Wearable ECG Device and Machine Learning for Heart Monitoring

Area of Science:

• Biomedical Engineering
• Cardiology
• Artificial Intelligence

Background:

• Cardiovascular diseases (CVD) are a major cause of death, increasing the need for effective monitoring solutions.
• Wearable devices offer a promising avenue for continuous cardiac activity assessment.
• Existing systems require enhancement for early disease prediction capabilities.

Purpose of the Study:

• To develop and evaluate an innovative single-lead ECG monitoring system integrated with machine learning for early cardiovascular disease prediction.
• To assess the performance of various machine learning algorithms in analyzing ECG data for anomaly detection and disease prediction.
• To establish a real-time data processing pipeline from a wearable device to a web-based interface.

Main Methods:

• Development of a wearable ECG system using ADS1298 and STM32L151xD microcontroller.
• Implementation of signal processing algorithms: band-pass filtering, K-means clustering, and PQRST analysis.
• Application of machine learning models including Isolation Forests for anomaly detection and comparative analysis of CNNs, SVM, XGBoost, and others for CVD prediction.

Main Results:

• The system successfully processed and analyzed ECG data in real-time.
• Isolation Forests demonstrated effectiveness in ECG anomaly detection.
• Convolutional Neural Networks (CNNs) achieved a high accuracy of 0.926 in predicting cardiovascular diseases, outperforming other tested methods.

Conclusions:

• The developed wearable ECG system, enhanced with machine learning, shows significant potential for early cardiovascular disease detection.
• CNNs are highly effective for processing ECG data and predicting CVD incidence.
• This integrated system offers a promising tool for remote cardiac monitoring and proactive healthcare.