Propriedades computacionais e físicas de grandes conjuntos de dados cardíacos humanos

Abstract

Since the beginning of humanity, when our hominines ancestors have learned how to create spears and manipulate fire, we have to deal with data flow and interpretation of it, as well as in several other areas, such as climate analysis in Meteorology, or with heart disease diagnosis with the evaluations of eletrocardiograms (ECG’s) and because of so many data checking we become able to comprehend even more the behavior of nature around us and consequently we raised our own longevity, simply for performing an efficient data checking. However, in such ECG’s analysis only are considered the principal data: The peaks and the spacing between values. Suspecting that the ignored small data (only considered noise until now) could indeed mean something that we didn’t see before (a hidden chaos), so we started this research and studied since the basics of statistic (mean, variance and correlation) until advanced filtration techniques, like Wavelets and Bézier (this one was left behind for being inefficient for our objectives), so we did the application of these in temporal series of biological signal from MIT Database using the MATLAB tool to filter (2.8) the noises and analyse the data once neglected. In the end, we achieved the filtration theoretically provided by the Wavelet filter (2.8) when we applied it on the biological signals from MIT-BIH Arrhythmia Database, with well defined peaks and with low noise in all detailed levels (section 3.4), in addition to achieve results that shows the proximity between computationally achieved data and the predicted peaks counting given by section (3.3). So, our expectation from now on is to keep going our study until creating an analytical detection method for every biological system in which it could be created universal numbers of easy diagnosis for diseases prevention (not only cardiovascular ones) from the previously ignored data, which it is the whole initial objective.