Evaluation of the analytic representation of long-record ECG and its HRV signals for congestive heart failure classification

Differential diagnosis of cardiac diseases is considered a real problem in cardiology. Moreover congestive heart disease [CHF] is one of the most life-threatening where it is characterized by neurologic complications, and decreased pulmonary flow. Analysis of long-record ECG trace and/or the extracted HRV signal need to consider the presence of non-stationary. In this work, Hilbert transform is applied to get the analytic representation of these signals. Instantaneous amplitude (envelop); phase; and frequency were calculated. K-means algorithm was applied on these outputs to classify CHF. Classification results were promising with ECG (92.1%) more than HRV (75.85).