Smart-ECG: Smart ECG Analysis

Objective: Our objective is to automatically interpret the most common and difficult-to-detect 20 types of arrhythmias and four anomalies with the help of artificial intelligence and then present this interpretation to a physician for confirmation.

The artificial intelligence software is designed to work independently of the ECG device and can process, convert, and interpret different input formats. This ensures that patient health is protected and hospital services are not limited by human error or skills. Additionally, it allows for remote diagnosis and effective monitoring.

Method: It is a multi-classification approach that can work with the raw form of data obtained from the ECG device, either with QRS waves or directly with Lead II.

Machine learning methods like RF and MLP and deep learning models like LSTM and CNN are covered.

Ensembling is done to increase in-class accuracy.

Innovative Features

• Artificial intelligence service capable of device-independent ECG Analysis and interpretation
• Ability to extract QRS component and other heart rate parameters from raw data so that model input parameters, essential in device independence, are uniform and single source.
• A platform where cardiac anomalies and arrhythmias can be added easily will be put into service.
• Storing and querying ECG Analysis comments
• Combined use of artificial intelligence methods (ensembling) to provide the best modeling.
• A process model suitable for ECG interpretations, all of which need to be done with remote data transfer, such as telemedicine
• Individual ECG Analysis, including the medical information of the patient’s current condition.