Assessment of risk level of cardiac pathologies by cardiac signal processing analysis

Abstract

Cardiovascular diseases are the leading cause of death worldwide. In this way, it is of high relevance to stratify high and low risk of developing sudden cardiac death (SCD) in those patients that present a cardiac pathology, and to predict this fatal development. The extraction of quantitative markers from the electrocardiogram (ECG) time-series data is nowadays a difficult but promising task. The signal processing methodologies can be applied for this purpose. In this way, the main objective of this project was to characterise differences between cardiac pathologies, discriminating between high and low risk of sudden cardiac death by using different Poincaré plot measures (non-linear dynamics analysis). Finally, it was also pursued to find an individual predictor from each of the Poincaré plot measures. Patient position, tilt versus supine, was analysed from RR beat series, showing a higher complexity in supine position (SD1, SD2, Ratio, Area). Tilt position did not present significant differences between low risk and high risk of suffering sudden cardiac death. Supine position, presented significant differences from SD2 parameter with a higher complexity in sudden cardiac death group. Finally, analysing RR and QT heart-beat series behaviour obtained from 24h-ECG Holter recordings by SD1, SD2, Ratio and Area parameters, it can be stated that SCD patients presented less complexity behaviour of their cardiac series recordings. Ischemic patient group (compared with non-ischemic), showed different responses during day and night, with a more complex behaviour during day period of the cardiac series. Furthermore, all RR series presented more predictability than QT series. Lastly, this analysis showed that QT beat series presented a decreasing slope from day to night, mainly in SD1, SD2 and Area, maybe due to the intrinsic information from the beginning of the depolarisation to the end of QT-wave repolarisation.