Aristotle University Of Thessaloniki (AUTH)
Μεταβλητότητα καρδιακής συχνότητας και μη γραμμική ανάλυση ως μέθοδοι καθορισμού βαρύτητας και πρόγνωσης σε βαρέως πάσχοντες που νοσηλεύονται σε πολυδύναμη ΜΕΘ
Abstract
dc:descriptionThe development of information technology and computer science during the last decades has made much easier the mathematical analysis and computation, something that was very difficult to perform in the past. The use of methological tools of other sciences in medicine has contributed to a change of view, concerning the nature of physiology and pathophysiology of different kind of diseases. The application of biosignal analysis in research and clinically oriented projects, concerning mainly the field of cardiology and neurosciences, was quite fruitful and meaningful as well. The parallel evolution of Intensive Care Units permitted the clinicians to create a huge data base from critically ill patients. The transformation of data to information first and then to knowledge, aiming at therapeutic interventions, remains a quite difficult task and demands the cooperation of other fields of science. From this cooperation until now, a new concept has emerged, concerning what is ‘normal’ or ‘abnormal’ in a complicated system with interrelated variables, such as the human body. The increased interrelation of sympathetic and parasympathetic system, at the level of cardiovascular control, is reflected to the heart rate variability of the ECG. At the same time, every organism is acting as a complex nonlinear system where both the periodicity, with loss of variability of ECG and the chaotic behavior constitute pathologic phenomena. Specially in the critical care setting, the transition of a patient to a septic state and multiple organ failure is accompanied by an intense periodicity within the signal, as it is evident from the use of different statistic metrics (approximate entropy-ApEn, α exponent). The purpose of our project was to study this kind of alterations in patients being treated in a multidisciplinary intensive care unit at G. Genimatas Hospital of Thessaloniki. We studied three cohorts of patients, one cohort of surgical patients without any past medical history (19 patients), another cohort of surgical patients with past medical history (19 patients) and finally a group of medical patients (15 patients). ECG data were retrieved from monitors and were analyzed with different kind of algorithms in a MATLAB environment. Heart rate variability, ratio of low to high frequency (LF/HF), ApEn and α2 exponent of patients signals were evaluated. ApEn is an index of periodicity of a signal, with low values indicating regularity and high values more variability. Exponent α2 is related to the non-linear dynamics of the system. An increase of its value is associated with a decrease in ApEn value and more regularity while a decrease is responsible of the inverse scenario. At the same time, every day SOFA score was assessed, as an index of severity of illness. The abovementioned variables of signal dynamics were correlated with different SOFA metrics, such as admission score, maximum, discharge score and ΔSOFA (SOFAadmission-SOFAmaximum) which is an index of morbidity of patients due to their ICU stay. Regression analysis with curve estimation revealed that in surgical patients of group 1, significant correlations came up between different ApEn values and minimum signal variance from one side and different SOFA metrics from the other side. In group 3 of medical patients, the only significant correlation was between the minimum LF/HF ratio and SOFA score while α2 exponent was correlated only with length of stay. Finally, in group 2 patients, who were a kind of combination of the other two groups, both ApEn and α2 exponent values were significantly correlated with different SOFA metrics (although r2 values were lower). However, the most significant correlations concerned the minimum LF/HF ratio and the admission variance. Finally, logistic regression analysis was performed in order to detect the possible influence of each measured variable separately, in the three groups of patients, upon mortality. Logistic regression analysis concluded that four statistically significant factors were affecting surgical patients’s mortality. The minimum log variance increases mortality up to 68%, ApEn up to 49%, and its minimum and admission values up to 71 and 52% respectively. In group 2 and 3 of patients the minimum log LF/HF is the only statistically significant parameter that increases mortality up to 62 and 46% respectively. In conclusion, minimum ApEn increases mortality of the three groups of patients by 61%.
Degree
thesis:*- Grantor dc:publisher
- Aristotle University Of Thessaloniki (AUTH)
- Year dc:date
- 2004
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Papaioannou, Vasilios
- Παπαϊωάννου, Βασίλειος
Subjects
dc:subject × 14Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/13941