FORM 2

THE PATENT ACT, 1970

(39 of 1970)
&
The Patent Rules, 2003
COMPLETE SPECIFICATION
[See section 10 and rule 13]

1. TITLE OF INVENTION:
Machine Learning-Based Digital Electrocardiograph Device
2. APPLICANT(S)

1. (A) Dr. Shanmugam Sundaramurthy

(B) Indian
(C) Assistant Professor, Department of Computing Technologies, SRM
Institute of Science and Technology, Kattankulathur, Chennai 603203,
Tamil Nadu, India.

2. (A) Dr. Syed Anwar Hussainy F

(B) Indian
(C) Assistant Professor, Department of CSE(AI&ML), FET-Jain University,
Bengaluru, Karnataka 562112, India.

3. (A) Dr. Sugumar R

(B) Indian
(C) Associate Professor, Department of CSE(AI&ML), FET-Jain
University, Bengaluru, Karnataka 562112, India.

4. (A) Dr. Arjun Paramarthalingam

(B) Indian
(C) Assistant Professor (Selection Grade), SF-18, Dept. of CSE,
University College of Engineering Villupuram, Kakuppam, Villupuram -
605103, Tamil Nadu, India.

5. (A) Dr. K. Ravishankar

(B) Indian
(C) Assistant Professor, School of Computing Department of Computer
science and Technology, Sathyabama institute of Science and
Technology, Chennai, Tamil Nadu, India.

3. PREAMBLE TO THE DESCRIPTION

The following specification particularly describes the invention and the manner in
which it is to be performed.

Page 1 of 17
 Field of the Invention

The present invention relates to a machine learning-based digital

electrocardiograph (ECG) device that integrates artificial intelligence (AI)
and machine learning (ML) algorithms with traditional ECG technology to
5 enhance the accuracy, speed, and diagnostic capabilities of heart disease
detection. More specifically, the invention pertains to real-time ECG signal
processing, automated cardiac anomaly detection, and AI-driven
predictive analysis, thereby improving early detection and classification of
cardiovascular conditions such as arrhythmias, ischemia, and atrial
10 fibrillation. The invention finds applications in telemedicine, wearable
health monitoring, and hospital-based cardiac diagnostics, offering
enhanced patient care, remote monitoring, and intelligent ECG
interpretation.

Background of the Invention

15 Cardiovascular diseases (CVDs) remain a leading cause of mortality

worldwide, necessitating accurate and timely detection of heart
abnormalities. Electrocardiography is a widely used, non-invasive
diagnostic tool that records the electrical activity of the heart. Traditional
ECG machines consist of electrodes placed on the skin, which capture
20 electrical impulses generated by the heart. These signals are then
amplified, filtered, and displayed on a monitor for manual interpretation by
medical professionals.

Existing ECG systems can be broadly categorized into the following types:

1. Conventional ECG Machines – Standalone hospital-grade devices used

25 in clinical settings that require trained professionals to interpret results.

2. Portable and Wearable ECG Devices – Compact devices designed for

remote monitoring, often integrated into smartwatches or handheld ECG
recorders.

Page 2 of 17
 3. Cloud-Based ECG Solutions – Systems that store ECG data in the
cloud for remote access and diagnosis by healthcare providers.

While these traditional solutions are widely used, they suffer from several
limitations that impact their effectiveness in real-world scenarios.

5 3. Limitations of Prior Art

3.1. Dependency on Manual Interpretation

Traditional ECG systems rely on manual analysis by cardiologists or

trained technicians. Human interpretation is prone to errors, especially
when diagnosing subtle ECG abnormalities. Misdiagnosis or missed
10 detections due to fatigue, subjectivity, or lack of immediate expertise can
lead to delayed or incorrect treatment.

3.2. Limited Real-Time Diagnostic Capability

Most conventional ECG devices record heart activity but do not have real-
time diagnostic intelligence. The lack of instant anomaly detection and
15 classification makes it difficult to identify life-threatening conditions such as
sudden arrhythmias, myocardial infarctions, or silent ischemia before
symptoms become severe.

3.3. Inability to Detect Complex Cardiac Patterns

Traditional ECG machines operate on rule-based algorithms that identify

20 abnormalities using predefined thresholds. However, such predefined
criteria are not always reliable, as heart diseases often manifest in subtle
variations rather than clear-cut abnormal waveforms. Machine learning-
based models can learn from large datasets and identify complex patterns
that rule-based systems might miss.

25 3.4. High False Positives and False Negatives

Page 3 of 17
 Existing automated ECG interpretation systems often produce high false
positives and false negatives, leading to unnecessary panic or missed
diagnoses. False positives cause unnecessary hospital visits and medical
costs, whereas false negatives may result in undiagnosed life-threatening
5 conditions.

3.5. Lack of Personalized ECG Analysis

Traditional ECG systems apply generalized diagnostic criteria to all

patients, ignoring individual variability in heart rhythms due to age,
genetics, lifestyle, or medical history. A one-size-fits-all approach limits
10 diagnostic accuracy, whereas an AI-driven ECG device can adapt to
individual patients by learning their unique cardiac profile.

3.6. Limited Remote Monitoring and Integration

While wearable ECG devices provide some level of remote monitoring,

they often have low accuracy and limited diagnostic intelligence.
15 Furthermore, many conventional ECG systems lack seamless integration
with digital health records, cloud storage, or mobile applications, making it
difficult for physicians to access and analyze historical ECG trends.

3.7. Poor Accessibility in Remote and Rural Areas

Many parts of the world, especially rural and underserved areas, lack
20 access to trained cardiologists. Traditional ECG machines are bulky and
require specialized infrastructure, making them impractical for low-
resource settings. AI-powered ECG interpretation could enable non-
specialists to diagnose cardiac conditions with higher confidence,
improving access to timely healthcare.

25 Need for an Improved Solution

Given these challenges, there is a critical need for an advanced ECG

device that integrates machine learning algorithms to:

Page 4 of 17
 Enhance diagnostic accuracy by learning from large datasets and
identifying complex cardiac conditions that traditional methods may miss.

Provide real-time analysis and alerts, enabling immediate intervention in

case of life-threatening conditions.

5 Reduce dependency on expert interpretation, making ECG analysis more

accessible and efficient, especially in remote or resource-limited settings.

Personalize ECG analysis by considering patient-specific characteristics,

thereby reducing false positives and false negatives.

Enable seamless remote monitoring and integration with mobile

10 applications, cloud storage, and telemedicine platforms for better patient
management.

Support edge AI processing, allowing real-time ECG interpretation without

requiring constant internet connectivity.

The present invention aims to address these issues by combining

15 traditional ECG signal acquisition with AI-driven analytics, providing a
smart, real-time, and automated ECG device capable of detecting
cardiovascular diseases with greater accuracy, speed, and accessibility.

Object of the Invention

The primary objective of the present invention is to develop a machine

20 learning-based digital electrocardiograph (ECG) device that enhances the
accuracy, speed, and diagnostic capability of heart disease detection. The
invention integrates artificial intelligence (AI) and machine learning (ML)
algorithms with traditional ECG technology to enable real-time analysis,
automated classification of heart conditions, and early detection of
25 cardiovascular diseases such as arrhythmias, ischemia, and atrial
fibrillation. This innovation aims to reduce dependency on manual

Page 5 of 17
 interpretation, minimize diagnostic errors, and improve patient outcomes
through intelligent cardiac monitoring.

Other Objectives:

- To provide a real-time ECG analysis system that automatically detects

5 and classifies heart abnormalities using a trained ML model, allowing for
instant feedback and immediate medical intervention.

- To reduce false positives and false negatives in ECG interpretation by

leveraging AI-driven algorithms that continuously learn from large datasets
and adapt to individual patient profiles.

10 - To integrate signal processing techniques that filter noise, correct

baseline drift, and extract key ECG features such as P-wave, QRS
complex, and T-wave morphology for more precise diagnostics.

- To enable personalized ECG monitoring, wherein the system adjusts its

analysis based on patient-specific factors such as age, medical history,
15 and lifestyle.

- To develop a wireless and portable ECG device that supports Bluetooth,

Wi-Fi, or 5G connectivity, enabling seamless data transmission to mobile
applications, cloud storage, or hospital information systems.

- To enhance remote patient monitoring and telemedicine applications by

20 allowing healthcare professionals to access ECG data from any location
and provide timely diagnosis.

- To incorporate edge AI processing, ensuring that real-time ECG analysis

and classification can occur on the device itself, without the need for
constant internet connectivity.

25 - To improve accessibility in remote and rural areas by providing an AI-

powered ECG system that can assist non-specialists in diagnosing cardiac
conditions with higher confidence.

Page 6 of 17
 - To develop an adaptive and retrainable machine learning model that
continuously updates itself based on newly acquired ECG data, improving
diagnostic accuracy over time.

- To integrate alerts and predictive analytics, where the system generates

5 early warnings for potential cardiac risks based on ECG trends, helping in
preventive healthcare.

By achieving these objectives, the invention aims to revolutionize cardiac

diagnostics, making ECG analysis faster, more accurate, and widely
accessible while leveraging AI for improved healthcare outcomes.

10 Summary of the Invention

The present invention relates to a machine learning-based digital

electrocardiograph (ECG) device that integrates artificial intelligence (AI)
and machine learning (ML) algorithms with traditional ECG technology to
enhance the accuracy, speed, and diagnostic capability of heart disease
15 detection. The device is designed to automatically analyze ECG signals in
real time, providing instant classification of cardiac conditions such as
arrhythmias, ischemia, atrial fibrillation, and other cardiovascular
abnormalities.

The invention comprises ECG electrodes for capturing cardiac signals, a

20 signal acquisition module for noise filtering and amplification, a processing
unit with an ML-based diagnostic module, and a wireless communication
interface for remote data transmission. The machine learning module is
trained on large ECG datasets and employs deep learning architectures,
including convolutional neural networks (CNNs) and recurrent neural
25 networks (RNNs), to detect complex cardiac patterns with higher accuracy
than conventional methods.

The system supports real-time analysis, risk prediction, and adaptive

learning, enabling personalized ECG monitoring by adjusting to patient-

Page 7 of 17
 specific characteristics. The device is further equipped with edge AI
capabilities, allowing real-time inference without the need for continuous
internet connectivity. Additionally, it provides seamless integration with
mobile applications, cloud storage, and telemedicine platforms, making it
5 suitable for hospitals, home healthcare, and remote monitoring
applications.

By leveraging AI-driven automation, this invention reduces dependency on

manual interpretation, minimizes false positives and false negatives, and
enhances early detection of cardiac conditions, ultimately improving
10 patient outcomes and accessibility to advanced cardiac diagnostics.

Detailed Description

The present invention discloses a machine learning-based digital

electrocardiograph (ECG) device that integrates artificial intelligence (AI)
and machine learning (ML) algorithms with traditional ECG technology to
15 enhance real-time cardiac monitoring, diagnostic accuracy, and early
detection of cardiovascular diseases. The system comprises multiple
interconnected components, including ECG electrodes, a signal
acquisition module, a processing unit with an ML-based diagnostic
module, a display interface, and a wireless communication module. These
20 components work synergistically to analyze ECG signals, classify cardiac
conditions, and provide actionable insights to medical professionals and
users.

System Components and Arrangement

1. ECG Electrode Unit

25 The system employs a set of dry or gel-based ECG electrodes placed on

the patient's skin to capture electrical signals generated by the heart.
These electrodes detect voltage fluctuations corresponding to the heart’s
rhythm and transmit the raw signals to the signal acquisition module.

Page 8 of 17
 2. Signal Acquisition Module

This module includes:

- Amplifiers: To boost the low-magnitude ECG signals for accurate

processing.

5 - Analog-to-Digital Converter (ADC): Converts the amplified analog signals

into digital format for further analysis.

- Noise Filtering Unit: Removes motion artifacts, baseline drift, and high-
frequency noise using digital filters (e.g., Butterworth or wavelet filters).

3. Processing Unit with Machine Learning Module

10 The processing unit consists of a microcontroller or an embedded AI chip

equipped with a trained machine learning model. The module performs the
following operations:

- Feature Extraction: Identifies key ECG characteristics, including P-wave,

QRS complex, T-wave, and RR intervals.

15 - Classification Algorithm: Uses deep learning models such as

Convolutional Neural Networks (CNNs) and Recurrent Neural Networks
(RNNs) to classify heart conditions (e.g., arrhythmia, atrial fibrillation,
myocardial infarction).

- Personalized ECG Analysis: Adapts to patient-specific variations using

20 reinforcement learning.

- Risk Prediction: Generates alerts if the model detects potential cardiac

abnormalities.

4. Display and User Interface

Page 9 of 17
 The system is integrated with an LCD screen, smartphone application, or
web dashboard to display real-time ECG waveforms, diagnostic results,
and historical trends. It provides:

- Visual ECG Graphs

5 - Automatic Interpretation & Alerts

- User-Friendly Navigation for Patients and Clinicians

5. Wireless Communication Module

The device supports Bluetooth, Wi-Fi, and 5G connectivity for seamless

transmission of ECG data to:

10 - Cloud Storage for Remote Monitoring

- Hospital Information Systems (HIS)

- Wearable Health Devices

6. Edge AI Processing

To enable real-time and offline ECG analysis, the processing unit supports
15 Edge AI technology, allowing:

- On-Device Inference Without Internet Dependence

- Faster Decision-Making in Emergency Cases

Method of Operation

Step 1: ECG Signal Acquisition

20 - The user places the ECG electrodes on predefined body locations.

- The electrodes capture electrical activity and transmit the signals to the
acquisition module.

Step 2: Signal Preprocessing and Noise Filtering

Page 10 of 17
 - The system amplifies and digitizes the ECG signals.

- A noise filtering algorithm removes unwanted signal distortions.

Step 3: Feature Extraction

- The processing unit extracts essential parameters such as heart rate

5 variability, QRS complex width, and ST-segment elevation.

Step 4: Machine Learning-Based Classification

- The ML model compares extracted features with a trained dataset.

- The system categorizes the ECG waveform as Normal, Arrhythmia, or

High-Risk.

10 Step 5: Diagnosis and Risk Prediction

- If abnormalities are detected, the system issues alerts via mobile

notifications or cloud-based platforms.

- Predictive analytics estimates the likelihood of future heart complications.

Step 6: Data Storage and Remote Monitoring

15 - The ECG recordings and analysis results are stored locally and
transmitted to cloud servers.

- Healthcare professionals can access real-time and historical ECG data

for further analysis.

Example Embodiments

20

Example 1: Real-Time ECG Monitoring in a Hospital Setting

A patient in a cardiac care unit is continuously monitored using the device.

The real-time ML-based analysis detects an early-stage arrhythmia,

Page 11 of 17
 prompting an immediate alert to doctors. The system suggests possible
interventions based on past ECG patterns, allowing timely medical
intervention.

Example 2: Remote ECG Monitoring for At-Home Patients

5 An elderly patient with a history of atrial fibrillation wears a portable ECG

patch embedded with the invention’s technology. The ML model detects
an abnormal heartbeat and sends an alert to a telemedicine platform,
enabling the physician to recommend an emergency check-up.

Example 3: Integration with Smartwatches and Mobile Health Apps

10 The invention is embedded in a smartwatch, allowing users to self-monitor

their heart health. The system provides daily ECG assessments, flags
irregularities, and syncs data with healthcare providers via a mobile app.

Advantages of the Invention

- Higher Diagnostic Accuracy: AI-driven ECG interpretation minimizes

15 errors compared to manual diagnosis.

- Early Detection & Prevention: Real-time predictive analysis helps prevent

severe cardiac events.

- Personalized Monitoring: The system adapts to individual ECG patterns,

improving diagnostic reliability.

20 - Remote Accessibility: Enables telemedicine applications and real-time

cardiac surveillance.

- Seamless Integration: Compatible with cloud storage, wearable devices,

and hospital networks.

Page 12 of 17
 The present invention revolutionizes ECG technology by integrating AI-
based diagnostics, making cardiac monitoring smarter, faster, and more
accessible.

10

15

20

Page 13 of 17
 Claims

We Claim:

1. A machine learning-based digital electrocardiograph (ECG) device,

comprising:

5 a set of ECG electrodes configured to capture electrical activity of a

patient's heart;

a signal acquisition module for filtering and amplifying ECG signals;

a processing unit including a machine learning (ML) module,

wherein the ML module is trained on ECG datasets to classify

10 abnormal and normal heart rhythms;

a real-time analytics engine for detecting arrhythmias and

cardiovascular conditions based on ECG patterns; and

a wireless communication interface to transmit analyzed data to a

remote device or healthcare system.

15 2. A method for real-time cardiac diagnosis using a machine learning-

based digital electrocardiograph device, the method comprising:

acquiring raw ECG signals via multiple electrodes;

preprocessing the ECG signals by noise filtering and baseline

correction;

Page 14 of 17
 extracting key ECG features including P-wave, QRS complex, and

T-wave morphology;

applying a trained ML algorithm to classify heart conditions such as

arrhythmias, ischemia, or atrial fibrillation; and

5 displaying real-time results on a user interface, alerting the user in

case of abnormalities.

3. A machine learning-based ECG device, wherein the ML module is

configured to:

continuously update and retrain its model using new ECG data;

10 employ a deep learning architecture, including convolutional neural

networks (CNNs) and recurrent neural networks (RNNs), for

detecting complex cardiac abnormalities; and

generate risk scores and predictive alerts for potential heart

conditions.

15 4. The device of claim 1, wherein the ECG electrodes include dry

electrodes or textile-based electrodes, enabling continuous, non-invasive

monitoring.

5. The device of claim 1, further comprising a cloud-based storage and

analytics system, wherein ECG recordings and ML-based diagnostics are

20 stored for further analysis and remote consultation.

Page 15 of 17
 6. The method of claim 2, wherein the ML model adapts based on

individual patient ECG variations, reducing false positives and improving

diagnostic accuracy.

7. The device of claim 1, wherein the wireless communication interface

5 supports Bluetooth, Wi-Fi, or 5G connectivity, enabling real-time ECG data

streaming to mobile applications or hospital databases.

8. The device of claim 3, wherein the ML module integrates with an edge

AI processor, allowing on-device inference without requiring continuous

internet access.

Dated this 25th day of February 2025

10 -- E-Signed --

KIRTIKUMAR M. PATEL (IN/PA/2999)

Applicant’s Patent Agent

Page 16 of 17
ABSTRACT

The present invention relates to a machine learning-based digital

electrocardiograph (ECG) device that integrates artificial intelligence (AI)
and machine learning (ML) algorithms to enhance real-time cardiac
monitoring, diagnostic accuracy, and early detection of cardiovascular
diseases. The device comprises ECG electrodes, a signal acquisition
module, a processing unit with an ML-based diagnostic module, and a
wireless communication interface. The ML module, trained on extensive
ECG datasets, detects arrhythmias, ischemia, and atrial fibrillation with
higher precision than conventional methods. The system supports real-
time analysis, predictive alerts, and adaptive learning, enabling
personalized patient monitoring. Additionally, it features edge AI
processing for offline operation and integrates with mobile applications
and cloud platforms for remote healthcare access. This invention reduces
manual interpretation errors and enhances early disease detection and
patient outcomes.

Page 17 of 17