrient component is short, normal or excessive according to several embodiments of the present invention.

[0054] Fig. 9 This is a hardware configuration diagram of a recipe plan recommendation server according to an embodiment of the present invention.

[0055] Fig.10 It is a flowchart for illustrating a method for recommending a recipe plan according to an embodiment of the present invention. DETAILED DESCRIPTION

[0056] The technical terms used in this specification are only for describing specific embodiments and are not intended to limit the present invention. Furthermore, unless a different definition is specifically given in this specification, the technical terms used in this specification should be interpreted as the meanings commonly understood by ordinary technicians in the technical field to which the present invention belongs, and should not be interpreted as overly general or overly narrowed. Furthermore, when a technical term used in this specification is an erroneous term that cannot correctly express the spirit of the present invention, it should be replaced with a technical term that can be correctly understood by ordinary technicians in the technical field to which the present invention belongs. Furthermore, the general terms used in the present invention should be interpreted according to the content defined in the dictionary or according to the context of the article, and should not be interpreted as overly narrowed.

[0057] Furthermore, unless it can be clearly distinguished in the context of the sentence, the singular expressions used in this specification also include the plural. In this application, the terms such as "including" or "having" should not be interpreted as necessarily including all the multiple elements or steps recorded in the specification, and may not include some of the elements or steps, or may further include other elements or steps.

[0058] Furthermore, the terms used in this specification that include ordinal numbers such as "first" and "second" may be used to describe various elements, but the elements shall not be limited to the terms. The purpose of using the terms is only to distinguish one element from other elements. For example, the first element may be named the second element without departing from the scope of the claims of the present invention, and similarly, the second element may be named the first element.

[0059] When an element is recorded as being "connected" or "accessed" to another element, it can be understood as being directly connected to or accessed to the other element, but it can also be understood as other elements being present in between. Conversely, when an element is recorded as being "directly connected" or "directly accessed" to another element, it should be understood as other elements not being present in between.

[0060] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. The same or similar elements are given the same figure mark without being affected by the graphic symbols, and repeated descriptions thereof are omitted. Moreover, when describing the present invention below, if it is considered that the specific description of the relevant known technology may confuse the main purpose of the present invention, its detailed description will be omitted. Moreover, the purpose of the drawings is only to make the spirit of the present invention easy to understand, and the drawings cannot be interpreted as limiting the spirit of the present invention. The spirit of the present invention should be extended to all changes, equivalents and even substitutes in addition to the drawings.

[0061] As mentioned above, existing services that recommend or suggest diets or meal plans simply propose the nutrients required for general health status based on static factors such as the user's height, weight and obesity. However, the nutrients required for health status may change at any time depending on factors such as the type and intensity of activities currently being performed or previously performed, the type and amount of food previously consumed, etc.

[0062] In order to overcome this restricted situation, the present invention proposes a tool that can recommend a recipe plan suitable for the user's real-time health status.

[0063] Figure 1 It is an exemplary diagram for explaining a recipe plan recommendation device according to an embodiment of the present invention.

[0064] like Figure 1 As shown, the recipe plan recommendation device according to one embodiment of the present invention uses an electrocardiograph (ECG reader, 100) to measure the user's ECG. Fig.10 (ECG), using artificial intelligence (AI) to analyze the user's measured ECG Fig.10 (ECG) can be used to identify the nutrients required according to the user's real-time health status, and a diet plan 20 (menu) based on the identified nutrients can be immediately recommended to the user.

[0065] Therefore, according to one embodiment of the present invention, the user only needs to measure the ECG Fig.10 (ECG) can be used to recommend a diet plan that is optimized for the user's health status at that point in time.

[0066] The following describes in detail a number of devices and methods for implementing the aforementioned features.

[0067] Figure 2 It is an illustrative diagram for explaining a recipe plan recommendation system according to an embodiment of the present invention.

[0068] like Figure 2 As shown, a recipe plan recommendation system of one embodiment of the present invention may include one or more electrocardiographs 100 (including a watch type 100a, a portable type 100b, ..., a dynamic (holter) type 100n), one or more user devices 200 (including a smart phone 200a, a laptop computer 200b, ..., 200n) and a recipe plan recommendation server 300.

[0069] As described above, the multiple elements of the recipe plan recommendation system of one embodiment of the present invention merely illustrate multiple elements that are functionally distinguished, so that two or more elements can be implemented in an integrated manner in an actual physical environment or one element can be implemented separately in an actual physical environment.

[0070] The following describes each element. The electrocardiograph 100 is a device that can measure and record the user's electrocardiogram. Fig.10 (ECG) device.

[0071] Specifically, the electrocardiograph 100 can induce electric potential changes based on myocardial electrical activity by electrodes in contact with the user and amplify the induced electric potential changes and record them in the form of a waveform.

[0072] The electrocardiograph 100 of one embodiment of the present invention can induce an electrocardiogram (ECG) according to one of standard limb leads, unipolar limb leads, and precordial leads. Fig.10 The ECG induction method is not limited thereto. The electrocardiograph 100 may have electrodes for a single lead, 6 leads, or 12 leads, but the number of leads of the electrocardiograph 100 is not limited thereto. Furthermore, the electrocardiograph 100 may have a watch type 100a, a portable type 100b, and a holter type 100n, but the type of the electrocardiograph 100 is not limited thereto.

[0073] The electrocardiogram (ECG) signal (i.e., the recorded waveform signal) measured by the electrocardiograph 100 can be directly transmitted to the recipe plan recommendation server 300 through the network function of the electrocardiograph 100 itself or input into the recipe plan recommendation server 300 via a third tool. For example, the third tool can be any one of a removable media, a user device 200, and a user's data input behavior, but is not limited thereto.

[0074] As a next element, the user device 200 can output information about the recipe schedule set by the recipe schedule recommendation server 300 .

[0075] Specifically, the user device 200 can receive information about the recipe schedule from the recipe schedule recommendation server 300. Also, the user device 200 can output the received information about the recipe schedule.

[0076] The user device 200 of one embodiment of the present invention is not limited to the user equipment (UE) specified in the 3rd Generation Partnership Project (3GPP) or the mobile station (MS) specified in the Institute of Electrical and Electronics Engineers (IEEE). Any device can be permitted by the present invention as long as it can send and receive data with the recipe plan recommendation server 300 and perform calculations based on the sent and received data.

[0077] As an example, the user device 100 can be a fixed computing device such as a desktop computer 200n, a workstation or a server, or a mobile computing device such as a smart phone 200a, a laptop computer 200b, a tablet computer, a phablet, a portable multimedia player (PMP), a personal digital assistant (PDA) or an e-book reader, but is not limited thereto.

[0078] As the next element, the recipe plan recommendation server 300 uses the electrocardiograph measured by the electrocardiograph 100 to Fig.10 The device can identify the nutrients required according to the user's real-time health status based on the ECG, and transmit information about the recipe plan 20 based on the identified nutrients to the user device 200.

[0079] The specific elements and actions of the recipe plan recommendation server 300 described above will be described later in conjunction with Figures 3 to 10 Please explain.

[0080] As described above, one or more electrocardiographs 100, one or more user devices 200 and a recipe plan recommendation server 300 constituting the recipe plan recommendation system can send and receive data using a network composed of one or more of a secure line, a public wired communication network or a mobile communication network that directly connects multiple devices.

[0081] For example, the public wired communication network may include Ethernet, x Digital Subscriber Line (xDSL), Hybrid Fiber Coax (HFC), Fiber To The Home (FTTH), but are not limited thereto. Furthermore, the mobile communication network may include Code Division Multiple Access (CDMA), Wideband CDMA (WCDMA), High Speed ​​Packet Access (HSPA), Long Term Evolution (LTE), and 5th generation mobile telecommunication, but are not limited thereto.

[0082] The elements of the recipe plan recommendation server 300 having the above-mentioned features will be described in detail below.

[0083] Figure 3 It is a logical configuration diagram of a recipe plan recommendation server according to an embodiment of the present invention. Figure 4 It is an example diagram of an artificial neural network (ANN) for illustrating an embodiment of the present invention. Figure 5 and Figure 6 is an illustrative diagram for illustrating a process of segmenting an electrocardiogram (ECG) signal according to some embodiments of the present invention. Figure 7 and Figure 8 It is an exemplary diagram for explaining the process of determining whether each nutrient component is short, normal or excessive according to several embodiments of the present invention.

[0084] like Figure 3 As shown, the recipe plan recommendation server 300 of one embodiment of the present invention may include a communication unit 305, an input and output unit 310, an artificial intelligence learning unit 315, an electrocardiogram decomposition unit 320, a nutritional component analysis unit 325 and a recipe plan setting unit 330.

[0085] As described above, the multiple elements of the recipe plan recommendation server 300 of one embodiment of the present invention merely illustrate multiple elements that are functionally distinguished, so that two or more elements can be implemented in an integrated manner in an actual physical environment or one element can be implemented separately in an actual physical environment.

[0086] Each element is described below. The communication unit 305 can transmit and receive data with the electrocardiograph 100 and the user device 200.

[0087] Specifically, the communication unit 305 can receive the electrocardiograph 100. Fig.10 Furthermore, the communication unit 305 can transmit information about the recipe planner 20 to the user device 200 .

[0088] As the next element, the input-output unit 310 can receive a command input from a manager or output a calculation result through a user interface (UI).

[0089] Specifically, the input-output unit 310 can receive the decomposed ECG. Fig.10 The input-output unit 310 can receive an input of a size of a grid (window) as a benchmark for (ECG) signals. The input-output unit 310 can receive an input of a valid range (validated rage) as a benchmark for removing outliers from multiple probability values ​​taken from artificial intelligence (AI). The input-output unit 310 can receive an input of a food dictionary (food dictionary) in which diets and nutrients are organized to match each other. The input-output unit 310 can receive an input of a threshold content (threshold content) as a warning value for further intake of each excess nutrient. The input-output unit 310 can receive an input of a minimum biotransformation time (minimum biotransformation time) as a benchmark for determining that a special condition has occurred in the user. Moreover, the input-output unit 310 can receive an input of a threshold distance (threshold distance) as a benchmark range for providing restaurant information.

[0090] On the other hand, the input-output unit 310 can receive inputs of activation functions of a generalized matrix decomposition layer intended for an artificial neural network (ANN), activation functions of a multi-layer perceptron layer intended for use, weights, and biases.

[0091] Furthermore, the input / output unit 310 can output the user's ECG. Fig.10 (ECG) signal. The input-output unit 310 can also output information about the diet plan 20 to be recommended to the user.

[0092] As the next element, the artificial intelligence learning unit 315 can Fig.10 Artificial intelligence (AI) learning that identifies nutritional deficiencies and excess nutrients corresponding to the user's health status based on (ECG) signals.

[0093] Specifically, the artificial intelligence (AI) that the artificial intelligence learning unit 315 drives to learn may include a first artificial neural network (ANN) about electrolyte, a second artificial neural network (ANN) about protein, and a third artificial neural network (ANN) about water.

[0094] The artificial intelligence learning unit 315 can make the first artificial neural network (ANN) learn based on the characteristics of the electrocardiogram (ECG) signal and the data on the electrolyte content input by the manager. The artificial intelligence learning unit 315 can make the second artificial neural network (ANN) learn based on the characteristics of the electrocardiogram (ECG) signal and the data on the protein content input by the manager. Moreover, the artificial intelligence learning unit 315 can make the third artificial neural network (ANN) learn based on the characteristics of the electrocardiogram (ECG) signal and the data on the water content input by the manager.

[0095] Here, an electrolyte is a substance that is dissolved in a solvent and then ionized to generate positive and negative ions. The electrolyte of one embodiment of the present invention may include sodium (Na), potassium (Ka) and calcium (Ca), but is not limited thereto. Protein is an organic substance realized by the combination of amino acids. The protein of one embodiment of the present invention may include albumin, but is not limited thereto.

[0096] The first artificial neural network (ANN), the second artificial neural network (ANN), and the third artificial neural network (ANN) as the objects of learning driven by the artificial intelligence learning unit 315 may each include an input layer, a hidden layer, and an output layer.

[0097] In particular, Figure 4 As shown, the hidden layers of the first artificial neural network (ANN) to the third artificial neural network (ANN) of one embodiment of the present invention can be constructed in the form of a mixture of generalized matrix factorization (GMF) and multi-layer perceptron (MLP).

[0098] For the sake of convenience in explanation, all or part of the first to third artificial neural networks (ANN) may be simply recorded as artificial neural networks (ANN).

[0099] In each artificial neural network (ANN) of an embodiment of the present invention, the generalized matrix factorization (GMF) layer is a layer used to learn the linearity relation between the characteristics of the electrocardiogram (ECG) signal and the nutrient content, which can be implemented using the following formula 1.

[0100] Equation 1

[0101] a(h T (p u ⊙q i ))

[0102] At this time, p u is a dense vector of electrocardiogram (ECG) signal features, q i is a dense vector of nutrient content, p u ⊙q i is the Hadamard product between the dense vector of ECG signal features and the dense vector of nutrient content, a is the activation function of the artificial neural network (ANN), h is the edge weight of the hidden layer, and h T is the vector product of the edge weights.

[0103] A dense vector is a vector that is reduced in dimension by embedding a vector expressed by one-hot encoding. In other words, a dense vector is a vector that is reduced in dimension by removing the value of 0 from a sparse vector expressed by one-hot encoding.

[0104] Furthermore, in each artificial neural network (ANN), the multi-layer perceptron layer is a layer used to learn the non-linearity relation between the characteristics of the electrocardiogram (ECG) signal and the nutrient content, which can be achieved using the following formulas 2 and 3.

[0105] Equation 2

[0106]

[0107] Equation 3

[0108]

[0109] At this time, p u is a dense vector of electrocardiogram (ECG) signal features, q iis a dense vector of nutrient content, z is the value of the concatenation of the dense vector of ECG signal features and the dense vector of nutrient content, a is the activation function of the artificial neural network (ANN), W is the weight of each layer of the multilayer perceptron (MLP), and W T is the vector product of 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 σ is the sum of the series.

[0110] Furthermore, each artificial neural network (ANN) can also be constructed in the form of a recurrent neural network having a recursive connection structure in which the probability value output by the output layer is added to the vector related to the nutrient content and then re-input into the input layer.

[0111] As mentioned above, each artificial neural network (ANN) of the artificial intelligence learning unit 315 has a hybrid form of generalized matrix factorization (GMF) and multi-layer perceptron (MLP) and can perform machine learning on the linear relationship between the characteristics of the electrocardiogram (ECG) signal and the nutrient content, and can also perform machine learning on nonlinear relationships.

[0112] Recombination Figure 3 The following elements are described: the electrocardiogram decomposition unit 320 can decompose the electrocardiogram measured by the electrocardiograph 100. Fig.10 (ECG) signal.

[0113] Specifically, the electrocardiogram decomposition unit 320 can obtain the electrocardiogram measured by the user. Fig.10 That is, the electrocardiogram decomposition unit 320 can directly receive the electrocardiogram (ECG) signal from the electrocardiograph 100 through the communication unit 305. Fig.10 (ECG) signal or directly receive the ECG signal through the input and output unit 310 Fig.10 (ECG) signal input.

[0114] The electrocardiogram decomposition unit 320 can separate the electrocardiogram into Fig.10 (ECG) signal is divided into multiple segmentation signals.

[0115] Basically, if Figure 5 As shown, the electrocardiogram decomposition unit 320 Fig.10 The ECG signal is cut individually (discretely) every time it reaches a certain size without repeated areas, so that the ECG signal can be Fig.10(ECG) signal is divided into multiple segment signals S1, S2, S3, ... In this case, it is used to cut the ECG signal individually. Fig.10 The size of the (ECG) signal may be a certain value preset by the input / output unit 310 , but is not limited thereto and may be a randomized value.

[0116] Differently, if Figure 6 As shown, the electrocardiogram decomposition unit 320 uses the electrocardiogram Fig.10 (ECG) signal as the object, while sliding the grid (window) along the time axis, cutting it can Fig.10 The (ECG) signal is segmented into a plurality of segment signals S1, S2, S3, ... having a size corresponding to the grid. In this case, the ECG signal is segmented into a plurality of segment signals S1, S2, S3, ... having a size corresponding to the grid. Fig.10 The size of the grid of the (ECG) signal may be a certain value preset by the input / output unit 310 .

[0117] On the other hand, ECG Fig.10 The (ECG) signal may contain noises generated during the measurement process due to factors such as the user's movements. Therefore, the electrocardiogram decomposition unit 320 can remove the noises contained in the segmented multiple segment signals S1, S2, S3, ...

[0118] Specifically, the electrocardiogram decomposition unit 320 can identify the displacement of the P wave, PQ wave, QRS wave, ST wave and T wave contained in each of the segmented multiple segmented signals S1, S2, S3, ... The electrocardiogram decomposition unit 320 can generate a normal distribution of the multiple displacements of the identified P wave, PQ wave, QRS wave, ST wave and T wave. Moreover, the electrocardiogram decomposition unit 320 can selectively remove only the segmented signals having the displacement of the P wave, PQ wave, QRS wave, ST wave or T wave contained in a preset noise range in the generated normal distribution.

[0119] Recombination Figure 3 The following elements are explained: the nutritional component analysis unit 325 can use artificial intelligence (AI) to analyze the ECG Fig.10 (ECG) signal to identify one or more lacked nutrients and surplus nutrients corresponding to the user's health status.

[0120] Here, the short nutrients are nutrients that need to be taken in high amounts immediately in view of the health status of the user. The excess nutrients are nutrients that do not need to be taken immediately and will not cause health problems in view of the health status of the user. Furthermore, the normal nutrients are nutrients taken in amounts that are necessary for enjoying daily life in view of the health status of the user.

[0121] First, the nutritional component analysis unit 325 can input the multiple segmented signals S1, S2, S3, ... divided by the electrocardiogram decomposition unit 320 into artificial intelligence (AI) respectively, and then obtain multiple probability values ​​P1, P2, P3, ..., Pn from the artificial intelligence (AI).

[0122] The artificial intelligence (AI) of one embodiment of the present invention includes first to third artificial neural networks (ANN), so the nutritional component analysis unit 325 can input multiple segmented signals S1, S2, S3, ... into the first to third artificial neural networks (ANN), respectively.

[0123] Specifically, the electrocardiogram decomposition unit 320 can extract one or more features according to a preset feature extraction benchmark based on the displacement of the P wave, PQ wave, QRS wave, ST wave and T wave contained in each of the divided multiple segmented signals S1, S2, S3, .... The electrocardiogram decomposition unit 320 can uniquely encode the extracted one or more features to generate a vector of electrocardiogram (ECG) signal features. The electrocardiogram decomposition unit 320 can embed the generated vector of electrocardiogram (ECG) signal features to generate a reduced-dimensional dense vector. The electrocardiogram decomposition unit 320 can input the generated dense vector into the generalized matrix decomposition layer and the multi-layer perceptron layer of each of the first artificial neural network (ANN) to the third artificial neural network (ANN). Moreover, the electrocardiogram decomposition unit 320 can obtain multiple probability values ​​P1, P2, P3, ..., Pn corresponding to the multiple segmented signals S1, S2, S3, ... from the first artificial neural network (ANN) to the third artificial neural network (ANN).

[0124] Next, the nutrient component analysis unit 325 can determine whether each nutrient component is in short supply, normal, or excessive based on the obtained multiple probability values ​​P1, P2, P3, ..., Pn.

[0125] Specifically, the nutrient component analysis unit 325 can determine whether electrolytes are in short supply, normal, or in excess based on multiple probability values ​​taken from the first artificial neural network (ANN). The nutrient component analysis unit 325 can determine whether proteins are in short supply, normal, or in excess based on multiple probability values ​​taken from the second artificial neural network (ANN). Moreover, the nutrient component analysis unit 325 can determine whether water is in short supply, normal, or in excess based on multiple probability values ​​taken from the third artificial neural network (ANN).

[0126] like Figure 7 and Figure 8 As shown, the nutritional component analysis unit 325 can remove outliers included in the plurality of probability values ​​P1, P2, P3, ..., Pn taken from each artificial neural network (ANN) based on a preset valid range. The nutritional component analysis unit 325 can determine whether the nutritional component is short, normal or excessive by using a soft voting result based on an average value Pa of the plurality of probability values ​​P1, P2, ..., Pm from which the outliers are removed or a hard voting result based on a majority decision of the plurality of probability values ​​P1, P2, ..., Pm from which the outliers are removed.

[0127] In this case, for soft voting, the nutrient component analysis unit 325 can calculate the average value Pa of the plurality of probability values ​​P1, P2, ..., Pm from which abnormal values ​​are removed. Furthermore, the nutrient component analysis unit 325 can determine a shortage when the calculated average value Pa is less than or equal to a preset shortage criterion (Criteria of lack, Cl), determine a surplus when the calculated average value Pa is greater than or equal to a surplus criterion (Criteria of surplus, Cs), and determine that the food is normal when the calculated average value Pa is greater than the shortage criterion Cl and less than the surplus criterion Cs.

[0128] Different from this, for hard voting, the nutrient component analysis unit 325 can judge shortage, normality or excess based on multiple probability values ​​P1, P2, ..., Pm with abnormal values ​​removed as objects. That is, the nutrient component analysis unit 325 can judge shortage when each probability value Px is less than or equal to the shortage reference Cl, judge excess when each probability value Px is greater than or equal to the excess reference Cs, and judge normality when each probability value Px is greater than the shortage reference Cl and less than the excess reference Cs. Moreover, the nutrient component analysis unit 325 can judge the most judgment result for multiple judgment results as the final result.

[0129] On the other hand, the nutritional component analysis unit 325 can receive the electrocardiogram (ECG) of the user measured at the first time and the second time with a difference within a preset minimum biotransformation time. Fig.10 (ECG) signal. The ECG signal is measured at the first moment with the user as the subject. Fig.10 (ECG) signal based on the excess nutrients identified and the electrocardiogram measured at the second moment for the user Fig.10 When the deficient nutrients identified based on the (ECG) signal are the same, the nutrient analysis unit 325 can determine that the user needs to be warned. If it is determined that the user needs to be warned, the nutrient analysis unit 325 can transmit a message to the user device 200 (UE) corresponding to the user's setting to inform that the body is abnormal.

[0130] Recombination Figure 3 Describing the following elements, the recipe plan setting unit 330 can set a recipe plan including one or more diets that can supplement the deficient nutrients identified by the nutrient component analysis unit 325 and avoid excess nutrients.

[0131] Specifically, the recipe plan setting unit 330 can identify two or more diets from a preset food dictionary that do not contain excess nutrients or only contain insufficient nutrients below a preset threshold content based on information about insufficient nutrients, excess nutrients, and normal nutrients identified by the nutrient analysis unit 325.

[0132] The recipe plan setting unit 330 can identify the human weight pre-set by the user and the ECG Fig.10 The time weight of the (ECG) signal at the time point at which it is measured.

[0133] In this case, the personal weight may correspond to the electrocardiogram (ECG) measured for the person set as the user's cohabitant. Fig.10 The time weight may be a value set according to the type of nutritional deficiency and nutritional excess identified based on the ECG signal. Fig.10 The value is set according to the season or meal time type corresponding to the time point of the (ECG) signal.

[0134] The recipe plan setting unit 330 can select a diet after reflecting the personal weight and the time weight from the two or more identified diets. Furthermore, the recipe plan setting unit 330 can identify one or more recipe plan templates including the selected diet and set a recipe plan using the identified one or more recipe plan templates. In this case, the recipe plan template may be a catalog including two or more diets pre-set based on the compatibility of the diets with each other.

[0135] When setting a recipe schedule using a diet template as described above, the recipe schedule setting unit 330 can further include information about the order of taking multiple ingredients and recommended intake amounts of one or more diets included in the diet template to set the recipe schedule. Moreover, the recipe schedule setting unit 330 can also set the recipe schedule by further including information about restaurants that are within a preset threshold distance from the user device 200 (UE) and can provide one or more diets included in the recipe schedule based on the location information of the user device 200 (UE).

[0136] Furthermore, the recipe schedule setting unit 330 can transmit information about the set recipe schedule to the user device 200 (UE) preset corresponding to the user.

[0137] The following further describes in detail the hardware used to implement the logical elements of the recipe plan recommendation server 300 having the above-mentioned features.

[0138] Fig. 9 This is a hardware configuration diagram of a recipe plan recommendation server according to an embodiment of the present invention.

[0139] like Fig. 9 As shown, the recipe plan recommendation server 300 according to an 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 an auxiliary storage 375 .

[0140] Specifically, the processor 350 can implement the actions and functions of the recipe plan recommendation server 300 based on the instructions of the software 380 a that implements the recipe plan recommendation method resident in the memory 355 .

[0141] The software 380 b implementing the recipe plan recommendation method stored in the auxiliary memory 375 may be resident (loaded) in the memory 355 .

[0142] The transceiver 360 can transmit and receive data with one or more of the electrocardiograph 100 and the user equipment 200 (UE).

[0143] The input-output device 365 can receive the input of the signal required when the recipe plan recommendation server 300 is running or output the calculation result to the outside according to the command of the processor 350.

[0144] The data bus 370 is connected to the processor 350 , the memory 355 , the transceiver 360 , the input / output device 365 , and the auxiliary memory 375 , respectively, and can function as a mobile channel for transmitting signals between the various elements.

[0145] The auxiliary memory 375 can store an application programming interface (API), a library file, and a resource file required for executing the software 380a that implements the recipe plan recommendation method of various embodiments of the present invention. The auxiliary memory 375 can store the software 380b that implements the recipe plan recommendation method of various embodiments of the present invention. In addition, the auxiliary memory 375 may include a database 385 for storing learning data of artificial intelligence (AI), a diet dictionary, a recipe plan template, and various setting values.

[0146] According to an embodiment of the present invention, the software 380a, 380b for implementing the recipe plan recommendation method resident in the memory 355 or stored in the auxiliary memory 375 may be a computer program stored in a storage medium for executing the following steps: the processor 350 obtains the electrocardiogram (ECG) of the user as the object through the transceiver 360 or the input / output device 365; Fig.10 (ECG) signal; processor 350 uses pre-learned artificial intelligence (AI) to analyze the ECG Fig.10 (ECG) signal to identify at least one nutrient deficiency and excess nutrient corresponding to the user's health status; and the processor 350 sets a recipe plan containing more than one diet that can supplement the identified nutrient deficiency and avoid excess nutrients.

[0147] More specifically, the processor 350 may include one or more of a central processing unit (CPU), an application-specific integrated circuit (ASIC), a chipset, and a logic circuit, but is not limited thereto.

[0148] The memory 355 may include one or more of a read-only memory (ROM), a random access memory (RAM), a flash memory, and a memory card, but is not limited thereto.

[0149] The input / output device 365 may include one or more input devices such as buttons, switches, keyboards, mice, joysticks, and touch screens, and output devices such as liquid crystal displays (LCDs), light emitting diodes (LEDs), organic LEDs (OLEDs), active matrix OLEDs (AMOLEDs), printers, and plotters, but is not limited thereto.

[0150] If the embodiments included in this specification are implemented in software, the methods can be implemented by multiple modules (processes, functions, etc.) each performing the functions described. Each module can reside in the memory 355 and be executed by the processor 350. The memory 355 can exist inside or outside the processor 350 and can be connected to the processor 350 by various well-known means.

[0151] Fig. 9 The elements shown can be implemented by various tools (such as hardware, firmware, software or a combination thereof). If implemented by hardware, an embodiment of the present invention can 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.

[0152] Furthermore, if implemented in firmware or software, an embodiment of the present invention can be implemented in the form of modules, steps, functions, etc. that perform the multiple functions or actions described above and stored in a readable storage medium through various computer tools. Here, the storage medium may contain program commands, data files, data structures, etc. alone or in combination.

[0153] The program commands recorded in the storage medium may be specially designed for the present invention or may be known and usable by ordinary technicians in the computer software industry. For example, the storage medium includes magnetic media such as hard disks, floppy disks and tapes; optical recording media such as compact disk read only memory (CD-ROM) and digital video disks (DVD); magneto-optical media such as floptical disks; and special hardware devices such as read-only memory (ROM), random access memory (RAM), and flash memory that store and execute program commands.

[0154] Program commands may include machine language codes produced by a compiler, or high-level language codes executed by a computer using tools such as an interpreter. The hardware device can be operated by one or more software to perform the actions of the present invention, and vice versa.

[0155] The operation of the recipe plan recommendation server 300 described above will be described in detail below.

[0156] Fig.10 It is a flowchart for illustrating a method for recommending a recipe plan according to an embodiment of the present invention.

[0157] like Fig.10 As shown, the recipe plan recommendation server 300 of one embodiment of the present invention can obtain the electrocardiogram (ECG) of the user as the object. Fig.10 (ECG) signal (step S100).

[0158] Specifically, the recipe plan recommendation server 300 can directly receive the electrocardiograph from the electrocardiograph 100. Fig.10 (ECG) signals or can directly receive signal input from a user or other person.

[0159] Next, the recipe plan recommendation server 300 can calculate the ECG time according to the sequence of time measured by the user. Fig.10 The (ECG) signal is segmented into a plurality of segment signals S1, S2, S3, ... (step S200).

[0160] According to one embodiment, the recipe plan recommendation server 300 Fig.10 The ECG signal is cut individually (discretely) every time it reaches a certain size without repeated areas, so that the ECG signal can be Fig.10 The ECG signal is divided into a plurality of segmented signals S1, S2, S3, ... According to another embodiment, the recipe plan recommendation server 300 uses the ECG signal to Fig.10 (ECG) signal as the object, while sliding the grid (window) along the time axis, cutting it can Fig.10 The (ECG) signal is segmented into a plurality of segment signals S1, S2, S3, ... having sizes corresponding to the grids.

[0161] On the other hand, the recipe plan recommendation server 300 can also remove noise contained in the divided plurality of segment signals S1, S2, S3, ...

[0162] Next, the recipe plan recommendation server 300 can utilize artificial intelligence (AI) to analyze the plurality of segmented signals S1, S2, S3, ... to obtain a plurality of probability values ​​P1, P2, P3, ..., Pn (step S300).

[0163] Specifically, the recipe plan recommendation server 300 can extract one or more features according to a preset feature extraction benchmark based on the displacement of the P wave, PQ wave, QRS wave, ST wave and T wave contained in each of the divided multiple segmented signals S1, S2, S3, .... The recipe plan recommendation server 300 can uniquely encode the one or more extracted features to generate a vector of electrocardiogram (ECG) signal features. The recipe plan recommendation server 300 can embed the generated vector of electrocardiogram (ECG) signal features to generate a reduced-dimensional dense vector. The recipe plan recommendation server 300 can input the generated dense vector into the generalized matrix decomposition layer and the multi-layer perceptron layer of each of the first artificial neural network (ANN) to the third artificial neural network (ANN). Moreover, the recipe plan recommendation server 300 can obtain multiple probability values ​​P1, P2, P3, ..., Pn corresponding to the multiple segmented signals S1, S2, S3, ... from the first artificial neural network (ANN) to the third artificial neural network (ANN).

[0164] Next, the recipe plan recommendation server 300 can determine whether each nutrient component is in short supply, normal, or in excess based on the obtained plurality of probability values ​​P1, P2, P3, . . . , Pn (step S400).

[0165] Specifically, the recipe plan recommendation server 300 can judge whether electrolytes are insufficient, normal, or excessive based on multiple probability values ​​taken from the first artificial neural network (ANN). The recipe plan recommendation server 300 can judge whether proteins are insufficient, normal, or excessive based on multiple probability values ​​taken from the second artificial neural network (ANN). Moreover, the recipe plan recommendation server 300 can judge whether water is insufficient, normal, or excessive based on multiple probability values ​​taken from the third artificial neural network (ANN).

[0166] To this end, the recipe plan recommendation server 300 can remove abnormal values ​​included in the plurality of probability values ​​P1, P2, P3, ..., Pn taken from each artificial neural network (ANN) based on a preset effective range. The recipe plan recommendation server 300 can use a soft voting result based on an average value Pa of the plurality of probability values ​​P1, P2, ..., Pm with abnormal values ​​removed or a hard voting result based on a majority decision of the plurality of probability values ​​P1, P2, ..., Pm with abnormal values ​​removed to determine whether a nutrient is insufficient, normal, or excessive.

[0167] Next, the recipe plan recommendation server 300 can set a recipe plan including one or more diets that can supplement the deficient nutrients and avoid the excess nutrients (step S500).

[0168] Specifically, the recipe plan recommendation server 300 can identify two or more diets that do not contain excess nutrients or only contain less than a preset threshold content and contain the missing nutrients from a preset diet dictionary based on the information about the identified missing nutrients, excess nutrients, and normal nutrients. The recipe plan recommendation server 300 can identify the preset personal weight corresponding to the user and the corresponding ECG weight. Fig.10 The time weight of the time point at which the (ECG) signal is measured. The recipe plan recommendation server 300 can reflect the personal weight and the time weight in the two or more identified diets and select a diet. The recipe plan recommendation server 300 can identify one or more recipe plan templates including the selected diet and set a recipe plan using the identified one or more recipe plan templates.

[0169] Finally, the recipe schedule recommendation server 300 can transmit information about the set recipe schedule to the user device 200 (UE) preset corresponding to the user (step S600).

[0170] As mentioned above, this specification and the accompanying drawings disclose the preferred embodiments of the present invention. However, in addition to the embodiments disclosed herein, multiple variations based on the technical spirit of the present invention can also be implemented, which is self-evident to a person of ordinary skill in the technical field to which the present invention belongs. Moreover, although specific terms are used in this specification and the accompanying drawings, these terms are used in a general sense only to facilitate the description of the technical content of the present invention and to help understand the present invention, and are not intended to limit the scope of the present invention. Therefore, the detailed descriptions above cannot be interpreted as restrictive in all aspects and can only be regarded as illustrative. The scope of the present invention should be selected based on a reasonable interpretation of the scope of the invention claims, and all changes within the equivalent range of the present invention are included in the scope of the present invention.

Claims

1. A recipe plan recommendation method, characterized in that: The following steps are involved: an electrocardiogram signal acquisition step, acquiring an electrocardiogram signal measured with the user as an object; an identification step, analyzing the electrocardiogram signal using pre-learned artificial intelligence to identify at least one nutrient deficiency and nutrient excess corresponding to the health status of the user; and A recipe plan setting step is to set a recipe plan that includes more than one diet that can supplement the identified deficient nutrients and avoid the excess nutrients.

2. The method for recommending a recipe plan according to claim 1, characterized in that: The identification step comprises the following steps: Segmenting the electrocardiogram signal into a plurality of segmented signals according to the sequence of time; Inputting the divided multiple segmented signals into the artificial intelligence respectively, and obtaining multiple probability values ​​from the artificial intelligence; as well as Based on the obtained multiple probability values, it is judged whether the nutrients are short, normal or excessive.

3. The method for recommending a recipe plan according to claim 2, characterized in that: In the step of dividing into a plurality of segment signals, a grid having a predetermined size is moved along a time axis and the electrocardiogram signal is divided into a plurality of segment signals having a size corresponding to the grid.

4. The method for recommending a recipe plan according to claim 2, characterized in that: The artificial intelligence includes a first artificial neural network about electrolytes, a second artificial neural network about proteins, and a third artificial neural network about water. In the step of making the judgment, whether electrolytes are in short supply, normal or in excess is judged based on multiple probability values ​​taken from the first artificial neural network, whether proteins are in short supply, normal or in excess is judged based on multiple probability values ​​taken from the second artificial neural network, and whether water is in short supply, normal or in excess is judged based on multiple probability values ​​taken from the third artificial neural network.

5. The method for recommending a recipe plan according to claim 4, characterized in that: The hidden layers of the first artificial neural network to the third artificial neural network include a generalized matrix decomposition for learning the linear relationship between the characteristics of the electrocardiogram signal and the nutritional content, and a multi-layer perceptron for learning the nonlinear relationship between the characteristics of the electrocardiogram signal and the nutritional content.

6. The method for recommending a recipe plan according to claim 2, characterized in that: In the step of making the judgment, abnormal values ​​contained in the multiple probability values ​​obtained are eliminated based on a preset effective range, and a soft voting result based on the average value of the multiple probability values ​​from which the abnormal values ​​have been eliminated, or a hard voting result based on the majority decision of the multiple probability values ​​from which the abnormal values ​​have been eliminated, is used to judge whether the nutrient is short, normal or in excess.

7. The method for recommending a recipe plan according to claim 1, characterized in that: After the recipe plan setting step, the method further includes the following steps: Information about the set recipe schedule is transmitted to a preset user device corresponding to the user.

8. The method for recommending a recipe plan according to claim 7, characterized in that: The recipe plan setting step includes the following steps: Identifying two or more diets from a preset diet dictionary that do not contain the excess nutrients or only contain the identified deficient nutrients at a content below a preset threshold; identifying a pre-set personal weight corresponding to the user and a time weight corresponding to a measurement time point of the electrocardiogram signal; and The personal weight and the time weight are reflected in the two or more identified diets, and a diet is selected.

9. The method for recommending a recipe plan according to claim 1, characterized in that: In the identification step, if the first moment and the second moment have a difference within a preset minimum bioconversion time, and if the excess nutrients identified based on the electrocardiogram signal measured with the user as the object at the first moment and the shortage nutrients identified based on the electrocardiogram signal measured with the user as the object at the second moment are the same as each other, it is determined that the user needs to be warned.

10. A computer program stored in a storage medium and combined with a computing device including a memory, a transceiver, an input-output device and a processor for processing commands resident in the memory for executing the following steps: The processor acquires an electrocardiogram signal measuring the user as an object through the transceiver or the input / output device; The processor analyzes the electrocardiogram signal using pre-learned artificial intelligence to identify at least one nutrient deficiency and nutrient excess corresponding to the health status of the user; and The processor sets a recipe plan that includes one or more diets that replenish the identified deficient nutrients and avoid the excess nutrients.

Images