This work proposes a method to characterize the respiratory pattern of patients with chronic heart failure (CHF) to determine non-periodic breathing (nPB), periodic breathing (PB) and Cheyne-Stokes respiration (CSR) through non-linear, symbolic analysis of biological signals. A total of 43 patients were examined for their cardiorespiratory profiles, their ECG and respiratory pattern signals were processed, analyzed and studied for parameters that could be of potential use in clinical decision making, specifically in patient classification. Patients in the study were characterized through their cardiorespiratory signals, applying joint symbolic dynamics (JSD) analysis to cardiac beat and respiratory interval durations. The most statistically significant parameters across all groups were identified through a Kruskal-Wallis two tailed test (α = 0.05) and a linear discriminant analysis (LDA) classification method based on such parameters was developed. The best result achieved with this classification method uses 10 features to discriminate patients with a 97.67% Accuracy (Acc). The best features to discriminate among groups are related to cardiorespiratory interaction rather than just respiration patterns alone. Results further support the idea that abnormal breathing patterns derive from physiological abnormalities in chronic heart failure.

This work proposes a method to characterize the respiratory pattern of patients with chronic heart failure (CHF) to determine non-periodic breathing (nPB), periodic breathing (PB) and Cheyne-Stokes respiration (CSR) through non-linear, symbolic analysis of biological signals. A total of 43 patients were examined for their cardiorespiratory profiles, their ECG and respiratory pattern signals were processed, analyzed and studied for parameters that could be of potential use in clinical decision making, specifically in patient classification. Patients in the study were characterized through their cardiorespiratory signals, applying joint symbolic dynamics (JSD) analysis to cardiac beat and respiratory interval durations. The most statistically significant parameters across all groups were identified through a Kruskal-Wallis two tailed test (α = 0.05) and a linear discriminant analysis (LDA) classification method based on such parameters was developed. The best result achieved with this classification method uses 10 features to discriminate patients with a 97.67% Accuracy (Acc). The best features to discriminate among groups are related to cardiorespiratory interaction rather than just respiration patterns alone. Results further support the idea that abnormal breathing patterns derive from physiological abnormalities in chronic heart failure.