Estimating the QT Interval from the Features Analysis of BCG and PCG signals

QT interval has clinical relevance since it is linked to ventricular arrhythmias and sudden death. Some drugs, systemic diseases, and genetics can modify it. Specialists usually obtain it manually from digital or printed electrocardiogram (ECG) recordings, which is a cumbersome task. For this reason, some algorithms have been proposed to make this diagnostic procedure efficient, faster, and cheaper. However, these approaches have limitations because a huge quantity of data and 12-lead ECG are required for classification methods. In this work, it is proposed to use other physiological signals with feasible acquisition procedures such as ballistocardiogram (BCG) and phonocardiogram (PCG) to extract information that could be related to QT interval or that could help to detect the T-wave end. This feature is a challenging one due to masking by U wave or by drifts in the isoelectric line. Two intervals were proposed here to correlate them with the QT interval:
$Q$
wave from ECG to K wave from BCG and H wave from BCG to S2 from PCG. The correlation coefficient between QT and Q-K intervals was 0.569, while the QT and H-S2 correlation was 0.653. Additionally, R2 values of 0.32 and 0.42 for the relationship between (QT, Q-K) and (QT, H-S2) were found. Bland Altman plots compared the proposed intervals with the QT interval where the 95% of the data fell within the agreement limits. These results are promising since both signals, BCG and PCG, could be used in a non-clinical setting in an easy way to help in the detection of QT interval. Clinical Relevance– QT prolongation is related to cardiovascular diseases, then the efforts to detect it effectively can have a health impact. BCG and PCG signals could provide additional information to identify the T-wave end and the overall QT interval.