Análisis de la interacción cardíaca y respiratoria en pacientes con cardiomiopatía y pacientes en proceso de extubación
- Autores: Andrés Ricardo Arcentales Viteri
- Directores de la Tesis: Beatriz Giraldo Giraldo (dir. tes.)
- Lectura: En la Universitat Politècnica de Catalunya (UPC) ( España ) en 2015
- Idioma: español
- Tribunal Calificador de la Tesis: Abel Torres Cebrián (presid.), Agustín Gutiérrez Gálvez (secret.), Francesc Clarià (voc.)
- Materias:
- Enlaces
- Tesis en acceso abierto en: TDX
- Resumen
Study and evaluation of patients with heart failure related with ischemic and dilated cardiomyopathy is interesting for the clinical practice. Similarly, the analysis of the behavior of the respiratory system of patients undergoing extubation process is still research topic, and is particularly important to determine the optimal moment for the weaning process. In Both cases, the study of the cardiac and respiratory systems, and their interaction can contribute to identify new medical process, and increase the knowledge of the pathologies studied. The main goal of this research is the analysis and characterization of the behavior of the cardiovascular system, the respiratory system, and the cardiorespiratory interaction in patients with chronic heart failure, and patients assisted by mechanical ventilation on the processes of weaning trials. We propose the extraction of information that allows to enhance the knowledge of the physiological behavior, and improve the diagnosis and stratification of these patients. Cardiomyopathy is a disease of the cardiac muscle reducing the heart’s pumping ability. Patients with chronic heart failure (CHF) usually present changes in the respiratory behavior, with periodical changes in the tidal volume, known as periodic breathing (PB). In order to study chronic heart failure patients, we propose to analyse the cardiorespiratory interaction through characteristics extracted, in time and frequency domain, from the respiratory flow (FLW), respiratory volume (VOL), electrocardiogram ECG, and blood pressure (BP) signals. The study was performed in 50 patients with dilated cardiomyopathy (19 patients), and ischemic cardiomyopathy (31 patients). A periodic respiratory pattern is considered as an indicator of the severity of these disease. Using the respiratory flow envelope, we calculated the modulation index (M) associated to the periodic breathing. First part of the study considered CHF patients classified based on the modulation index. The study was performed considering 35 segments of signals non modulated (GN, M<10%), and 12 signal segments high modulated (GH, M>75%). Considering these time segments, we studied parameters extracted from the cardiac, respiratory and cardiorespiratory interaction related to the envelope of the respiratory flow signal. The main objective is to analyze differences on the cardiovascular behavior between non modulated and modulated patients. Studying respiratory signals, the main differences were found on the end expiratory lung volume (EELV). Considering the parameters extracted from the ECG, were obtained difference on the spectrum of the upward and downward slope of the QRS complex. Differences were also found on the heart rate variability. From the analysis of the magnitude squared coherence (MSC) between the series and the envelope the main differences were found in the frequency domain on the band of very low frequency. This result suggest that the periodic changes on the tidal volume are presented in the cardiac behavior. CHF patients were also studied through an unsupervised classification based on the K-means method. Time and frequency features extracted from the series were used to perform a 2 clusters classification. Based on these cluster were analyzed the clinical information. Additional, were study the modulation index on these clusters. Results suggest a strong correlation between feature extracted form ECG series and the modulation of the respiratory. Clusters calculated form BP series are related with the blood volume on the ventricle. Mechanical ventilation guarantee a correct alveolar ventilation in patients with respiratory failure. Weaning trial is the process of transfer the respiratory effort from the mechanical ventilator to the patient. This study investigated the contribution of spectral signals of heart rate variability (HRV) and respiratory flow, and their coherence to classifying patients on weaning process from mechanical ventilation. A total of 121 candidates for weaning, undergoing spontaneous breathing tests, were analyzed: 73 were successfully weaned (GS), 33 failed to maintain spontaneous breathing so were reconnected (GF), and 15 were extubated after the test but reintubated within 48 h (GR). The power spectral density and MSC of HRV and respiratory flow signals were estimated. Dimensionality reduction was performed using principal component analysis (PCA) over the spectral signals. Considering this new space formed by the PCA were calculated a fuzzy K-nearest neighbor classification. Best classification index, applying PCA and fKNN methods, were obtained considering the spectral signal of the MSC between HRV and FLW. The classifiers present a good balance between sensibility and specificity, and high accuracy of 92% comparing GS vs. GF, 86% classifying GS vs. GR, and 83% classifying GS vs. GFR. Reintubated patients is the most complex group for the analysis and classification, because at the beginning of the trial the behavior is similar to the success patients, but before 48 h the evolution of the respiratory pattern is more comparable to the failure patients. However, applying our method the index of accuracy, sensibility and specificity comparing GR against the other groups are above 80%. Spectral analysis of weaning trial patients were complemented with a nonlinear study based on the recurrence plot (RP) method. Additional, to the individual RP analysis of the HRV series, inspiratory time series (TI ), and respiratory time series (TTot), were performed the study of the cardiorespiratory interaction through the cross recurrence plot (CRP) and joint recurrence plot (JRP). One of the main issues for the application of the recurrence technique is the correct selection of the embedding dimension (?). We propose the application of the method based on the analysis of ROC curves comparing observational noise against the signals of the study. This method allow the selection of the optimal ? for each data series. Recurrence matrix where characterized based on the parameters extracted applying the recurrence quantification analysis (RQA). Comparing the different groups that the main differences was observed on the determinism and stationarity of the signals. Patients of the success group show higher values of determinism, suggesting that time series extracted form de GE patients during the weaning trials are more stationary comparing with the orders groups. Analyzing the result obtained from JRP, it concludes that patients from GS show a higher mutual recurrence between HRV and the time series (TI and (TTot. Based on the RQA parameters obtained for each time series extracted, was proposed a supervised classification applying support vector machine technique (SVM). The best classification obtained were abode the 80% in accuracy. The results obtained suggest that recurrence plot and especially joint recurrence plot techniques could improve the discrimination and characterization of patients on weaning trials