Granted Patent B2

US 12,744,124 · App. 18/732,380 · Granted Sep 22, 2026

Inventors: Evangelos Hytopoulos (San Mateo, CA); Justin Cambra (Oakland, CA)

Assignee: iRhythm Technologies, Inc.

G16H50/20G16H10/60G16H50/30G16H50/70

Abstract

Claims (32)

1 . A computing system for inferring an onset of cardiac arrythmia of a user, the computing system comprising:

a receiver configured to receive cardiac signals of a user detected by a sensor applied to the user; and

a hardware processor configured to apply the cardiac signals to a first machine learning model, wherein the first machine learning model is configured to infer an onset of cardiac arrhythmia based on the cardiac signals of the user, and wherein the first machine learning model is generated using a machine learning training process comprising:

accessing historical patient data for a plurality of patients, wherein the historical patient data comprises historical cardiac signal data for the plurality of patients;

accessing a status of cardiac arrhythmia for each of the plurality of patients, wherein the status of cardiac arrhythmia for the plurality of patients is determined subsequent to a time when the historical cardiac signal data were recorded for each patient of the plurality of patients; and

training the first machine learning model based at least in part on the historical cardiac signal data for the plurality of patients and the status of cardiac arrhythmia of each patient of the plurality of patients to infer an onset of cardiac arrhythmia, wherein the user is different than the plurality of patients,

wherein the hardware processor is further configured to apply the cardiac signals of the user to a second machine learning model, wherein the second machine learning model is trained to remove artifacts from the cardiac signals of the user to obtain filtered cardiac signals, and wherein applying the cardiac signals to the first machine learning model comprises applying the filtered cardiac signals as input to the first machine learning model.

2 . The computing system of claim 1 , wherein the second machine learning model is further trained to reduce noise from the cardiac signals of the user to obtain the filtered cardiac signals.

3 . The computing system of claim 1 , wherein inferring the onset of cardiac arrhythmia comprises determining a probability of an occurrence of cardiac arrythmia at a later time.

4 . The computing system of claim 1 , wherein at least a portion of the historical patient data does not indicate cardiac arrhythmia at a time of recording the historical cardiac signal data.

5 . The computing system of claim 1 , wherein the first machine learning model is further configured to process photoplethysmography (PPG) data, heartbeat data, or heart rate acceleration data to infer the onset of cardiac arrythmia.

6 . The computing system of claim 1 , wherein the cardiac arrhythmia comprises at least one of: ventricular tachycardia, supraventricular tachycardia, ectopy, or ventricular fibrillation.

7 . The computing system of claim 1 , wherein the historical patient data further comprises the status of cardiac arrhythmia for each of the plurality of patients.

8 . The computing system of claim 1 , wherein the sensor comprises an electrode.

9 . A method of training a machine learning model to infer an onset of cardiac arrhythmia, the method comprising:

by a hardware processor configured to execute computer-implemented instructions:

accessing historical patient data for a plurality of patients, wherein the historical patient data comprises historical cardiac signal data for the plurality of patients, wherein the historical cardiac signal data comprises cardiac signals, and wherein at least a portion of the historical cardiac signal data does not indicate cardiac arrhythmia at a time of recording the cardiac signals;

accessing a status of cardiac arrhythmia for each patient of the plurality of patients, wherein the status of cardiac arrhythmia for each of the plurality of patients is determined subsequent to a time when the cardiac signals were recorded for the patient;

training a first machine learning model based on the historical cardiac signal data for the plurality of patients and the status of cardiac arrhythmia of each patient of the plurality of patients to infer an onset of cardiac arrhythmia of a user, wherein the user is not included in the plurality of patients;

applying cardiac signals of the user to a second machine learning model that is trained to remove artifacts from the cardiac signals of the user to obtain filtered cardiac signals; and

applying the filtered cardiac signals to the first machine learning model to infer the onset of cardiac arrhythmia of the user.

10 . The method of claim 9 , wherein the second machine learning model is further trained to reduce noise from the cardiac signals of the user.

11 . The method of claim 9 , wherein inferring the onset of cardiac arrhythmia comprises determining a probability of an occurrence of cardiac arrythmia within a particular time period.

12 . The method of claim 9 , wherein training the first machine learning model further comprises training the first machine learning model to infer an occurrence of hospitalization, heart failure onset, onset of a stroke, or death of the user.

13 . The method of claim 9 , further comprising training a second machine learning model to infer an occurrence of hospitalization, heart failure onset, an onset of a stroke, or death of the user, wherein an input to the second machine learning model comprises cardiac signals of a longer duration than cardiac signals provided to the first machine learning model.

14 . The method of claim 9 , further comprising determining an atrial fibrillation burden from cardiac signals of the user, wherein the atrial fibrillation burden comprises an amount of time spent in atrial fibrillation by the user during a period of time.

15 . A non-transitory computer storage medium storing computer-executable instructions that, when executed by a processor, cause the processor to perform operations comprising:

accessing historical patient data for a plurality of patients, wherein the historical patient data comprises historical cardiac signal data for the plurality of patients, wherein the historical cardiac signal data comprises cardiac signals, and wherein at least a portion of the historical cardiac signal data does not indicate cardiac arrhythmia at a time of recording the cardiac signals;

accessing a status of cardiac arrhythmia for each patient of the plurality of patients, wherein the status of cardiac arrhythmia for each of the plurality of patients is determined subsequent to a time when the cardiac signals were recorded for the patient;

training a first machine learning model based on the historical cardiac signal data for the plurality of patients and the status of cardiac arrhythmia of each patient to infer an onset of cardiac arrhythmia of a user, wherein the user is not included in the plurality of patients;

applying cardiac signals of the user to a second machine learning model that is trained to remove artifacts or reduce noise from the cardiac signals of the user to obtain filtered cardiac signals; and

applying the filtered cardiac signals to the first machine learning model to infer the onset of cardiac arrhythmia of the user.

Assignments (1)

Continuity (3)

References Cited (400)