Refined multiscale fuzzy entropy based on standard deviation for biomedical signal analysis

Area of Science:

• Biomedical Signal Processing
• Complexity Science
• Time Series Analysis

Background:

• Multiscale entropy (MSE) is widely used for biomedical time series complexity.
• Traditional MSE struggles with short signals and primarily uses the mean.
• Recent research explores alternative statistical moments like variance in MSE.

Purpose of the Study:

• Introduce refined composite multiscale fuzzy entropy using standard deviation (RCMFEσ) and mean (RCMFEμ).
• Quantify dynamical properties (spread and mean) across multiple time scales.
• Enhance complexity analysis for biomedical signals, particularly short or noisy ones.

Main Methods:

• Developed RCMFEσ and RCMFEμ algorithms.
• Utilized synthetic signals to compare RCMFE methods with traditional MSE.
• Applied RCMFEσ and RCMFEμ to magnetoencephalograms (Alzheimer's disease) and electroencephalograms (epilepsy).

Main Results:

• RCMFEσ and RCMFEμ demonstrated greater stability and reliability than standard MSE.
• Both RCMFEσ and RCMFEμ effectively captured distinct dynamical properties.
• RCMFEσ-based features yielded higher classification accuracies in disease-specific EEG/MEG data.

Conclusions:

• RCMFEσ and RCMFEμ provide robust alternatives for biomedical time series complexity analysis.
• The choice between RCMFEσ and RCMFEμ depends on the specific dynamical properties being investigated.
• These refined methods offer improved diagnostic potential in neurological disorders.