A study on standard and atrial lead system for improved screening of P-wave using random forest classifier

The challenging aspect of diagnosing atrial arrhythmias is the accurate detection of P-wave. P-wave visibility helps in better clinical decision-making and is essential in detecting atrial arrhythmias. Improved screening of P-wave requires improved recording or signal acquisition methods. In this study, an Atrial Lead System (ALS) is developed for improved recording of P-wave amplitude in sinus rhythm volunteers. The ALS is compared with the existing Standard Limb Lead (SLL) system for significant changes in signal strength in bipolar and augmented unipolar leads. The study developed a Random Forest (RF) classification model to classify SLL and ALS ECG. The RF model exhibited a 10-fold cross-validation accuracy of 0.92. The sensitivity, precision, and F1 score of the RF model are 0.96, 0.91, and 0.92, respectively. The developed model predicts the test data with an accuracy of 0.89 and a precision of 0.95 for ALS than SLL (0.85). The classification accuracy obtained is mainly due to a greater difference in P-wave amplitude (μV) between the leads. This difference is due to improved recording of P-wave in ALS than SLL system. The study concludes that the RF model automatically detects ALS for better visualization of P-wave.