WO2024038930A1 - 복수의 심전도를 이용한 딥러닝기반 건강상태 예측 시스템 - Google Patents
복수의 심전도를 이용한 딥러닝기반 건강상태 예측 시스템 Download PDFInfo
- Publication number
- WO2024038930A1 WO2024038930A1 PCT/KR2022/012364 KR2022012364W WO2024038930A1 WO 2024038930 A1 WO2024038930 A1 WO 2024038930A1 KR 2022012364 W KR2022012364 W KR 2022012364W WO 2024038930 A1 WO2024038930 A1 WO 2024038930A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- ecg data
- ecg
- unit
- deep learning
- data
- Prior art date
Links
- 238000013135 deep learning Methods 0.000 title claims abstract description 27
- 230000036541 health Effects 0.000 title abstract description 6
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 claims abstract description 61
- 201000010099 disease Diseases 0.000 claims abstract description 60
- 238000005259 measurement Methods 0.000 claims abstract description 41
- 230000003862 health status Effects 0.000 claims description 36
- 238000012774 diagnostic algorithm Methods 0.000 claims description 5
- 239000000284 extract Substances 0.000 claims description 4
- 238000003745 diagnosis Methods 0.000 abstract description 10
- 238000004422 calculation algorithm Methods 0.000 abstract description 7
- 238000002565 electrocardiography Methods 0.000 description 132
- 229940079593 drug Drugs 0.000 description 5
- 239000003814 drug Substances 0.000 description 5
- 238000000034 method Methods 0.000 description 5
- 238000000691 measurement method Methods 0.000 description 4
- 208000010125 myocardial infarction Diseases 0.000 description 4
- 238000004393 prognosis Methods 0.000 description 4
- 230000006866 deterioration Effects 0.000 description 3
- 238000012351 Integrated analysis Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 230000006698 induction Effects 0.000 description 2
- 208000014674 injury Diseases 0.000 description 2
- 238000007477 logistic regression Methods 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000007637 random forest analysis Methods 0.000 description 2
- 230000035882 stress Effects 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 210000001519 tissue Anatomy 0.000 description 2
- 230000008733 trauma Effects 0.000 description 2
- 238000002255 vaccination Methods 0.000 description 2
- 206010008805 Chromosomal abnormalities Diseases 0.000 description 1
- 208000031404 Chromosome Aberrations Diseases 0.000 description 1
- 208000017667 Chronic Disease Diseases 0.000 description 1
- 206010010356 Congenital anomaly Diseases 0.000 description 1
- 206010027476 Metastases Diseases 0.000 description 1
- 208000008589 Obesity Diseases 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000003542 behavioural effect Effects 0.000 description 1
- 239000008280 blood Substances 0.000 description 1
- 210000004369 blood Anatomy 0.000 description 1
- 230000036772 blood pressure Effects 0.000 description 1
- 230000035606 childbirth Effects 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 235000005911 diet Nutrition 0.000 description 1
- 230000037213 diet Effects 0.000 description 1
- 210000002249 digestive system Anatomy 0.000 description 1
- 208000016097 disease of metabolism Diseases 0.000 description 1
- 208000035475 disorder Diseases 0.000 description 1
- 230000035622 drinking Effects 0.000 description 1
- 210000005069 ears Anatomy 0.000 description 1
- 230000002124 endocrine Effects 0.000 description 1
- 208000030172 endocrine system disease Diseases 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 210000004072 lung Anatomy 0.000 description 1
- 210000001595 mastoid Anatomy 0.000 description 1
- 230000003340 mental effect Effects 0.000 description 1
- 230000004060 metabolic process Effects 0.000 description 1
- 230000009401 metastasis Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 210000002346 musculoskeletal system Anatomy 0.000 description 1
- 230000001613 neoplastic effect Effects 0.000 description 1
- 210000000653 nervous system Anatomy 0.000 description 1
- 230000000474 nursing effect Effects 0.000 description 1
- 235000020824 obesity Nutrition 0.000 description 1
- 229910052760 oxygen Inorganic materials 0.000 description 1
- 239000001301 oxygen Substances 0.000 description 1
- 230000035935 pregnancy Effects 0.000 description 1
- 210000002345 respiratory system Anatomy 0.000 description 1
- 230000000391 smoking effect Effects 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
- 210000002229 urogenital system Anatomy 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
- A61B5/346—Analysis of electrocardiograms
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
- A61B5/346—Analysis of electrocardiograms
- A61B5/349—Detecting specific parameters of the electrocardiograph cycle
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
- A61B5/346—Analysis of electrocardiograms
- A61B5/349—Detecting specific parameters of the electrocardiograph cycle
- A61B5/352—Detecting R peaks, e.g. for synchronising diagnostic apparatus; Estimating R-R interval
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
- A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/09—Supervised learning
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/50—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for simulation or modelling of medical disorders
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/20—Ensemble learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
- G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0499—Feedforward networks
Definitions
- the present invention can improve the accuracy of measurement, diagnosis, examination, and prediction of health status by comparing and analyzing multiple electrocardiograms taken at different times of the same subject, and can determine the presence or absence of disease in time series. It is about a deep learning-based health status prediction system using multiple electrocardiograms that can analyze and predict trends and disease severity.
- an electrocardiogram is a graphical recording of electrical potentials related to heartbeat on the surface of the body.
- an exercise stress electrocardiogram there is an exercise stress electrocardiogram and an activity electrocardiogram.
- This type of electrocardiogram is used for examination and diagnosis of circulatory diseases and has the advantages of being simple, relatively inexpensive, non-invasive, and easily recorded repeatedly.
- the standard 12-lead ECG used in medical institutions attaches 6 electrodes to the front of the chest and 3 electrodes to each extremity, collects all 12-lead ECG information, and makes a comprehensive judgment to diagnose the disease.
- a 12-lead electrocardiogram is a recording of the heart's potential in 12 electrical directions centered on the heart, and through this, heart-related diseases limited to one area can be read.
- the technical task to be achieved by the idea of the present invention is to improve the accuracy of measurement, diagnosis, examination, and prediction of health status by comparing and analyzing multiple electrocardiograms taken at different times of the same examinee based on deep learning.
- the goal is to provide a deep learning-based health status prediction system using multiple electrocardiograms.
- an embodiment of the present invention includes an electrocardiogram measuring unit that measures electrocardiograms of the same examinee at different time points with time differences to obtain N pieces of electrocardiogram data; Integrate semantic feature values or output values from the N ECG data provided from the ECG measurement unit through a pre-built diagnostic algorithm by learning the N ECG data at different times and the disease dataset corresponding to the ECG data.
- a prediction unit that generates disease information that predicts the presence or absence of the disease, its trend, and the degree of the disease
- an input unit that converts and inputs the N pieces of ECG data from the ECG measurement unit to the prediction unit.
- the input unit may integrate the N pieces of ECG data input from the ECG measurement unit into single ECG data.
- the input unit may integrate the N pieces of ECG data into each ECG data unit.
- the input unit may divide the N ECG data into each lead and integrate the ECG data for each of the divided leads.
- the input unit may divide the N ECG data into each lead, divide them into bit units, and then pair and integrate the bit units of the same lead.
- the prediction unit may extract semantic feature values of each of the N ECG data provided from the input unit, analyze the semantic feature values in an integrated manner, and compare and analyze the N ECG data to generate the disease information. .
- the present invention based on deep learning, it is possible to increase the accuracy of measurement, diagnosis, examination, and prediction of health status by comparing and analyzing multiple electrocardiograms taken at different times of the same subject, and to improve the accuracy of disease in time series. It has the effect of analyzing and predicting the presence, trend, and degree of disease.
- Figure 1 shows the configuration of a deep learning-based health status prediction system using a plurality of electrocardiograms according to an embodiment of the present invention.
- FIG. 2 illustrates an ECG integration method using a deep learning-based health status prediction system using the plurality of ECGs of FIG. 1.
- FIG. 3 illustrates a flowchart of a prediction method using a deep learning-based health status prediction system using a plurality of electrocardiograms in FIG. 1.
- the deep learning-based health status prediction system using a plurality of electrocardiograms includes an electrocardiogram measurement unit 110 that acquires N pieces of electrocardiogram data by measuring electrocardiograms of the same examinee at different time points with time differences. ), Semantic feature values or output values from the N ECG data provided from the ECG measurement unit 110 through a pre-built diagnostic algorithm by learning the N ECG data at different time points and the disease dataset corresponding to the ECG data.
- N ECG data from the prediction unit 120 and the ECG measurement unit 110 which integrates and generates disease information that predicts the presence or absence of the disease, its trend, and the degree of the disease, is converted and input to the prediction unit 120.
- the main idea is to more accurately measure, diagnose, examine, and predict health status from a plurality of electrocardiogram data at different times for the same examinee, including an input unit 130 that performs a medical examination.
- the ECG measurement unit 110 measures the ECG of the same examinee at different time points with a time difference and acquires N pieces of ECG data.
- the ECG measurement unit 110 can measure N ECG data for each measurement at two or more points in time by measuring the same examinee over a long period of time, such as monthly, quarterly, or yearly.
- the ECG measurement unit 110 includes a wearable ECG patch, a smartwatch, a 6-lead ECG bar that measures for a short period of time, or an ECG device installed in a medical institution that can measure ECG in contact or non-contact during daily life, and can measure asynchronous or synchronous ECG. Can be measured and provided to the input unit 130.
- the prediction unit 120 learns N ECG data from different time points of the same examinee, which is big data accumulated in the medical institution server, and a dataset of diseases corresponding to the ECG data, and uses a pre-built diagnostic algorithm.
- Semantic feature values or output values from N pieces of ECG data provided from the ECG measurement unit 110 are integrated to generate disease information that predicts the presence or absence of the disease, its trend, and the degree of the disease.
- the semantic feature value is a spatial time-series feature value extracted from ECG data, and based on this, the prediction unit 120 compares and analyzes N ECG data at different times to determine the current value at the time the ECG data was last measured. Health status or future health status can be measured, diagnosed, examined, and predicted.
- the prediction unit 120 extracts semantic feature values of N ECG data at different time points provided from the input unit 130, which will be described later, integrates and analyzes the semantic feature values, and analyzes the N ECG data at different time points.
- Time-series disease information can also be generated by comparing and analyzing data.
- various deep learning algorithms such as CNN, LSTM, RNN, and MLP are used, or machine learning methods such as logistic regression, principle-based model, random forest, and support vector machine are used. You can also use .
- the input unit 130 converts N ECG data from the ECG measurement unit 110 and inputs it to the prediction unit 120.
- the input unit 130 may integrate N ECG data from different time points of the same subject input from the ECG measurement unit 110 into a single ECG data and input it to the prediction unit 120.
- the input unit 130 can integrate N pieces of ECG data from different time points into the same length of time axis for each ECG data unit.
- N ECG data from different time points are not measured with the same length of the time axis, they are cut to the length of the minimum time axis, or through a deep learning algorithm, an ECG of the insufficient length of the remaining ECG data is generated according to the length of the maximum time axis. You can also adjust the overall length by doing this.
- the input unit 130 determines the individual characteristics of the ECG data for each lead of N ECG data at different time points (T1, T2), and inputs each N ECG data according to the identified characteristics. It may be divided by lead, and the ECG data for each lead may be integrated and input into the prediction unit 120.
- the input unit 130 divides N ECG data at different times into each lead, divides the ECG data for each lead into bits, and then divides the ECG data for each lead into bits.
- Bit units can also be paired, integrated, and input to the prediction unit 120.
- a noise removal unit 140 is further included to minimize noise in the ECG data provided from the ECG measurement unit 110 to the input unit 130 and thereby increase the reliability of the disease information generated by the prediction unit 120.
- the noise removal unit 140 may generate ECG data for each lead with less noise and unique style data of the ECG data for each lead, based on a plurality of ECG data, for example, standard 12-lead ECG data accumulated in medical institutions.
- the unique style is extracted from the ECG data measured by the ECG measurement unit 110 by reflecting the characteristics of the examinee and the characteristics of the measurement method, and the extracted unique style It is possible to convert and generate ECG data of a specific guidance style that does not contain noise.
- the noise removal unit 140 reflects the characteristics of the examinee due to the age, gender, disease, etc., the attachment position of the electrodes, and the characteristics of the measurement method due to the ECG device, etc., to create a unique style for each induction of each ECG data. It can be determined with accuracy, and based on this, ECG data for each lead can be converted and generated more accurately.
- the ECG data generating unit 150 When measuring by the ECG measuring unit 110, if the ECG data contains a lot of noise in some leads or sections of the ECG data or the electrode contact is lost and the measurement is not performed properly, the ECG data generating unit 150 Through this, it is possible to generate a noise-free electrocardiogram and fill in the missing electrocardiogram data to perform more accurate measurement, diagnosis, examination, and prediction of health status.
- the prediction unit 120 includes diseases of the circulatory system, endocrine, nutritional and metabolic diseases, neoplastic diseases, mental and behavioral disorders, diseases of the nervous system, diseases of the eyes and appendages, and diseases of the ears and mastoids.
- deformation and chromosomal abnormalities can be diagnosed and predicted.
- the prediction unit 120 damage caused by physical trauma can be confirmed, prognosis can be confirmed, pain can be measured, the risk of death or worsening due to trauma can be predicted, and concurrent complications can be detected or predicted. It can also identify specific conditions that appear before and after birth.
- the prediction unit 120 as a healthcare area, aging, sleep, weight, blood pressure, blood sugar, oxygen saturation, metabolism, stress, tension, fear, drinking, smoking, problem behavior, lung capacity, exercise volume, and pain management , obesity, body mass, body composition, diet, type of exercise, lifestyle pattern recommendation, emergency management, chronic disease management, medication prescription, test recommendation, checkup recommendation, nursing, telehealth management, telemedicine, vaccination and post-vaccination management, etc. It may be possible to measure, diagnose, examine, and predict the health status of the person being examined, which can lead to services.
- the disease can be diagnosed more accurately by comparing the electrocardiogram taken at a previous time and the electrocardiogram taken at a subsequent time. If the electrocardiogram at the time showed findings of myocardial infarction due to ST-segment elevation, etc., but the examination was normal, but the electrocardiogram at a follow-up time, such as 1 year later, showed findings of myocardial infarction due to the same ST-segment elevation as before, there is a possibility of myocardial infarction. This is less. In other words, since it is impossible to have a myocardial infarction for a long period of one year, the subject should be regarded as a normal patient.
- FIG. 3 illustrates a flowchart of a prediction method by a deep learning-based health status prediction system using a plurality of electrocardiograms in FIG. 1, which is briefly described in detail as follows.
- the ECG of the same examinee can be measured at different time points with a time difference, and N pieces of ECG data can be obtained and provided to the input unit 130 (S110).
- the ECG measurement unit 110 can measure N ECG data for each measurement at two or more points in time by measuring the same examinee over a long period of time, such as monthly, quarterly, or yearly.
- the ECG measurement unit 110 includes a wearable ECG patch, a smartwatch, a 6-lead ECG bar that measures for a short period of time, or an ECG device installed in a medical institution that can measure ECG in contact or non-contact during daily life, and can measure asynchronous or synchronous ECG. Can be measured and provided to the input unit 130.
- N pieces of ECG data from the ECG measurement unit 110 are converted and input into the prediction unit 120 through the input unit 130 (S120).
- the input unit 130 may integrate N ECG data from different time points of the same subject input from the ECG measurement unit 110 into a single ECG data and input it to the prediction unit 120.
- the input unit 130 can integrate N pieces of ECG data from different time points into the same length of time axis for each ECG data unit.
- N ECG data from different time points are not measured with the same length of the time axis, they are cut to the length of the minimum time axis, or through a deep learning algorithm, an ECG of the insufficient length of the remaining ECG data is generated according to the length of the maximum time axis. You can also adjust the overall length by doing this.
- the input unit 130 determines the individual characteristics of the ECG data for each lead of N ECG data at different time points (T1, T2), and inputs each N ECG data according to the identified characteristics. It may be divided by lead, and the ECG data for each lead may be integrated and input into the prediction unit 120.
- the input unit 130 divides N ECG data at different times into each lead, divides the ECG data for each lead into bits, and then divides the ECG data for each lead into bits.
- Bit units can also be paired, integrated, and input to the prediction unit 120.
- the noise of the ECG data provided from the ECG measurement unit 110 to the input unit 130 is minimized through the noise removal unit 140 to further increase the reliability of the disease information generated by the prediction unit 120. It may further include a step (S125), where, for example, the noise removal unit 140 is based on a plurality of ECG data, for example, standard 12-lead ECG data accumulated in a medical institution, ECG data for each lead and each lead with less noise.
- the unique style is derived from the ECG data measured by the ECG measurement unit 110 by reflecting the characteristics of the examinee and the characteristics of the measurement method. can be extracted and generated by converting it into ECG data of a specific guidance style that does not contain noise through the extracted unique style.
- the noise removal unit 140 reflects the characteristics of the examinee due to the age, gender, disease, etc., the attachment position of the electrodes, and the characteristics of the measurement method due to the ECG device, etc., to create a unique style for each induction of each ECG data. It can be determined with accuracy, and based on this, ECG data for each lead can be converted and generated more accurately.
- the ECG data generating unit 150 When measuring by the ECG measuring unit 110, if the ECG data contains a lot of noise in some leads or sections of the ECG data or the electrode contact is lost and the measurement is not performed properly, the ECG data generating unit 150 Through this, it is possible to generate a noise-free electrocardiogram and fill in the missing electrocardiogram data to perform more accurate measurement, diagnosis, examination, and prediction of health status.
- a diagnostic algorithm is built in advance by learning N ECG data from different time points of the same subjects, which is big data accumulated in the medical institution server, and a dataset of diseases corresponding to the ECG data.
- semantic feature values or output values are integrated from the N pieces of ECG data provided from the ECG measurement unit 110 to generate disease information that predicts the presence or absence of the disease, its trend, and the degree of the disease (S130).
- the semantic feature value is a spatial time-series feature value extracted from ECG data, and based on this, the prediction unit 120 compares and analyzes N ECG data at different times to determine the current value at the time the ECG data was last measured. Health status or future health status can be measured, diagnosed, examined, and predicted.
- the prediction unit 120 extracts semantic feature values of N ECG data at different time points provided from the input unit 130, which will be described later, integrates and analyzes the semantic feature values, and analyzes the N ECG data at different time points.
- Time-series disease information can also be generated by comparing and analyzing data.
- various deep learning algorithms such as CNN, LSTM, RNN, and MLP are used, or machine learning methods such as logistic regression, principle-based model, random forest, and support vector machine are used. You can also use .
- the prognosis prediction unit it is possible to measure, diagnose, examine, and predict not only the health status due to the individual diseases mentioned above, but also the health status that appears in complex, and predict the worsening or alleviation of the health status of the person being examined. , it is possible to predict short-term and long-term prognosis, predict the condition of transition or complication from one disease to another, and learn about the improvement or deterioration of the health condition according to the analysis and prediction of specific drugs and electrocardiograms to determine the health status. Accordingly, a specific drug may be recommended (S140).
- the accuracy of measurement, diagnosis, examination, and prediction of health status can be improved by comparing and analyzing multiple electrocardiograms taken at different times of the same subject, and can improve the accuracy of health status measurement, diagnosis, examination, and prediction.
- the presence, trend, and degree of disease can be analyzed and predicted.
Landscapes
- Health & Medical Sciences (AREA)
- Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Medical Informatics (AREA)
- Public Health (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- Pathology (AREA)
- Cardiology (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Surgery (AREA)
- Heart & Thoracic Surgery (AREA)
- Data Mining & Analysis (AREA)
- Veterinary Medicine (AREA)
- Animal Behavior & Ethology (AREA)
- Artificial Intelligence (AREA)
- Databases & Information Systems (AREA)
- Epidemiology (AREA)
- Primary Health Care (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- Theoretical Computer Science (AREA)
- Physiology (AREA)
- General Physics & Mathematics (AREA)
- Fuzzy Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Software Systems (AREA)
- Psychiatry (AREA)
- Signal Processing (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
Description
Claims (6)
- 동일 피검진자의 시차를 둔 서로 다른 시점의 심전도를 각각 측정하여 N개의 심전도 데이터를 각각 획득하는 심전도 측정부;서로 다른 시점의 N개의 심전도 데이터 및 상기 심전도 데이터에 해당하는 질환의 데이터셋을 학습하여 미리 구축된 진단 알고리즘을 통해, 상기 심전도 측정부로부터 제공되는 상기 N개의 심전도 데이터로부터 시멘틱 특징값 또는 출력값을 통합하여 질환의 유무와 추이와 질환의 정도를 예측하는 질환정보를 생성하는, 예측부; 및상기 심전도 측정부로부터의 상기 N개의 심전도 데이터를 상기 예측부로 변환하여 입력하는 입력부;를 포함하는,복수의 심전도를 이용한 딥러닝기반 건강상태 예측 시스템.
- 제1항에 있어서,상기 입력부는, 상기 심전도 측정부로부터 입력되는 상기 N개의 심전도 데이터를 단일 심전도 데이터로 통합하는 것을 특징으로 하는,복수의 심전도를 이용한 딥러닝기반 건강상태 예측 시스템.
- 제2항에 있어서,상기 입력부는 상기 N개의 심전도 데이터를 각 심전도 데이터 단위로 통합하는 것을 특징으로 하는,복수의 심전도를 이용한 딥러닝기반 건강상태 예측 시스템.
- 제2항에 있어서,상기 입력부는 상기 N개의 심전도 데이터를 각각 유도별로 구분하고, 상기 각 구분된 유도별 심전도 데이터를 유도별로 통합하는 것을 특징으로 하는,복수의 심전도를 이용한 딥러닝기반 건강상태 예측 시스템.
- 제2항에 있어서,상기 입력부는 상기 N개의 심전도 데이터를 각 유도별로 구분하고 비트단위로 분할한 후 동일 유도의 비트단위를 상호 짝지어 통합하는 것을 특징으로 하는,복수의 심전도를 이용한 딥러닝기반 건강상태 예측 시스템.
- 제1항에 있어서,상기 예측부는, 상기 입력부로부터 제공되는 상기 N개의 심전도 데이터의 시멘틱 특징값을 각각 추출하고, 상기 시멘틱 특징값을 통합 분석하여 상기 N개의 심전도 데이터를 비교분석하여 상기 질환정보를 생성하는 것을 특징으로 하는,복수의 심전도를 이용한 딥러닝기반 건강상태 예측 시스템.
Priority Applications (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202280056116.XA CN117915834A (zh) | 2022-08-18 | 2022-08-18 | 利用复数个心电图的基于深度学习的健康状态预测系统 |
PCT/KR2022/012364 WO2024038930A1 (ko) | 2022-08-18 | 2022-08-18 | 복수의 심전도를 이용한 딥러닝기반 건강상태 예측 시스템 |
JP2024508509A JP2024533041A (ja) | 2022-08-18 | 2022-08-18 | 複数の心電図を用いたディープラーニングに基づく健康状態予測システム |
EP22953377.3A EP4371492A1 (en) | 2022-08-18 | 2022-08-18 | Deep learning-based health condition prediction system using plurality of electrocardiograms |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/KR2022/012364 WO2024038930A1 (ko) | 2022-08-18 | 2022-08-18 | 복수의 심전도를 이용한 딥러닝기반 건강상태 예측 시스템 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2024038930A1 true WO2024038930A1 (ko) | 2024-02-22 |
Family
ID=89941921
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/KR2022/012364 WO2024038930A1 (ko) | 2022-08-18 | 2022-08-18 | 복수의 심전도를 이용한 딥러닝기반 건강상태 예측 시스템 |
Country Status (4)
Country | Link |
---|---|
EP (1) | EP4371492A1 (ko) |
JP (1) | JP2024533041A (ko) |
CN (1) | CN117915834A (ko) |
WO (1) | WO2024038930A1 (ko) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR101857624B1 (ko) * | 2017-08-21 | 2018-05-14 | 동국대학교 산학협력단 | 임상 정보를 반영한 의료 진단 방법 및 이를 이용하는 장치 |
KR102362679B1 (ko) * | 2021-01-27 | 2022-02-14 | 주식회사 뷰노 | 심전도 신호 기반의 만성질환 예측 방법 |
KR20220040516A (ko) * | 2020-09-22 | 2022-03-30 | 주식회사 바디프랜드 | 심전도를 이용한 딥러닝 기반 관상동맥질환 예측 시스템 |
KR20220098600A (ko) * | 2021-01-04 | 2022-07-12 | 주식회사 바디프랜드 | 시차를 둔 단일유도 심전도를 통한 딥러닝기반의 복수 유도 심전도 생성 시스템 |
KR20230025963A (ko) * | 2021-08-17 | 2023-02-24 | 주식회사 메디컬에이아이 | 복수의 심전도를 이용한 딥러닝기반 건강상태 예측 시스템 |
-
2022
- 2022-08-18 CN CN202280056116.XA patent/CN117915834A/zh active Pending
- 2022-08-18 EP EP22953377.3A patent/EP4371492A1/en active Pending
- 2022-08-18 WO PCT/KR2022/012364 patent/WO2024038930A1/ko active Application Filing
- 2022-08-18 JP JP2024508509A patent/JP2024533041A/ja active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR101857624B1 (ko) * | 2017-08-21 | 2018-05-14 | 동국대학교 산학협력단 | 임상 정보를 반영한 의료 진단 방법 및 이를 이용하는 장치 |
KR20220040516A (ko) * | 2020-09-22 | 2022-03-30 | 주식회사 바디프랜드 | 심전도를 이용한 딥러닝 기반 관상동맥질환 예측 시스템 |
KR20220098600A (ko) * | 2021-01-04 | 2022-07-12 | 주식회사 바디프랜드 | 시차를 둔 단일유도 심전도를 통한 딥러닝기반의 복수 유도 심전도 생성 시스템 |
KR102362679B1 (ko) * | 2021-01-27 | 2022-02-14 | 주식회사 뷰노 | 심전도 신호 기반의 만성질환 예측 방법 |
KR20230025963A (ko) * | 2021-08-17 | 2023-02-24 | 주식회사 메디컬에이아이 | 복수의 심전도를 이용한 딥러닝기반 건강상태 예측 시스템 |
Also Published As
Publication number | Publication date |
---|---|
CN117915834A (zh) | 2024-04-19 |
JP2024533041A (ja) | 2024-09-12 |
EP4371492A1 (en) | 2024-05-22 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105228508B (zh) | 一种测定用于分类的危险评分的系统 | |
Nepi et al. | Validation of the heart-rate signal provided by the Zephyr bioharness 3.0 | |
Altıntop et al. | A novel approach for detection of consciousness level in comatose patients from EEG signals with 1-D convolutional neural network | |
WO2023022485A1 (ko) | 비동기 심전도를 이용한 건강상태 예측 시스템 | |
WO2023022484A9 (ko) | 단일유도 심전도기기를 활용한 건강상태 예측 시스템 | |
WO2024038930A1 (ko) | 복수의 심전도를 이용한 딥러닝기반 건강상태 예측 시스템 | |
KR20230025963A (ko) | 복수의 심전도를 이용한 딥러닝기반 건강상태 예측 시스템 | |
Orphanidou et al. | Signal quality assessment in physiological monitoring: requirements, practices and future directions | |
TWI756793B (zh) | 一種通道資訊處理系統 | |
TWI748485B (zh) | 一種資訊處理系統及其方法 | |
WO2023022519A1 (ko) | 2유도 심전도 데이터를 이용한 복수개의 표준 심전도 데이터 생성 시스템 | |
WO2023022493A1 (ko) | 딥러닝 기반 심전도 데이터의 노이즈 제거 시스템 | |
WO2023022521A1 (ko) | 딥러닝기반 모델 및 원칙기반 모델 통합 심전도 판독 시스템 | |
US20240350067A1 (en) | System for generating plurality of pieces of standard electrocardiogram data using 2-lead electrocardiogram data | |
Behar et al. | Editorial on Remote Health Monitoring: from chronic diseases to pandemics | |
US20240350094A1 (en) | Deep learning-based electrocardiogram data noise removal system | |
WO2024014844A1 (ko) | 심전도에 기반하여 체성분을 측정하는 방법 및 그 장치 | |
WO2023022516A1 (ko) | 딥러닝 알고리즘을 기반으로 복수개의 표준 심전도 데이터를 생성하는 방법 | |
WO2022181948A1 (ko) | 웨어러블 심전도 리듬 분석을 이용한 심장 위험도 산출 시스템 및 방법 | |
Braojos et al. | Level Accuracy of Automatic and Real Time Detection of Atrial Fibrillation with A New Wireless ECG Recorder (The SmartCardia) |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
ENP | Entry into the national phase |
Ref document number: 2024508509 Country of ref document: JP Kind code of ref document: A |
|
WWE | Wipo information: entry into national phase |
Ref document number: 202280056116.X Country of ref document: CN |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2022953377 Country of ref document: EP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 202447010915 Country of ref document: IN |
|
ENP | Entry into the national phase |
Ref document number: 2022953377 Country of ref document: EP Effective date: 20240215 |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 22953377 Country of ref document: EP Kind code of ref document: A1 |