Electrocardiogram analysis apparatus, electrocardiogram analyzing method, and non-transitory computer-readable storage medium

A method for analyzing an electrocardiogram executed by a computer, which supports determination of an advisability of doing an additional examination or administering a drug on a person to be analyzed, comprising:

• acquiring, by a processor, a machine learning model that is stored in the memory and is realized by machine learning that uses first training electrocardiogram data of a patient who has been determined to have paroxysmal arrhythmia during a non-paroxysmal period during which a doctor determined by visual inspection in accordance with medical diagnostic criteria that no episode of the paroxysmal arrhythmia occurs, the non-paroxysmal period being a predetermined period of at least one of before or after a period during which the doctor determined in accordance with the medical diagnostic criteria that an episode of the paroxysmal arrhythmia occurs and second training electrocardiogram data of a person who has been determined not to have the paroxysmal arrhythmia so as to determine whether there is a sign of the paroxysmal arrhythmia in the non-paroxysmal period, the machine learning model outputting, when an electrocardiogram data of a person who has not been determined whether to have the paroxysmal arrhythmia is input, information regarding whether the input electrocardiogram data is an electrocardiogram data of a person having the paroxysmal arrhythmia;
• inputting, by the processor, electrocardiogram data of the person to be analyzed who has not been determined whether to have the paroxysmal arrhythmia, which is a subject of analysis, into the machine learning model stored in the memory;
• displaying, by the processor, on an information terminal, abnormality information and a portion of the non-paroxysmal period in electrocardiogram data associated with each other, where the abnormality information corresponds to information output from the machine learning model stored in the memory and is about whether the person to be analyzed has the paroxysmal arrhythmia, and the portion of the non-paroxysmal period in electrocardiogram data is inputted into the machine learning model in order to acquire the abnormality information;
• and
• receiving, by a communication part, the electrocardiogram data from an electrocardiograph and transmitting at least a part of the electrocardiogram data from the output control part to an external device.

A system including an electrocardiograph and an electrocardiogram analysis apparatus, which supports determination of an advisability of doing an additional examination or administering a drug on a person to be analyzed, the electrocardiograph being configured to generate electrocardiogram data, and

• the electrocardiogram analysis apparatus comprising:
• a processor;
• a memory;
• a machine learning part that has a machine learning model, which is stored in the memory, realized by machine learning that uses first training electrocardiogram data of a patient who has been determined to have paroxysmal arrhythmia during a non-paroxysmal period during which a doctor determined by visual inspection in accordance with medical diagnostic criteria that no episode of the paroxysmal arrhythmia occurs, the non-paroxysmal period being a predetermined period of at least one of before or after a period during which the doctor determined in accordance with the medical diagnostic criteria that an episode of the paroxysmal arrhythmia occurs and second training electrocardiogram data of a person who has been determined not to have the paroxysmal arrhythmia so as to determine whether there is a sign of the paroxysmal arrhythmia in the non-paroxysmal period, the machine learning model outputting, when an electrocardiogram data of a person who has not been determined whether to have the paroxysmal arrhythmia is input, information regarding whether the input electrocardiogram data is an electrocardiogram data of a person having the paroxysmal arrhythmia;
• an input processing part that inputs, by the processor, electrocardiogram data of the person to be analyzed who has not been determined whether to have the paroxysmal arrhythmia, which is a subject of analysis, into the machine learning model stored in the memory;
• an output control part that displays, by the processor, on an information terminal, abnormality information and a portion of the non-paroxysmal period in electrocardiogram data associated with each other, where the abnormality information corresponds to information output from the machine learning model stored in the memory and is about whether the person to be analyzed has the paroxysmal arrhythmia, and the portion of the non-paroxysmal period in electrocardiogram data is inputted into the machine learning model in order to acquire the abnormality information;
• and
• a communication part configured to receive the electrocardiogram data from the electrocardiograph and transmit at least a part of the electrocardiogram data from the output control part to an external device.