Code and data used for estimating the Rescaled Range and Hurst exponent values for Normal and Abnormal Heart Sound Signals. This project was done as my undergraduate reseach project (BSc. Mechanical Engineering, University of Ibadan)
This project was inspired by the Physionet 2016 Computing in Cardiology challenge, which was aimed at accurately classifying Heart Signals from Phonocardiograms (PCG) as Normal/Abnormal for Low-to-Middle-Income-Countries.
The results of this study were published in the Journal of Engineering Research and Reports. You can read the full paper here
Heart acoustics can be used as an early diagnostic tool, to quickly and accurately identify medical patients who may be at risk of unfavourable cardiovascular outcomes. However, at present, there remains no universally accepted standard or technique for detecting abnormalities using Phonocardiograms. To address this, a large database containing normal and abnormal heart sound recordings of patients were studied with the rescaled range technique and the Hurst exponent, a measure of persistence for non-linear dynamic data.
Using the Hurst exponent (H) as a benchmark, we compared the values of the Hurst exponent for normal heart sound recordings with that of the abnormal heart sound recordings. For this study, the recording length was limited to the first 10 seconds for all 578 distinct recordings, which were selected randomly from the database. Furthermore, two Hurst exponent values were obtained for each recording, by subjecting them to time intervals of 1 and 2 milliseconds respectively. The results from this study show that heart sound recordings are persistent (H > 0.5) for normal and abnormal heart sound recordings, with the normal recordings being slightly more persistent.
More information about the Physionet 2016 Challengeand the Database can be gotten here https://www.physionet.org/content/challenge-2016/1.0.0/