# US 2022/0015711 A1 · System and method for automated analysis and detection of cardiac arrhythmias from electrocardiograms

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2022/0015711 A1
- Document ID: US20220015711A1
- Status: Published application
- Kind: A1
- Application number: 17/380,765
- Filed: 2021-07-20
- Published: 2022-01-20
- Earliest claimed: 2020-07-20
- Assignees: Board of Regents, The University of Texas System
- Applicants: Board of Regents, The University of Texas System
- Inventors: Vignesh Kalidas; Lakshman S. Tamil
- Web page: https://patentgrep.com/patent/US20220015711A1

## CPC

- A61B 5/7203: {for noise prevention, reduction or removal}
- A61B 5/0006: {ECG or EEG signals}
- A61B 5/0022: {Monitoring a patient using a global network, e.g. telephone networks, internet}
- A61B 5/352: Detecting R peaks, e.g. for synchronising diagnostic apparatus; Estimating R-R interval
- A61B 5/361: Detecting fibrillation
- A61B 5/364: Detecting abnormal ECG interval, e.g. extrasystoles, ectopic heartbeats
- A61B 5/366: Detecting abnormal QRS complex, e.g. widening
- A61B 5/7207: {of noise induced by motion artifacts}
- A61B 5/725: {using specific filters therefor, e.g. Kalman or adaptive filters (specific diagnostics methods using using bioelectric or biomagnetic signals A61B5/316)}
- A61B 5/7267: {involving training the classification device}
- G06N 3/045: Combinations of networks
- G06N 3/0455: Auto-encoder networks; Encoder-decoder networks
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/084: Backpropagation, e.g. using gradient descent
- G06N 3/088: Non-supervised learning, e.g. competitive learning
- G06N 3/0895: Weakly supervised learning, e.g. semi-supervised or self-supervised learning
- G06N 3/09: Supervised learning
- G06N 5/01: Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
- G06N 7/01: Probabilistic graphical models, e.g. probabilistic networks
- G06N 20/20: Ensemble learning
- G16H 15/00: ICT specially adapted for medical reports, e.g. generation or transmission thereof
- G16H 40/63: for local operation
- G16H 40/67: for remote operation
- G16H 50/20: for computer-aided diagnosis, e.g. based on medical expert systems
- G16H 50/30: for calculating health indices; for individual health risk assessment
- G16H 50/70: for mining of medical data, e.g. analysing previous cases of other patients
- A61B 5/363: Detecting tachycardia or bradycardia
- A61B 5/726: {using Wavelet transforms}
- G06N 3/048: Activation functions
- A61B: DIAGNOSIS; SURGERY; IDENTIFICATION (usefulness limited to only animals A61D)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G16H: HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA

## Abstract

The present disclosure presents arrhythmia analysis systems and methods. One such method comprises processing an acquired ECG waveform signal to remove noise artifacts and form a denoised ECG waveform signal; processing the denoised ECG waveform signal to remove low quality segments and form a high quality ECG waveform signal; analyzing the high quality ECG waveform to detect a presence of a beat-independent ventricular arrhythmia; processing the denoised ECG signal to extract beat (R-peak) locations corresponding to QRS complexes from the denoised ECG signal; analyzing the denoised ECG signal to detect a presence of a beat-dependent ventricular arrhythmia based on the extracted beat (R-peak) locations; analyzing the denoised ECG signal to detect a presence of one or more supraventricular arrhythmias based on the extracted beat locations of the ECG signal; and outputting a report containing one or more arrhythmias detected by the analyzing steps. Other methods/systems are also provided.

## Claims

### Claim 1

An arrhythmia analysis method comprising:
acquiring, by at least one computing device, an electrocardiogram (ECG) waveform signal of a subject at a set sampling frequency rate;
processing, by the at least one computing device, the acquired ECG waveform signal to remove low frequency noise and high frequency noise artifacts and form a denoised ECG waveform signal;
processing, by the at least one computing device, the denoised ECG waveform signal to remove low quality segments and form a high quality ECG waveform signal;
analyzing, by the at least one computing device, the high quality ECG waveform to detect a presence of a beat-independent ventricular arrhythmia within the high quality ECG signal;
processing, by the at least one computing device, the denoised ECG signal to extract beat (R-peak) locations corresponding to QRS complexes from the denoised ECG signal;
analyzing, by the at least one computing device, the denoised ECG signal to detect a presence of a beat-dependent ventricular arrhythmia within the denoised ECG signal based on the extracted beat (R-peak) locations of the denoised ECG signal;
analyzing, by the at least one computing device, the denoised ECG signal to detect a presence of one or more supraventricular arrhythmias within the denoised ECG signal based on the extracted beat locations of the ECG signal; and
outputting, by the at least one computing device, a report containing one or more arrhythmias detected by the analyzing steps.

### Claim 2 (depends on 1)

The method of claim 1, wherein the low frequency noise is removed by application of stationary wavelet transforms to the acquired ECG signal.

### Claim 3 (depends on 1)

The method of claim 1, wherein the high frequency noise is removed by application of Convolutional Autoencoder techniques.

### Claim 4 (depends on 1)

The method of claim 1, wherein the low quality segments are corrupted by electrode motion artifacts and are removed by application of a convolutional neural network based binary classifier.

### Claim 5 (depends on 1)

The method of claim 1, wherein beat-independent ventricular arrhythmia is analyzed using random forest classifier machine learning techniques.

### Claim 6 (depends on 1)

The method of claim 1, wherein a convolutional autoencoder is trained to identify potential QRS complexes and an adaptive thresholding is applied to the potential QRS complexes to accurately identify the beat locations corresponding to the QRS complexes present in the denoised ECG signal.

### Claim 7 (depends on 1)

The method of claim 1, wherein the denoised ECG signal is processed to perform premature ventricular complexes (PVC) classifications of the denoised ECG signal using semi-supervised autoencoder and random forest machine learning techniques.

### Claim 8 (depends on 7)

The method of claim 7, wherein the presence of one or more beat-dependent ventricular arrhythmias is detected using logical analysis of patterns of the PVC classifications of the denoised ECG signal.

### Claim 9 (depends on 1)

The method of claim 1, wherein a presence of an atrial fibrillation supraventricular arrhythmia is detected using machine learning employing 8-state Markov models and random forest classifier techniques.

### Claim 10 (depends on 1)

The method of claim 1, wherein the analysis of the denoised ECG signal to detect the presence of one or more supraventricular arrhythmias comprises performing supraventricular ectopic beats (SVEBs) classifications of the denoised ECG signal using semi-supervised autoencoder and random forest machine learning techniques.

### Claim 11 (depends on 10)

The method of claim 10, wherein the presence of one or more supraventricular arrhythmia is detected using logical analysis of patterns of SVEB beats within the denoised ECG signal.

### Claim 12 (depends on 1)

The method of claim 1, wherein the at least computing device is configured to detect one or more beat-independent ventricular arrhythmias including sustained ventricular tachycardia, ventricular fibrillation, and ventricular flutter.

### Claim 13 (depends on 1)

The method of claim 1, wherein the at least computing device is configured to detect one or more beat-dependent ventricular arrhythmias including ventricular bigeminy, ventricular trigeminy, ventricular quadrigeminy, ventricular runs and ventricular couplets.

### Claim 14 (depends on 1)

The method of claim 1, wherein the at least computing device is configured to detect one or more supraventricular arrhythmias including supraventricular bigeminy, supraventricular trigeminy, supraventricular quadrigeminy, supraventricular runs, supraventricular couplets, and sinus bradycardia.

### Claim 15 (depends on 1)

The method of claim 1, further comprising:
receiving, by the at least one computing device, electrocardiogram data representing an electrocardiogram waveform signal of a subject at first sampling frequency rate; and
resampling, by the at least one computing device, the electrocardiogram waveform signal at the first sampling frequency rate to the set sampling frequency rate to form the acquired electrocardiogram waveform signal.

### Claim 16

An arrhythmia analysis system comprising:
a computing device having a memory and a processor, wherein the processor is configured to:
acquire an electrocardiogram (ECG) waveform signal of a subject at a set sampling frequency rate;
process the acquired ECG waveform signal to remove low frequency noise and high frequency noise artifacts and form a denoised ECG waveform signal;
process the denoised ECG waveform signal to remove low quality segments corrupted by electrode motion artifacts and form a high quality ECG waveform signal;
analyze the high quality ECG waveform to detect a presence of a beat-independent ventricular arrhythmia within the high quality ECG signal;
process the denoised ECG signal to extract beat (R-peak) locations corresponding to QRS complexes from the denoised ECG signal;
analyze the denoised ECG signal to detect a presence of a beat-dependent ventricular arrhythmia within the denoised ECG signal based on the extracted beat (R-peak) locations of the denoised ECG signal;
analyze the denoised ECG signal to detect a presence of one or more supraventricular arrhythmias within the denoised ECG signal based on the extracted beat locations of the ECG signal; and
output a report containing one or more arrhythmias detected by the analyzing steps.

### Claim 17 (depends on 16)

The system of claim 16, further comprising an ECG machine that is configured to sense the ECG waveform signal.

### Claim 18 (depends on 16)

The system of claim 16, wherein the computing device performs the analyzing steps and delivers the output report as hosted services over the Internet.

### Claim 19 (depends on 16)

The system of claim 16, wherein the ECG waveform signal comprises a single-lead ECG waveform signal.

### Claim 20 (depends on 16)

The system of claim 16, wherein the at least computing device is configured to detect one or more beat-independent ventricular arrhythmias including sustained ventricular tachycardia, ventricular fibrillation, and ventricular flutter, wherein the at least computing device is configured to detect one or more beat-dependent ventricular arrhythmias including ventricular bigeminy, ventricular trigeminy, ventricular quadrigeminy, ventricular runs, and ventricular couplets, and wherein the at least computing device is configured to detect one or more supraventricular arrhythmias including atrial fibrillation, supraventricular bigeminy, supraventricular trigeminy, supraventricular quadrigeminy, supraventricular runs, supraventricular couplets, and sinus bradycardia.

### Claim 21 (depends on 16)

The system of claim 16, further comprising a client computing device that is configured to transmit the electrocardiogram (ECG) waveform signal of the subject to the computing device and is configured to receive the report from the client computing device, wherein the client computing devices comprises a smart phone or a smart watch.
