Accuracy evaluation of ECG waves detection and segmentation

Wearable wireless devices are vital in regulating lifestyles and facilitating telemedicine and home healthcare monitoring. Despite abundant low-cost sensor options flooding the market, only a handful have received validation and certification for clinical use. This study offers initial insights from a detailed evaluation of an electrocardiography (ECG) tool designed for patients' self-managed monitoring of cardiovascular diseases. The primary focus of this assessment is on examining the functionality of the MAX30003 integrated circuit and the essential characteristics embedded within its firmware. The study also investigates six cardiac conditions requiring thorough and continual ECG monitoring. To assess accuracy, the experimental setup utilizes a National Instruments USB6001, applying seven distinct signals-Typical, Hyper- and Hypo-calcemia, Hyper- and Hypo-kalaemia, Ventricular and Atrial Tachycardia-under three overlapping and non-overlapping noise sources: white noise, powerline noise, and their combination. Results from the accuracy evaluation demonstrate the integrated circuit's performance, revealing a normalized root mean square error lower than 10% in the worst-case scenario (signal with power line and noise) and an average of 2.5% in the best-case scenario (no noise). Similarly, repeatability and reproducibility are higher overall for TYP ECG. PLN is the noise that least affects repeatability, while white noise is the least attenuated one. Moreover, it achieves a 95% detection rate for P, R, and S peaks, 98% precise segmentation for the QRS complex, and 95% precision for P and S peaks under no-noise conditions. In noise conditions, precise segmentation rates are slightly lower, at 87% for the QRS complex and 91% for P and S-peaks. Preliminary results from healthy subjects align with those from systemic signals, indicating promising segmentation performance. These findings provide crucial insights into the potential effectiveness of the MAX30003 for accurate ECG monitoring across diverse cardiovascular scenarios.