ECG features

Provides standard linear time-domain, linear frequency-domain, and non-linear ECG processing functions (Support vector machine-based arrhythmia classification using reduced features of heart rate variability signal 2.2.2 https://pdfs.semanticscholar.org/0c5d/2c9a7540dd3ee6f708e3671d8c9352c2ff8b.pd ). All features are based on the R-R intervals.

The functions are designed to work with records from physionet.org. Records from physionet.org can be downloaded and saved using retrieve_physio_files.py.

Linear time-domain functions

• Mean.
• Root mean square successive difference (RMSSD).
• Standard deviation between normal-normal (R-R) intervals (SDNN).
• Standard deviation between successive differences (SDSD).
• Probability successive normal-normal (R-R) intervals differ by greater than t (standard t = 50, 10, or 5) (pNN).

Linear frequency-domain functions

• Power spectral analysis (PSA). Ratio of the low-frequency (LF) and high-frequency (HF) bands.

Non-linear

• Cardiac-Sympathetic index (CSI).
• Approximate entropy (ApEn).
• Spectral entropy (SpEn).
• Largest lyapunov exponent (LLE).
• Detrended fluctuation analysis (DFA).
• Sequential trend analysis (STA).