Method of processing electrocardiogram signal

Abstract

A method of processing an electrocardiogram signal by an electrocardiogram signal processing device including at least one processor is provided. The method includes calculating one or more time intervals between peaks of an electrocardiogram signal, determining peaks having time interval less than a preset first value as a first abnormal period of the electrocardiogram signal, and determining whether or not the electrocardiogram signal without the first abnormal period is abnormal by comparing the electrocardiogram signal without the first abnormal period with a pattern of a preset region having a measurement time of an R peak of a normal electrocardiogram signal using a similarity determination method. The preset first value is set by using an average value of time intervals between normal R-peaks.

What is claimed is: 
     
         1 . A method of processing an electrocardiogram signal by an electrocardiogram signal processing device including at least one processor, the method comprising:
 calculating one or more time intervals between peaks of an electrocardiogram signal;   determining peaks having time interval less than a preset first value as a first abnormal period of the electrocardiogram signal; and   determining whether or not the electrocardiogram signal without the first abnormal period is abnormal by comparing the electrocardiogram signal without the first abnormal period with a pattern of a preset region having a measurement time of an R peak of a normal electrocardiogram signal using a similarity determination method,   wherein the preset first value is set by using an average value of time intervals between normal R-peaks.   
     
     
         2 . The method of  claim 1 , further comprising:
 selecting normal R-peaks of the electrocardiogram signal upon determination that the electrocardiogram signal is normal.   
     
     
         3 . The method of  claim 2 , further comprising, after the normal R-peaks are selected, obtaining a heart rate or heart rate variability based on time intervals between the normal R-peaks. 
     
     
         4 . The method of  claim 2 , further comprising, after the normal R-peaks are selected, storing signal periods including the normal R-peaks in a memory or transmitted to an external device. 
     
     
         5 . The method of  claim 1 , wherein the step of determining of whether or not the electrocardiogram signal is abnormal further includes determining a period of an abnormal electrocardiogram signal included in a window, and
 during the period, a complexity of the electrocardiogram signal, calculated for a window of a preset size, falls out of a preset range such that the period is determined as an abnormal electrocardiogram signal period.   
     
     
         6 . The method according to  claim 5 , further comprising:
 generating a machine learning model based on waveforms of periods except for the first abnormal period of the electrocardiogram signal, and   updating the machine learning model by using waveforms of the first abnormal period in the electrocardiogram signal.   
     
     
         7 . The method according to  claim 1 , wherein the step of determining of whether or not the electrocardiogram signal without the first abnormal period is abnormal further includes:
 determining whether or not the electrocardiogram signal is abnormal by using a machine learning model generated based on waveforms of periods except for a previously obtained abnormal period of the electrocardiogram signal, and   updating the machine learning model by using waveforms of the first abnormal period in the electrocardiogram signal.   
     
     
         8 . The method according to  claim 1 , wherein the step of determining whether or not the electrocardiogram signal without the first abnormal period is abnormal further comprises determining whether or not the electrocardiogram signal is abnormal by using a machine-learning algorithm based on waveforms of the electrocardiogram signal acquired from a plurality of objects. 
     
     
         9 . The method according to  claim 1 , wherein the step of determining whether or not the electrocardiogram signal without the first abnormal period is abnormal further comprises determining whether or not the electrocardiogram signal is abnormal by using a machine-learning technique based on waveforms of the electrocardiogram signal acquired from a plurality of objects. 
     
     
         10 . A method of processing an electrocardiogram signal by an electrocardiogram signal processing device including at least one processor, the method comprising:
 calculating one or more time intervals between peaks of an electrocardiogram signal;   determining peaks having time interval less than a preset first value as a first abnormal period of the electrocardiogram signal; and   determining whether or not the electrocardiogram signal without the first abnormal period is abnormal by comparing a complexity of the electrocardiogram signal without the first abnormal period with a complexity of a preset region having a measurement time of an R peak of a normal electrocardiogram signal using a method of calculating the complexity,   wherein the preset first value is set by using an average value of time intervals between normal R-peaks.   
     
     
         11 . The method of  claim 10 , wherein the method of calculating the complexity is Shannon entropy, turning point ratio (TPR), root mean square of successive difference (RMSSD), or a combination thereof. 
     
     
         12 . The method of  claim 10 , further comprising:
 selecting the normal R-peaks of the electrocardiogram signal upon determination that the electrocardiogram signal is normal.   
     
     
         13 . The method of  claim 12 , further comprising, after the normal R-peaks are selected, obtaining a heart rate or a heart rate variability based on time intervals between the normal R-peaks. 
     
     
         14 . The method of  claim 11 , further comprising, after the normal R-peaks are selected, storing signal periods including the normal R-peaks in a memory or transmitted to an external device. 
     
     
         15 . The method of  claim 10 , wherein the step of determining of whether or not the electrocardiogram signal is abnormal further includes determining a period of an abnormal electrocardiogram signal included in a window, and
 during the period, a complexity of the electrocardiogram signal, calculated for a window of a preset size, falls out of a preset range such that the period is determined as an abnormal electrocardiogram signal period.   
     
     
         16 . The method according to  claim 15 , further comprising:
 generating a machine learning model based on waveforms of periods except for the first abnormal period of the electrocardiogram signal, and   updating the machine learning model by using waveforms of the first abnormal period in the electrocardiogram signal.   
     
     
         17 . The method according to  claim 10 , wherein the step of determining of whether or not the electrocardiogram signal without the first abnormal period is abnormal further includes:
 determining whether or not the electrocardiogram signal is abnormal by using a machine learning model generated based on waveforms of periods except for a previously obtained abnormal period of the electrocardiogram signal, and   updating the machine learning model by using waveforms of the first abnormal period in the electrocardiogram signal.   
     
     
         18 . The method according to  claim 10 , wherein the step of determining whether or not the electrocardiogram signal without the first abnormal period is abnormal further comprises determining whether or not the electrocardiogram signal is abnormal by using a machine-learning algorithm based on waveforms of the electrocardiogram signal acquired from a plurality of objects. 
     
     
         19 . The method according to  claim 10 , wherein the step of determining whether or not the electrocardiogram signal without the first abnormal period is abnormal further comprises determining whether or not the electrocardiogram signal is abnormal by using a machine-learning technique based on waveforms of the electrocardiogram signal acquired from a plurality of objects.