Personalization of artificial intelligence models for analysis of cardiac rhythms

A method for operating processing circuitry of a computing system, the method comprising:

• obtaining, by the processing circuitry of the computing system, a cardiac electrogram (EGM) strip for a current patient;
• applying, by the processing circuitry of the computing system, a first cardiac rhythm classifier neural network (CRC) with a segment of the cardiac EGM strip as input to generate, based on the segment of the cardiac EGM strip, first data indicating an aspect of a cardiac rhythm of the current patient, wherein the aspect of the cardiac rhythm of the current patient is an occurrence of a cardiac arrhythmia in the cardiac rhythm of the current patient, the first CRC is trained on training cardiac EGM strips from a first population that includes a plurality of patients, and the first data is based on a first probability that the segment of the cardiac EGM strip represents at least one occurrence of the cardiac arrhythmia;
• applying, by the processing circuitry of the computing system, a second CRC with the segment of the cardiac EGM strip as input to generate, based on the segment of the cardiac EGM strip, second data indicating the aspect of the cardiac rhythm of the current patient, wherein the second CRC is a version of the first CRC that is trained on training cardiac EGM strips from a second population that is smaller than the first population and the second data is based on a second probability that the segment of the cardiac EGM strip represents at least one occurrence of the cardiac arrhythmia;
• and
• generating, by the processing circuitry of the computing system, output data indicating a cardiac arrhythmia occurred in the cardiac rhythm of the current patient based on the second data.

A computing system comprising:

• a storage medium configured to store a cardiac electrogram (EGM) strip for a current patient;
• and
• processing circuitry configured to:
• apply a first cardiac rhythm classifier neural network (CRC) with a segment of the cardiac EGM strip as input to generate, based on the segment of the cardiac EGM strip, first data indicating an aspect of a cardiac rhythm of the current patient, wherein, the aspect of the cardiac rhythm of the current patient comprises a morphological aspect of occurrences of a cardiac arrhythmia that occur in the cardiac rhythm of the current patient, the first CRC is trained on training cardiac EGM strips from a first population that includes a plurality of patients, and the first data, generated by the first CRC, comprises first data indicating the morphological aspect of an occurrence of the cardiac arrhythmia represented in the segment of the cardiac EGM strip;
• apply a second CRC with the segment of the cardiac EGM strip as input to generate, based on the segment of the cardiac EGM strip, second data indicating the aspect of the cardiac rhythm of the current patient, wherein the second CRC is a version of the first CRC that is trained on training cardiac EGM strips from a second population that is smaller than the first population and the second data, generated by the second CRC, comprises second data indicating the morphological aspect of the occurrence of the cardiac arrhythmia represented in the segment of the cardiac EGM strip;
• and
• generate output data including comparison data based on a comparison of the first data indicating the morphological aspect of the occurrence of the cardiac arrhythmia represented in the segment of the cardiac EGM strip and the second data indicating the morphological aspect of the occurrence of the cardiac arrhythmia represented in the segment of the cardiac EGM strip.

A computing system comprising:

• a storage medium configured to store a cardiac electrogram (EGM) strip for a current patient;
• and
• processing circuitry configured to:
• apply a first cardiac rhythm classifier neural network (CRC) with a segment of the cardiac EGM strip as input to generate, based on the segment of the cardiac EGM strip, first data indicating an aspect of a cardiac rhythm of the current patient, wherein the aspect of the cardiac rhythm of the current patient is an occurrence of a cardiac arrhythmia in the cardiac rhythm of the current patient, the first CRC is trained on training cardiac EGM strips from a first population that includes a plurality of patients, and the first data is based on a first probability that the segment of the cardiac EGM strip represents at least one occurrence of the cardiac arrhythmia;
• apply a second CRC with the segment of the cardiac EGM strip as input to generate, based on the segment of the cardiac EGM strip, second data indicating the aspect of the cardiac rhythm of the current patient, wherein the second CRC is a version of the first CRC that is trained on training cardiac EGM strips from a second population that is smaller than the first population and the second data is based on a second probability that the segment of the cardiac EGM strip represents at least one occurrence of the cardiac arrhythmia;
• and
• generate output data indicating a cardiac arrhythmia occurred in the cardiac rhythm of the current patient based on the second data.

A non-transitory computer-readable data storage medium having instructions stored thereon that, when executed, cause a computing system to:

• obtain a cardiac electrogram (EGM) strip for a current patient;
• apply a first cardiac rhythm classifier neural network (CRC) with a segment of the cardiac EGM strip as input to generate, based on the segment of the cardiac EGM strip, first data indicating an aspect of a cardiac rhythm of the current patient, wherein, the aspect of the cardiac rhythm of the current patient comprises a morphological aspect of occurrences of a cardiac arrhythmia that occur in the cardiac rhythm of the current patient, the first CRC is trained on training cardiac EGM strips from a first population that includes a plurality of patients, and the first data, generated by the first CRC, comprises first data indicating the morphological aspect of an occurrence of the cardiac arrhythmia represented in the segment of the cardiac EGM strip;
• apply a second CRC with the segment of the cardiac EGM strip as input to generate, based on the segment of the cardiac EGM strip, second data indicating the aspect of the cardiac rhythm of the current patient, wherein the second CRC is a version of the first CRC that is trained on training cardiac EGM strips from a second population that is smaller than the first population and the second data, generated by the second CRC, comprises second data indicating the morphological aspect of the occurrence of the cardiac arrhythmia represented in the segment of the cardiac EGM strip;
• and
• generate output data including data based on a comparison of the first data indicating the morphological aspect of the occurrence of the cardiac arrhythmia represented in the segment of the cardiac EGM strip and the second data indicating the morphological aspect of the occurrence of the cardiac arrhythmia represented in the segment of the cardiac EGM strip.