Patient specific ECG classification using machine learning

In recent years, considerable research has been done to assist cardiologists with their task of diagnosing the ECG recordings. The ?elds of research range from novelty detectors to fully automated ECG-diagnosing systems. A wide range of techniques have been used for the purpose that include statisti

Project Title

Patient specific ECG classification using machine learning

Project Area of Specialization

Artificial Intelligence

Project Summary

In recent years, considerable research has been done to assist cardiologists with their task of diagnosing the ECG recordings. The ?elds of research range from novelty detectors to fully automated ECG-diagnosing systems. A wide range of techniques have been used for the purpose that include statistical pattern recognition, expert systems, arti?cial neural networks (ANN), wavelet transform, and fuzzy and neuro-fuzzy systems.

The morphological diagnosis of the ECG is a pattern recognition task. The ECG interpretation is comprised of two distinct and sequential phases: feature extraction and classi?cation. A set of signal measurements containing information for the characterization of the waveform are obtained by shapeidenti?cation methods. These waveform descriptors are used to allocate the ECG to one or more diagnostic classes in the classi?cation phase. These classi?ers may be heuristic and use rules of-thumb or employ syntactic or fuzzy logic as reasoning tools. The classi?er may also be statistical with the use of complex and even abstract-signal features as waveform descriptors and different discriminate function models for class allocation.

Project Objectives

• To design a Real-Time Patient-Specific ECG Classification Using Machine Learning
• To investigate and validate the accuracy of different ML classifiers
• To develop an embedded recommendation instrument for cardio specialist   

Project Implementation Method

In this project, we proposed a patient-specific ECG heartbeat classifier with an adaptive implementation of 1D Convolutional Neural Networks (CNNs) that are able to fuse the two major blocks of the traditional ECG classification into a single learning body: feature extraction and classification.

Such a compact implementation for each patient over a simple CNN not only negates the necessity to extract hand-crafted manual features, or any kind of pre- and post-processing, also makes it a primary choice for a real-time implementation for
heart monitoring and anomaly detection. Besides the speed and computational efficiency achieved, the proposed method only requires 1D convolutions (multiplications and additions) that make any hardware implementation simpler and cheaper.

In addition to that once a dedicated CNN is trained for an individual patient, it can solely be used to classify his/her long ECG records such as Holter registers in a fast and accurate manner.

Benefits of the Project

The proposed project demonstrate a fast and accurate patient-specific electrocardiogram (ECG) classification and monitoring system. Methods: An adaptive implementation of Convolutional Neural Networks (CNNs) is inherently used to fuse the two major blocks of the ECG classification into a single learning body: feature extraction and classification. Therefore, for each patient an individual and simple CNN will be trained by using relatively small common and patient-specific training data, and thus such a patient-specific feature extraction ability can further improve the classification performance. 

Technical Details of Final Deliverable

• Software Simulation results
• Comparative Study
• HW/SW integrated system

Final Deliverable of the Project

HW/SW integrated system

Core Industry

Medical

Other Industries

IT , Health

Core Technology

Artificial Intelligence(AI)

Other Technologies

Clean Tech

Sustainable Development Goals

Good Health and Well-Being for People

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com