Early Prediction of Paroxysmal Atrial Fibrillation using Wavelet Transform Methods

Abstract

Paroxysmal Atrial fibrillation is one of the most common complaints of heart disorders that occur as a result of random vibrations of the atria. PAF episode show a serious increase with age, and the next steps are more difficult especially for the elderly. So, diagnosing in the early stages of this disorder is very important for the PAF patients to stop the progression of the disease and to improve the quality of life. For his reason, in this studyitisaimedtobedetectedwhichin5minutesbeforethePAF episodes. The 30-minute data is divided into 10 parts in 5 minutes with 50% overlap. For each part, wavelet transform methods and wavelet entropy are calculated over heart rate variability data. Using these measurements, it is determined whether there is a statistically significant difference between the parts and the early detection performance of PAF was obtained using the k-nearest neighbors classifier. As a result, PAF episode can be statistically distinguished before it occurs and it is determined that the k-nn classifier has about 72% performance 12.5 minutes earlier than a PAF episode.