Enhanced cardiovascular diagnostics using wearable ECG and bioimpedance monitoring with LightGBM classifier

Cardiovascular diseases (CVDs) are the leading cause of death worldwide, necessitating advanced diagnostic and monitoring tools. Traditional methods of cardiac monitoring face challenges such as limited availability, high costs, and continuous physician oversight. Recent advancements in mobile health (mHealth) technologies, including wearable devices and mobile applications, offer promising solutions for continuous and real-time monitoring of vital signs such as ECG, bioimpedance, and physical activity. This study focuses on integrating these monitoring modalities to enhance the accuracy and reliability of cardiovascular diagnostics. Specifically, we explore the use of the MAX30001 device for precise ECG and bioimpedance measurements in wearable applications. Machine learning techniques, particularly LightGBM, are employed to classify cardiac conditions based on the collected data. The LightGBM classifier achieved a test set accuracy of 94.49 %, with precision, recall, and F1-scores above 0.95 for all classes. The model's performance was further validated through cross-validation (CV), yielding a 5-fold CV accuracy of 95.86 % and a 10-fold CV accuracy of 96.16 %. The ROC curve analysis showed excellent discriminatory ability with AUC values close to 1. These findings highlight the potential applications of advanced mHealth solutions in providing continuous, accurate, and real-time monitoring of cardiovascular health, which can lead to better patient management and outcomes through timely and informed interventions. • Analysis of ECG, bioimpedance, and physical activity data combination is promising. • LightGBM demonstrated high accuracy in classifying data combinations. • ECG and BioZ signal parameters demonstrate interconnections with physical activity.