| Abstract [eng] |
Atrial fibrillation is rapidly growing problem of aging world population. To make matters worse, atrial fibrillation is usually asymptomatic. Because of this it is estimated, that up to 2% of all population may have atrial fibrillation. Atrial fibrillation is progressive disease and if it is diagnosed in a late stage, its treatment becomes complicated. Therefore, it is of utmost importance to select people who has the greatest risk of atrial fibrillation. In Gladstone et. al. article, published in 2015, it was shown that number of atrial premature beats in twenty-four hours is related to the risk of atrial fibrillation. Atrial premature beats usually are detected using electrocardiogram, however this method is uncomfortable for long term monitoring. Therefore, it was deiced to investigate idea of detecting atrial premature beats using only photoplethysmogram. During the investigation a method for detecting atrial premature beats in photoplethysmogram was created. Method was tested using Physionet MIMIC II data base and achieved 0.78 Matthews correlation coefficient. Furthermore, method showed 99% specificity when tested on signals acquired from healthy volunteers. Therefore, it was proved than created method is suitable for detecting atrial premature beats and can be installed in a small wearable device with plethysmography sensor. |
