System and method for analyzing health condition of user based on multimodal biological signal

Abstract

The present invention relates to a system and method for analyzing a user's health status based on multimodal biosignals. A health status analysis system according to one embodiment comprises a patch-type biosignal measuring device attached to a user to collect at least two of multiple biosignals among electrocardiogram, brainwave, motion signal, photoplethysmography signal, and body temperature, and an analysis server that analyzes the user's health status by performing at least one analysis among artificial intelligence analysis and statistical analysis using multiple biosignals, wherein the analysis server extracts morphological information as feature information through conversion of multiple biosignals into time-standardized images based on a stored artificial intelligence machine learning algorithm, and analyzes the health status based on the extracted feature information.

Description

Technology Field

[0001] The present invention relates to a health status analysis system and a method thereof, and more specifically, to a technical concept that utilizes multimodal sensor-based IoT technology to transmit a user's biosignals from an analysis server and provides the user's health status analyzed by artificial intelligence through the analysis server. Background Technology

[0002] According to demographic changes, the elderly population aged 65 or older is projected to reach 10,508,000 in 2025, accounting for 20% of the total population of 56.21 million. With the number of patients in intensive care units increasing exponentially, there is a continuous upward trend in the capacity of nursing hospitals and facilities.

[0003] Cardiovascular and cerebrovascular diseases account for 24.3% of all causes of death and have a direct correlation with population aging.

[0004] According to information released by hospitals, the in-hospital mortality rate for patients in intensive care units is 18%, which is about twice that of developed countries, and in 2018, one-third of all deaths in the country occurred in nursing hospitals and nursing facilities for the elderly.

[0005] Intensive care units that have obtained a grade of 2 or higher in the appropriateness assessment account for only about one-third of the total, and there is a need for qualitative improvement in the level of medical services. As of 2016, the proportion of medical expenses for hospitalization of the elderly is about 47.6%, meaning that about 14% of the elderly population uses about half of the total medical expenses.

[0006] The low reimbursement rates for intensive care units contribute to deficits in tertiary and general hospitals, and while nursing hospitals contribute the most to job creation among medical institutions, their inpatient revenue per 100 beds is at the lowest level.

[0007] Representative technologies for solving the aforementioned problems include monitoring technologies using ECG (Electrocardiogram), PPG (Photoplethysmogram), Actigraph, and 3D depth cameras in terms of patient monitoring technology, and through these, equipment for monitoring the patient's condition in real time is continuously being developed.

[0008] However, existing patient monitoring equipment faces significant difficulties in accurately determining a patient's condition because it collects and utilizes only a single biosignal. Prior art literature

[0009] Korean Published Patent No. 10-2020-0062685, "Method for collecting patient monitoring information using Internet of Things service" Korean Published Patent No. 10-2020-0002241, "Method, device, and computer program for monitoring biosignals" Korean Published Patent No. 10-2020-0002201, "Smart healthcare system through artificial intelligence" Korean Published Patent No. 10-2017-0129689, "Mobile wearable monitoring system" The problem to be solved

[0010] The present invention aims to provide a health status analysis system and a method capable of simultaneously collecting multiple biosignals of a user using a patch-type biosignal monitoring device and analyzing the user's health status thereby.

[0011] In addition, the present invention aims to provide a health status analysis system and a method capable of more accurately and comprehensively determining a user's health status through artificial intelligence and statistical-based analysis using collected multiple biosignals. means of solving the problem

[0012] A health status analysis system according to one embodiment of the present invention may include a patch-type biosignal measuring device attached to a user to collect at least two of multiple biosignals among electrocardiogram, brainwave, motion signal, photoplethysmography signal, and body temperature, and an analysis server that analyzes the user's health status by performing at least one analysis among artificial intelligence analysis and statistical analysis using multiple biosignals, wherein the analysis server extracts morphological information as feature information through conversion of multiple biosignals into time-standardized images based on a stored artificial intelligence machine learning algorithm, and analyzes the health status based on the extracted feature information.

[0013] According to one aspect, the biosignal measuring device may include a first sensor unit configured with at least two sensors and configured with a wired connection to the patch body and configured with a single sensor.

[0014] According to one aspect, the first sensor unit may include an electrocardiogram sensor, a motion detection sensor, and a photoplethysmography detection sensor.

[0015] According to one side, the second sensor unit may include at least one of a body temperature sensing sensor and a brainwave sensing sensor.

[0016] According to one side, the analysis server analyzes a user’s disease corresponding to at least one of the user’s physical health and mental health and whether the user’s disease is in a critical state based on extracted feature information, and can feed back information on the user’s disease, whether the critical state is in a critical state, and fluctuations in multiple biosignals according to the critical state to a pre-configured user terminal device.

[0017] According to one aspect, the user terminal device may provide at least one service among a user's nursing management service, a disease data management service, a disease data visualization service, a disease data statistics service, and an emergency push notification service based on at least one information among the user's disease, whether it is in an emergency state, and information on fluctuations in multiple biosignals.

[0018] According to one side, the analysis server can analyze the user's mental health based on the output data of a stress classification algorithm that takes extracted feature information as input.

[0019] According to one side, a biosignal measuring device can generate a normalized signal based on the time domain of multiple biosignals, convert the generated normalized signal into a time-standardized image, generate a compressed signal corresponding to the multiple biosignals by applying weights based on an artificial intelligence machine learning algorithm to the converted image, and transmit the compressed signal corresponding to the multiple biosignals to an analysis server.

[0020] According to one side, the analysis server can generate a restored signal from a compressed signal corresponding to multiple biosignals using applied weights, and machine learn the weights so that the difference between the generated normalized signal and the generated restored signal corresponds to a pre-set threshold range, thereby extracting morphological information of multiple biosignals as feature information.

[0021] A health status analysis method according to one embodiment of the present invention may include the steps of: collecting at least two multiple biosignals among electrocardiogram, brainwave, motion signal, photoplethysmography signal, and body temperature from a patch-type biosignal measuring device attached to a user; and analyzing the user's health status by performing at least one analysis among artificial intelligence analysis and statistical analysis using multiple biosignals at an analysis server, wherein morphological information is extracted as feature information through conversion of multiple biosignals into time-standardized images based on a previously stored artificial intelligence machine learning algorithm, and the health status is analyzed based on the extracted feature information.

[0022] According to one aspect, the biosignal measuring device may include a first sensor unit configured with at least two sensors and configured with a wired connection to the patch body and configured with a single sensor. Effects of the invention

[0023] According to one embodiment, the present invention can simultaneously collect multiple biosignals of a user using a patch-type biosignal monitoring device and analyze the user's health status thereby.

[0024] According to one embodiment, the present invention can more accurately and comprehensively determine a user's health condition through artificial intelligence and statistics-based analysis using collected multiple biosignals. Brief explanation of the drawing

[0025] FIG. 1 is a drawing illustrating a health condition analysis system according to one embodiment. FIGS. 2a and 2b are drawings illustrating the structural features of a biosignal measuring device according to one embodiment. FIG. 3 is a drawing illustrating the components of a biosignal measuring device according to one embodiment. FIG. 4 is a diagram illustrating an artificial intelligence encoder of a biosignal measuring device according to one embodiment. FIG. 5 is a diagram illustrating the components of an analysis server according to one embodiment. FIG. 6 is a diagram illustrating a health condition analysis method according to one embodiment. Specific details for implementing the invention

[0026] Specific structural or functional descriptions of embodiments according to the concept of the present invention disclosed herein are provided merely for the purpose of explaining embodiments according to the concept of the present invention, and embodiments according to the concept of the present invention may be implemented in various forms and are not limited to the embodiments described herein.

[0027] Embodiments according to the concept of the present invention may be subject to various modifications and may take various forms; therefore, embodiments are illustrated in the drawings and described in detail in this specification. However, this is not intended to limit the embodiments according to the concept of the present invention to specific disclosed forms, and includes modifications, equivalents, or substitutions that fall within the spirit and scope of the present invention.

[0028] Terms such as "first" or "second" may be used to describe various components, but the components should not be limited by the terms. For the sole purpose of distinguishing one component from another, for example, without departing from the scope of rights according to the concept of the present invention, the first component may be named the second component, and similarly, the second component may be named the first component.

[0029] When it is stated that one component is "connected" or "connected" to another component, it should be understood that while it may be directly connected or connected to that other component, there may also be other components in between. Conversely, when it is stated that one component is "directly connected" or "directly connected" to another component, it should be understood that there are no other components in between. Expressions describing the relationships between components, such as "between," "exactly between," or "directly adjacent to," should be interpreted in the same way.

[0030] The terms used herein are used merely to describe specific embodiments and are not intended to limit the invention. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this specification, terms such as “comprising” or “having” are intended to specify the existence of the described features, numbers, steps, actions, components, parts, or combinations thereof, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0031] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art to which the present invention pertains. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined in this specification.

[0033] Hereinafter, embodiments will be described in detail with reference to the attached drawings. However, the scope of the patent application is not limited or restricted by these embodiments. Identical reference numerals in each drawing indicate identical components.

[0035] FIG. 1 is a drawing illustrating a health condition analysis system according to one embodiment.

[0036] Referring to FIG. 1, a health status analysis system (100) according to one embodiment can simultaneously collect multiple biosignals of a user using a patch-type biosignal monitoring device and analyze the user's health status through this.

[0037] In addition, the health status analysis system (100) can more accurately determine the user's health status from various angles through artificial intelligence and statistical-based analysis using multiple collected biosignals.

[0038] To this end, the health status analysis system (100) may include a biosignal measuring device (110), a gateway (120), and an analysis server (130).

[0039] A biosignal measuring device (110) according to one embodiment is attached to a user in the form of a patch and can collect at least two or more multiple biosignals among an electrocardiogram, brain waves, motion signals, photoplethysmography signals and body temperature.

[0040] For example, the biosignal measuring device (110) is a patch-type device attached to the chest area of ​​a user and can collect multiple biosignals, including a first biosignal (e.g., electrocardiogram, motion signal, and photoplethysmography signal) which is collected from the chest area of ​​the user and used as a main indicator to diagnose the user's health condition such as cardiovascular disease, and a second biosignal (e.g., brainwave and body temperature) which is detected in an area other than the user's chest and used as an auxiliary indicator to diagnose the user's health condition.

[0041] In other words, the present invention utilizes multiple biosignals (a first biosignal and a second biosignal) rather than a single biosignal to analyze the user's health condition more accurately and from multiple angles.

[0042] The biosignal measuring device (110) can compress and encrypt the collected multiple biosignals using an artificial intelligence encoder and transmit them to a gateway (120) using short-range wireless communication (e.g., low-power Bluetooth).

[0043] According to one aspect, the biosignal measuring device (110) can generate a normalized signal based on the time domain of multiple biosignals, convert the generated normalized signal into a time-standardized image, apply weights based on an artificial intelligence machine learning algorithm to the converted image to generate a compressed signal corresponding to multiple biosignals, and transmit the compressed signal corresponding to multiple biosignals to the gateway (120).

[0044] For example, as a preprocessing step for applying measured biosignals to a deep learning model, an artificial intelligence encoder can divide and normalize biosignals into signals of equal length using sliding window technology. For instance, the window length can be 2 seconds and the update period can be 1 second.

[0045] Here, the sliding window can analyze signals with relatively short measurement times, and through a normalization process, various signal amplitudes and offsets can be set so that the deep learning model is not affected. The artificial intelligence encoder will be further explained later using the example Fig. 4.

[0046] The biosignal measuring device (110) can amplify multiple biosignals and remove noise before generating a compressed signal corresponding to multiple biosignals, and may further include a noise removal filter for this purpose. For example, the biosignal measuring device (110) can remove noise by exception processing signals that exceed a predetermined threshold based on the magnitude of the unit signal.

[0047] A gateway (120) according to one embodiment can transmit a compressed signal corresponding to multiple transmitted biosignals to an analysis server (130).

[0048] The gateway (120) is linked with the biosignal measuring device (110) and the analysis server (130) to support battery low event transmission, process monitoring, and automatic failover.

[0049] Additionally, the gateway (120) supports automatic node search, registration, and connection of the biosignal measuring device (110), and can support a low-power Bluetooth communication interface and a battery level check of the biosignal measuring device (110).

[0050] An analysis server (130) according to one embodiment analyzes the user's health condition by performing at least one of an artificial intelligence analysis using multiple biosignals and a statistical analysis, and can extract morphological information as feature information by converting multiple biosignals into time-standardized images based on a stored artificial intelligence machine learning algorithm, and can analyze the health condition based on the extracted feature information.

[0051] According to one side, the analysis server (130) receives a compressed signal corresponding to multiple biosignals from the gateway (120), decodes the compressed signal corresponding to multiple biosignals using an artificial intelligence decoder to extract multiple biosignals, and performs machine learning and analysis processes on the extracted multiple biosignals using an artificial intelligence algorithm, thereby optimizing the detection performance of heart conditions using a public biosignal data set measured in various environments, or can perform model learning and performance evaluation on multiple biosignals measured from the biosignal measurement device (110) by utilizing the structure of an artificial intelligence for removing biosignal artifacts developed for multimodal biosignals.

[0052] In addition, the analysis server (130) can utilize morphological information of multiple biosignals as feature information to reflect various patterns of multiple biosignals in a risk assessment model based on an artificial intelligence algorithm.

[0053] Specifically, the analysis server (130) receives a compressed signal corresponding to multiple biosignals from the gateway (120), generates a restored signal from the compressed signal corresponding to multiple biosignals using applied weights, and machine learns the weights so that the difference between the generated normalized signal and the generated restored signal corresponds to a pre-set threshold range, thereby extracting morphological information of multiple biosignals as feature information.

[0054] For example, the analysis server (130) can determine a classification model to distinguish major cardiovascular diseases such as myocardial infarction and coronary artery disease by using machine learning models such as CNN (Convolutional Neural Network) and DBN (Deep Belief Network), or by using various machine learning techniques such as Gradient Boost and XGBooST.

[0055] In addition, the analysis server (130) can evaluate the utility of the classification model by calculating the accuracy of the classification model through result comparison using machine learning techniques such as SVM (support vector machine) and RF (random forest) based on the public biosignal dataset.

[0056] According to one side, the analysis server (130) can determine the signal quality of the restored multiple biosignals and determine whether to extract feature information in response to the result of determining the signal quality.

[0057] For example, the analysis server (130) may further include an artificial intelligence algorithm for quality estimation and can use it to determine the signal quality of the restored multiple biosignals.

[0058] Specifically, the analysis server (130) can estimate that the signal quality is better when the reference signal provided and the multiple biosignals are similar based on an artificial intelligence algorithm for quality estimation generated based on the correlation between the characteristics of the biosignal and the signal quality to be estimated.

[0059] Here, the artificial intelligence algorithm for quality estimation can determine the signal quality level as 1 to 10, and can be defined so that the closer the signal quality level is to 1, the higher the signal quality.

[0060] That is, the analysis server (130) can determine that the signal quality is good if the signal quality level of the multiple biosignals estimated based on an artificial intelligence algorithm for quality estimation is above a preset threshold, and extract feature information from the restored multiple biosignals, and if the signal quality level of the multiple biosignals is below the preset threshold, determine that the reliability of the signal according to the signal quality is low, and not use the restored multiple biosignals, and perform the signal quality determination and feature extraction process for the multiple biosignals received at the next time.

[0061] According to one aspect, the analysis server (130) can analyze the user’s disease corresponding to at least one of the user’s physical health and mental health and whether the user’s disease is in a critical state based on extracted feature information, and feed back the user’s disease, whether the critical state is in a critical state and information on the fluctuation of multiple biosignals according to the critical state to a pre-configured user terminal device (140).

[0062] For example, the analysis server (130) can determine cardiovascular diseases related to the heart condition of a supported person that has been learned and analyzed using an artificial intelligence algorithm, and provide the determined cardiovascular diseases and analysis results to a user terminal device (140) to assist medical staff in diagnosing cardiovascular diseases or to help easily recognize changes in the patient's prognosis.

[0063] Here, the user terminal device (140) can provide at least one service among the user's nursing management service, disease data management service, disease data visualization, disease data statistics service, and emergency push notification service based on at least one information among the user's disease, whether it is in an emergency state, and information on fluctuations in multiple biosignals.

[0064] According to one aspect, the analysis server (130) may collect user activity information corresponding to feature information extracted from the user terminal device (140) and may analyze the user's health status by considering the collected activity information. For example, the activity information may include user information regarding 'sleep', 'exercise', and 'normal activity', but is not limited thereto.

[0065] For example, the user terminal device (140) may be linked with the biosignal measuring device (110) via wireless communication to generate user activity information corresponding to multiple biosignals collected through the biosignal measuring device (110), and to this end, the user terminal device (140) may include at least one sensing means among a microphone, a motion detection sensor, and an image sensor.

[0066] Additionally, the analysis server (130) can standardize multiple biosignals of the user collected over a predetermined period of time to set reference information corresponding to each of the multiple activity information, wherein the reference information corresponding to each of the multiple activity information may refer to a preset normal range value (minimum value, maximum value, change value, etc.) of each of the user's multiple biosignals that occurs in response to the multiple activity information. For example, when the user is in a 'sleep' state, the analysis server (130) can standardize the user's sleep pattern (sleep time and wake time), breathing volume, movement volume, and heart rate information that occur during sleep to set reference information.

[0067] In other words, the analysis server (130) can determine the user's health status through an analysis process based on the user's first reference information and feature information set in advance in response to the sleep state when the user's activity information corresponding to the feature information is in a 'sleep' state, and can determine the user's health status through an analysis process based on the user's second reference information and feature information set in advance in response to the exercise state when the user's activity information is in a 'sleep' state.

[0068] Meanwhile, the analysis server (140) can analyze the user's mental health based on the output data of a stress classification algorithm that takes extracted feature information as input.

[0069] For example, the analysis server (140) may include a clustering algorithm that takes feature information corresponding to a heart rate signal among the extracted feature information as input and outputs a stress level as a result, wherein the feature information corresponding to the heart rate signal may include a heart rate signal in a resting state and a heart rate signal exposed to noise.

[0070] The analysis server (140) can classify stress by clustering the heart rate signal in a resting state and the heart rate signal exposed to noise using a k-means clustering algorithm.

[0071] Specifically, the analysis server (140) can calculate the RS amplitude difference, which is the difference between the R peak value and the S peak value, and output the heart rate signal in a resting state and the heart rate signal exposed to noise as stress levels relative to the RS amplitude difference according to the pre-set stress signal classification criteria.

[0072] Here, the classification criteria for the preset stress signal can distinguish it as a heart rate signal exposed to noise if the RS amplitude difference is greater than or equal to the preset amplitude threshold, and as a heart rate signal in a resting state if it is less than the amplitude threshold.

[0073] In other words, the analysis server (140) can support relatively easy identification of the stress index by K-means clustering each of the resting heart rate signal and the noise-exposed heart rate signal based on the feature points of each of the resting heart rate signal and the noise-exposed heart rate signal.

[0075] FIGS. 2a and 2b are drawings illustrating the structural features of a biosignal measuring device according to one embodiment.

[0076] Referring to FIGS. 2a and 2b, the biosignal measuring device (200) is a patch-type device and may include a first sensor unit (210) provided in a patch body and composed of at least two sensors, and a second sensor unit (220) connected to the patch body by a wire and composed of at least one sensor.

[0077] For example, the first sensor unit (210) may include an electrocardiogram sensor (211) and a photoplethysmography sensor (212), and may further include a motion detection sensor.

[0078] Additionally, the second sensor unit (220) may include a body temperature sensing sensor (221), but is not limited thereto and may further include a brainwave sensing sensor.

[0079] In other words, the biosignal measuring device (200) can simultaneously collect multiple biosignals through a first sensor unit (210) that collects a first biosignal from the chest area of ​​the user and a second sensor unit (220) that collects a second biosignal collected from an area other than the chest.

[0081] FIG. 3 is a drawing illustrating the components of a biosignal measuring device according to one embodiment.

[0082] Referring to FIG. 3, the biosignal measuring device (300) may include a biosignal monitoring unit (310) and a control unit (320).

[0083] The biosignal monitoring unit (310) is a patch-type device attached to the upper part of the user's heart or near the heart, and can collect multiple biosignals of the user and generate a compressed signal corresponding to the multiple biosignals through an artificial intelligence encoder using monitoring information containing the collected multiple biosignals.

[0084] Specifically, the biosignal monitoring unit (310) can generate a normalized signal based on the time domain of multiple biosignals through an artificial intelligence encoder, convert the generated normalized signal into a time-standardized image, and apply weights based on an artificial intelligence machine learning algorithm to the converted image to generate a compressed signal corresponding to multiple biosignals.

[0085] The control unit (320) can transmit a compressed signal corresponding to multiple biosignals to a server via a nearby gateway using wireless communication.

[0086] Additionally, the control unit (320) can simulate data traffic generated when measuring at least one of multiple biosignals and determine the operating state of the biosignal measuring device based on the simulation.

[0087] In addition, the control unit (320) can periodically check the battery status to determine a low battery status and provide alarm information.

[0089] FIG. 4 is a diagram illustrating an artificial intelligence encoder of a biosignal measuring device according to one embodiment.

[0090] Referring to FIG. 4, the artificial intelligence encoder (400) is composed of an encoding block (410) and a deep learning neural network (420), and the encoding block (410) may be composed of a deep learning neural network layer, a pooling layer, and a reshape layer.

[0091] The artificial intelligence encoder (400) can compress signals for morphological information through conversion of multiple biosignals into time-standardized images and output the compressed signals as monitoring information.

[0092] The artificial intelligence encoder (400) may be composed of a convolutional neural network layer and a bidirectional long-term short-term memory layer, and for example, the artificial intelligence encoder (400) may output signal information compressed by about 64 times.

[0093] For example, morphological information obtained through transformation into a time-standardized image can be generated based on signals that have been divided and normalized into signals of equal length using sliding window technology.

[0095] FIG. 5 is a diagram illustrating the components of an analysis server according to one embodiment.

[0096] Referring to FIG. 5, the analysis server (500) may include a monitoring information collection unit (510), a signal extraction unit (520), an artificial intelligence processing unit (530), and a control unit (540).

[0097] The monitoring information collection unit (510) can collect compressed signals corresponding to multiple biosignals of the user.

[0098] For example, a compressed signal corresponding to multiple biosignals may be a biosignal transmitted through a gateway by a biosignal measuring device according to one embodiment.

[0099] In addition, the monitoring information collection unit (510) can store the collected compressed signals in a separate database.

[0100] The signal extraction unit (520) can extract multiple biosignals for analyzing the user's heart condition from the compressed signal.

[0101] For example, the signal extraction unit (520) can convert the encrypted and compressed signal into multiple biosignals using an artificial intelligence decoder.

[0102] The artificial intelligence processing unit (530) can extract the time domain and frequency domain of multiple biosignals extracted based on a previously stored artificial intelligence machine learning algorithm as feature information, determine multiple heart abnormality type models using the extracted feature information, calculate the distinction accuracy for the determined multiple heart abnormality type models, and determine the cardiovascular disease of the supported subject using the determined heart abnormality type models and the disclosed cardiovascular disease data based on the calculated accuracy.

[0103] Specifically, the artificial intelligence processing unit (530) divides the frequency domain into specific time intervals based on the time domain, normalizes the divided frequency domain into binary numbers to generate a normalized signal, converts the generated normalized signal into an image, generates a compressed signal by applying weights based on an artificial intelligence machine learning algorithm stored in the converted image, generates a restored signal from the compressed signal using the applied weights, and machine learns the weights so that the difference between the generated normalized signal and the generated restored signal corresponds to a preset threshold range to extract feature information corresponding to the morphological features of the electrocardiogram signal.

[0104] Additionally, the artificial intelligence processing unit (530) can machine learn feature information to determine at least one of a plurality of heart abnormality type models, such as a tachycardia model, a bradycardia model, an atrial fibrillation model, a left bundle branch block model, a right bundle branch block model, an atrial premature contraction model, a ventricular premature contraction model, and a cardiac arrest model.

[0105] For example, the artificial intelligence processing unit (530) can calculate the distinction accuracy for the determined multiple heart abnormality type models

[0183] .

[0106] Specifically, the artificial intelligence processing unit (530) can calculate the distinction accuracy for multiple heart abnormality type models based on the following mathematical formula 1.

[0107] [Mathematical Formula 1]

[0108] Accuracy = (TP + TN) / (TP + TN + FP + FN)

[0109] In mathematical formula 1, TP can represent the case where an abnormal state is accurately distinguished as an abnormal state, TN can represent the case where a normal state is distinguished as a normal state, FP can represent the case where a normal state is distinguished as an abnormal state, and FN can represent the case where a normal state is distinguished as an abnormal state.

[0110] More specifically, the artificial intelligence processing unit (530) can classify the heart abnormality into a TP (True Positive) case in which the heart abnormality is classified as a heart abnormality, an FN (False Negative) case in which the heart abnormality is classified as normal, an FP (False Positive) case in which normal is classified as a heart abnormality, and a TN (True Negative) case in which normal is classified as normal, by using an open data set and at least one of a tachycardia model, a bradycardia model, an atrial fibrillation model, a left bundle branch block model, a right bundle branch block model, an atrial premature contraction model, and a ventricular premature contraction model and a normal heart state model.

[0111] Additionally, the artificial intelligence processing unit (530) can calculate the distinction accuracy for at least one of the tachycardia model, bradycardia model, atrial fibrillation model, left bundle branch block model, right bundle branch block model, atrial premature contraction model, and ventricular premature contraction model based on the ratio of the combination of the TP (True Positive) case value and the TN (True Negative) case value to the combination of the TP (True Positive) case value, the FN (False Negative) case value, the FP (False Positive) case value and the TN (True Negative) case value.

[0112] Additionally, the artificial intelligence processing unit (530) can classify the cardiac arrest interval into a TP (True Positive) case in which the cardiac arrest interval is classified as a cardiac arrest interval, an FN (False Negative) case in which the cardiac arrest interval is classified as a normal interval, an FP (False Positive) case in which the normal interval is classified as a cardiac arrest interval, and a TN (True Negative) case in which the normal interval is classified as a normal interval by using an open data set, a cardiac arrest model, and a normal heart state model.

[0113] Next, the artificial intelligence processing unit (530) can calculate the critical situation distinction accuracy based on the ratio of the combination of the TP (True Positive) case value and the TN (True Negative) case value to the combination of the TP (True Positive) case value, the FN (False Negative) case value, the FP (False Positive) case value and the TN (True Negative) case value.

[0114] Additionally, the artificial intelligence processing unit (530) classifies the artifact signal into a TP (True Positive) case in which the artifact signal is classified as an artifact signal, an FN (False Negative) case in which the artifact signal is classified as a normal signal, an FP (False Positive) case in which the normal signal is classified as an artifact signal, and a TN (True Negative) case in which the normal signal is classified as a normal signal, and calculates the artifact removal accuracy based on the ratio of the combination of the TP (True Positive) case value and the TN (True Negative) case value to the combination of the TP (True Positive) case value, the FN (False Negative) case value, the FP (False Positive) case value and the TN (True Negative) case value.

[0115] For example, the artificial intelligence processing unit (530) can automatically calculate the distinction accuracy by comparing an open data set and a heart abnormality type model based on artificial intelligence learning technology.

[0116] The artificial intelligence processing unit (530) includes signal quality management artificial intelligence, arrhythmia detection artificial intelligence, and cardiovascular disease evaluation artificial intelligence, and can manage the signal quality of biosignals transmitted from a biosignal measurement device and evaluate and detect arrhythmia and cardiovascular disease.

[0117] Meanwhile, the control unit (540) can control the provision of the cardiovascular disease determined by the artificial intelligence processing unit (530) to the user terminal device. Here, the device being controlled may correspond to a communication device.

[0118] For example, a user terminal device may receive analysis information related to cardiovascular disease and provide cardiovascular disease determination information and analysis results included in the received analysis information through a display.

[0119] After at least one of the electrocardiogram signal and the motion signal is measured, the control unit (540) calculates the number of measured electrocardiogram signals and the number of measured motion signals, and compares the calculated number of electrocardiogram signals and the calculated number of motion signals with a threshold value to check the data reception status of the electrocardiogram signal and the motion signal.

[0120] Additionally, if the reception status of the electrocardiogram signal and the motion signal is poor, the control unit (540) may request additional data or additional measurement from the biosignal measuring device.

[0121] Additionally, the control unit (540) can control the feedback of at least one of the information regarding cardiovascular disease and emergency status and the information on fluctuations in biosignals to the user terminal device.

[0123] FIG. 6 is a diagram illustrating a health condition analysis method according to one embodiment.

[0124] In other words, FIG. 6 is a drawing illustrating the operation method of a health condition analysis device according to an embodiment described through FIG. 1 to 5, and any descriptions described through FIG. 6 below that overlap with those described through FIG. 1 to 5 will be omitted.

[0125] Referring to FIG. 6, in step 610, a health status analysis method according to one embodiment can collect at least two or more multiple biosignals among electrocardiogram, brainwave, motion signal, photoplethysmography signal and body temperature from a patch-type biosignal measuring device attached to a user.

[0126] For example, the biosignal measuring device may include a first sensor unit configured with at least two sensors and provided in a patch body, and a second sensor unit configured with one sensor and connected to the patch body via a wire.

[0127] Here, the first sensor unit includes an electrocardiogram sensor, a motion detection sensor, and a photoplethysmography detection sensor, and the second sensor unit may include at least one of a body temperature detection sensor and a brainwave detection sensor.

[0128] Next, in step 620, the health status analysis method according to one embodiment can transmit multiple biosignals collected from a patch-type biosignal measuring device to an analysis server.

[0129] Specifically, in step 620, the health status analysis method according to one embodiment can generate a normalized signal based on the time domain of multiple biosignals in a biosignal measuring device, convert the generated normalized signal into a time-standardized image, generate a compressed signal corresponding to multiple biosignals by applying weights based on an artificial intelligence machine learning algorithm to the converted image, and transmit the compressed signal corresponding to multiple biosignals to an analysis server.

[0130] For example, in step 620, the health status analysis method according to one embodiment can transmit a compressed signal corresponding to multiple biosignals from a biosignal measuring device to a gateway, and transmit a compressed signal corresponding to multiple biosignals received from the gateway to an analysis server.

[0131] Next, in step 630, the health status analysis method according to one embodiment analyzes the user's health status by performing at least one analysis among artificial intelligence analysis using multiple biosignals and statistical-based analysis on an analysis server, wherein morphological information is extracted as feature information through conversion of multiple biosignals into time-standardized images based on a stored artificial intelligence machine learning algorithm, and the health status is analyzed based on the extracted feature information.

[0132] For example, in step 630, the health status analysis method according to one embodiment can generate a restored signal from a compressed signal corresponding to multiple biosignals using applied weights on an analysis server, and machine learn the weights so that the difference between the generated normalized signal and the generated restored signal corresponds to a pre-set threshold range, thereby extracting morphological information of multiple biosignals as feature information.

[0133] According to one aspect, in step 630, the health status analysis method according to one embodiment can analyze a user’s disease corresponding to at least one of the user’s physical health and mental health and whether the user’s disease is in a critical state based on extracted feature information at an analysis server, and feed back information on the user’s disease, whether the critical state is in a critical state, and changes in multiple biosignals according to the critical state to a pre-configured user terminal device.

[0134] In addition, at step 630, the health status analysis method according to one embodiment may provide at least one service among a user's nursing management service, a disease data management service, a disease data visualization service, a disease data statistics service, and an emergency push notification service based on at least one information among the user's disease, whether it is in an emergency state, and information on fluctuations in multiple biosignals at a user terminal device.

[0135] Meanwhile, in step 630, the health status analysis method according to one embodiment can analyze the user's mental health based on the output data of a stress classification algorithm that takes extracted feature information as input at an analysis server.

[0137] Ultimately, by utilizing the present invention, multiple biosignals of a user can be collected simultaneously using a patch-type biosignal monitoring device, and the user's health status can be analyzed thereby.

[0138] In addition, by using the present invention, the user's health condition can be determined more accurately and from multiple perspectives through artificial intelligence and statistics-based analysis using collected multiple biosignals.

[0140] Although the embodiments have been described above with reference to the limited drawings, those skilled in the art can make various modifications and variations from the description above. For example, suitable results can be achieved even if the described techniques are performed in a different order than described, and / or the components of the described device, structure, apparatus, circuit, etc. are combined or assembled in a form different from described, or replaced or substituted by other components or equivalents.

[0141] Therefore, other implementations, other embodiments, and equivalents to the claims also fall within the scope of the claims set forth below. Explanation of the symbols

[0142] 100: Health status analysis system 110: Biosignal measurement device 120: Gateway 130: Analysis Server

Claims

Claim 1 A patch-type biosignal measuring device attached to a user, which collects two or more multiple biosignals including an electrocardiogram and additionally including at least one of brain waves, motion signals, photoplethysmography signals, and body temperature; and an analysis server that analyzes the user's health status by performing at least one analysis among artificial intelligence analysis and statistical-based analysis using the multiple biosignals, wherein the analysis server extracts morphological information as feature information by converting the multiple biosignals into time-standardized images based on a stored artificial intelligence machine learning algorithm, and analyzes the health status based on the extracted feature information; wherein the analysis server analyzes the user's mental health based on output data of a stress classification algorithm that calculates a stress index by K-means clustering the resting heart rate signal and the noise-exposed heart rate signal, respectively, based on feature points of the resting heart rate signal and the noise-exposed heart rate signal among the extracted feature information, wherein the analysis server calculates the RS amplitude difference, which is the difference between the R peak value and the S peak value in the multiple biosignals, and, according to a pre-set stress signal classification criterion, the resting heart rate signal and the noise-exposed heart rate signal are classified into stress levels relative to the RS amplitude difference A health status analysis system that analyzes the mental health of the user by outputting and classifying the RS amplitude difference as a heart rate signal exposed to noise if it is greater than or equal to a preset amplitude threshold, and as a heart rate signal in a resting state if it is less than the amplitude threshold. Claim 2 In claim 1, the biosignal measuring device comprises a first sensor unit configured with at least two sensors and provided in a patch body, and a second sensor unit configured with one sensor and connected to the patch body via a wire. Claim 3 In paragraph 2, the first sensor unit comprises a health condition analysis system including an electrocardiogram sensor, a motion detection sensor, and a photoplethysmography detection sensor. Claim 4 In paragraph 2, the second sensor unit comprises at least one of a body temperature sensing sensor and a brainwave sensing sensor, in a health condition analysis system. Claim 5 A health status analysis system according to claim 1, wherein the analysis server analyzes the user’s disease corresponding to at least one of the user’s physical health and mental health and whether the user’s disease is in a critical state based on the extracted feature information, and feeds back the user’s disease, the critical state, and information on the variation of the multiple biosignals according to the critical state to a pre-configured user terminal device. Claim 6 In claim 5, the user terminal device is a health status analysis system that provides at least one service among a nursing management service, a disease data management service, a disease data visualization service, a disease data statistics service, and an emergency push notification service based on at least one of the information among the user's disease, whether the user is in an emergency state, and the fluctuation information of the multiple biosignals. Claim 7 delete Claim 8 A health status analysis system according to claim 1, wherein the biosignal measuring device generates a normalized signal based on the time domain of the multiple biosignals, converts the generated normalized signal into the time-standardized image, generates a compressed signal corresponding to the multiple biosignals by applying weights based on the artificial intelligence machine learning algorithm to the converted image, and transmits the compressed signal corresponding to the multiple biosignals to the analysis server. Claim 9 A health state analysis system according to claim 8, wherein the analysis server generates a restored signal from a compressed signal corresponding to the multiple biosignals using the applied weights, and machine learns the weights such that the difference between the generated normalized signal and the generated restored signal corresponds to a pre-set threshold range, thereby extracting the morphological information of the multiple biosignals as the feature information. Claim 10 A patch-type biosignal measuring device comprises the steps of: collecting two or more multiple biosignals attached to a user, including an electrocardiogram, and additionally including at least one of an electrocardiogram, brainwave, motion signal, photoplethysmography signal, and body temperature; and, at an analysis server, performing at least one analysis among artificial intelligence analysis and statistical-based analysis using the multiple biosignals to analyze the user's health status, wherein morphological information is extracted as feature information through conversion of the multiple biosignals into time-standardized images based on a stored artificial intelligence machine learning algorithm, and the health status is analyzed based on the extracted feature information; wherein the step of analyzing the health status comprises, wherein the analysis server analyzes the user's mental health based on output data of a stress classification algorithm that calculates a stress index by K-means clustering the resting heart rate signal and the noise-exposed heart rate signal, respectively, based on the feature points of the resting heart rate signal and the noise-exposed heart rate signal among the extracted feature information, wherein the RS amplitude difference, which is the difference between the R peak value and the S peak value in the multiple biosignals, is calculated, and according to a pre-set classification criterion for stress signals, the resting heart rate signal and A health state analysis method further comprising outputting a heart rate signal exposed to noise as a stress level relative to the RS amplitude difference, and analyzing the mental health of the user by classifying it as a heart rate signal exposed to noise if the RS amplitude difference is greater than or equal to a preset amplitude threshold, and as a heart rate signal in a resting state if it is less than the amplitude threshold. Claim 11 In claim 10, the above biosignal measuring device comprises a first sensor unit configured with at least two sensors and provided in a patch body, and a second sensor unit configured with one sensor and connected to the patch body via a wire.

Images