Electrocardiogram wave segmentation using machine learning

Inventor: Benjamin A. Teplitzky (Rochester, MN)

Assignee: BOSTON SCIENTIFIC CARDIAC DIAGNOSTICS, INC.

H04N21/426 A61B5/349 A61B5/366 H04N21/42676 H04N21/6118

View Patent ↗

Loading inventors, assignments & file history…

Monitor This Case

Get email alerts when status or documents change.

Order Certified Copies

Most orders are placed with the USPTO same day — all within 24 business hours.

Order via The Patent Place →

Pre-filled with this patent's details

Quick Facts

Patent No.: US 12701286
App. No.: 17/974,025
Granted: Aug 4, 2026
Kind: B2

External Links

USPTO Patent Center↗ Full Patent PDF↗

Abstract

A method includes analyzing a first segment of electrocardiogram measurements corresponding to a first heartbeat and a second heartbeat of a patient to identify a set of features of the first segment of electrocardiogram measurements. The method further includes adjusting the set of features based at least on the first segment and a second segment of electrocardiogram measurements corresponding to the first heartbeat. The method further includes predicting, based on the adjusted set of features, a region in the second segment of electrocardiogram measurements comprising a p-wave generated by the second heartbeat.

Claims (36)

1 . A method comprising:

dividing electrocardiogram (ECG) measurements of a patient into a plurality of segments comprising a first segment of the ECG measurements corresponding to a first heartbeat of the patient and a second segment of the ECG measurements corresponding to the first heartbeat and a second heartbeat of the patient;

analyzing the second segment of the ECG measurements to identify a set of features of the second segment of the ECG measurements, wherein the set of features comprise of portions of heartbeats;

adjusting a duration of at least one feature of the set of features based at least on the first segment of the ECG measurements and the second segment of the ECG measurements;

determining, based on the adjusted duration of the at least one feature of the set of features, a region in the second segment of the ECG measurements comprising a p-wave generated by the second heartbeat; and

determining, based on the determined region in the second segment, whether a cardiac abnormality is occurring in a patient.

2 . The method of claim 1 , further comprising determining, based on the adjusted duration of the at least one feature of the set of features, regions in the second segment of the ECG measurements comprising a t-wave generated by the second heartbeat and a QRS complex generated by the second heartbeat.

3 . The method of claim 1 , wherein the region in the second segment of the ECG measurements comprises both the p-wave and a t-wave generated by the first heartbeat.

4 . The method of claim 1 , wherein heartbeats of the first segment of the ECG measurements and heartbeats of the second segment of the ECG measurements are of a same duration.

5 . The method of claim 1 , wherein the analyzing the second segment of the ECG measurements to identify the set of features is performed using a deep neural network to identify changes with larger magnitudes as the set of features within the second segment than magnitudes of a corresponding set of features within the first segment.

6 . The method of claim 1 , wherein the adjusting the durations of the at least one feature of the set of features is performed using a long short-term memory circuit to adjust the duration of the at least one feature such that the at least one feature aligns with the p-wave in the second segment.

7 . The method of claim 1 , further comprising determining a treatment for determined cardiac abnormality occurring in the patient.

8 . The method of claim 1 , wherein determining the region comprising the p-wave comprises determining a start time of the p-wave and an end time of the p-wave.

9 . The method of claim 1 , wherein the second heartbeat occurs after the first heartbeat.

10 . An apparatus comprising:

a memory; and

a hardware processor communicatively coupled to the memory and configured to:

divide electrocardiogram (ECG) measurements of a patient into a plurality of segments comprising a first segment of the ECG measurements corresponding to a first heartbeat of the patient and a second segment of the ECG measurements corresponding to the first heartbeat and a second heartbeat of the patient,

analyze the second segment of the ECG measurements to identify a set of features of the second segment of the ECG measurements, wherein the set of features comprise of portions of heartbeats,

adjust a duration of at least one feature of the set of features based at least on the first segment of the ECG measurements and the second segment of the ECG measurements,

determine, based on the adjusted duration of the at least one feature of the set of features, a region in the second segment of the ECG measurements comprising a p-wave generated by the second heartbeat; and

determine, based on the determined region in the second segment, whether a cardiac abnormality is occurring in a patient.

11 . The apparatus of claim 10 , wherein the hardware processor is further configured to determine, based on the adjusted duration of the at least one feature of the set of features, regions in the second segment of the ECG measurements comprising a t-wave generated by the second heartbeat and a QRS complex generated by the second heartbeat.

12 . The apparatus of claim 10 , wherein the region in the second segment of the ECG measurements comprises both the p-wave and a t-wave generated by the first heartbeat.

13 . The apparatus of claim 10 , wherein heartbeats of the first segment of the ECG measurements and heartbeats of the second segment of the ECG measurements are of a same duration.

14 . The apparatus of claim 10 , wherein the analyze the second segment of the ECG measurements to identify the set of features is performed using a deep neural network to identify changes with larger magnitudes as the set of features within the second segment than magnitudes of a corresponding set of features within the first segment.

15 . The apparatus of claim 10 , wherein adjusting the duration of the at least one feature of the set of features is performed using a long short-term memory circuit to adjust the duration of the at least one feature such that the at least one feature aligns with the p-wave in the second segment.

16 . The apparatus of claim 10 , wherein the hardware processor is further configured to determine a treatment for determined cardiac abnormality occurring in the patient.

17 . The apparatus of claim 10 , wherein the determining the region comprising the p-wave comprises determining a start time of the p-wave and an end time of the p-wave.

18 . The apparatus of claim 10 , wherein the second heartbeat occurs after the first heartbeat.

19 . A method comprising:

analyzing a first segment of electrocardiogram (ECG) measurements corresponding to a first heartbeat and a second heartbeat of a patient to identify a set of features of the first segment of ECG measurements, wherein the set of features comprise of portions of heartbeats;

adjusting a duration of at least one feature of the set of features based at least on the first segment and a second segment of ECG measurements corresponding to the first heartbeat;

determining, based on the adjusted duration of the at least one feature of the set of features, a region in the second segment of ECG measurements comprising a p-wave generated by the second heartbeat; and

determining, based on the determined region in the second segment, whether a cardiac abnormality is occurring in a patient.

20 . The method of claim 19 , further comprising determining, based on the adjusted duration of the at least one feature of the set of features, regions in the second segment of ECG measurements comprising a t-wave generated by the second heartbeat and a QRS complex generated by the second heartbeat.