Assess the effect of artifact denoising on entropy biomarkers

Investigate the robustness of entropy-based flow biomarkers under different EEG artifact-removal strategies and compare feature stability before and after preprocessing, because wavelet-based hard-threshold denoising may distort signal-complexity characteristics used for entropy estimation.

Background

The study uses wavelet-based hard-threshold denoising to suppress ocular artifacts before extracting slope, distribution, spectral, and cross-channel entropy features from consumer-grade EEG recordings. Because entropy measures quantify signal complexity, preprocessing may alter the properties on which the biomarkers depend.

The authors acknowledge that they did not explicitly quantify the effect of denoising on the extracted entropy features. Establishing feature stability across artifact-removal methods and before versus after preprocessing would clarify whether the reported flow-state classification performance reflects neural information, preprocessing-induced changes, or a combination of both.

References

Lastly, although wavelet-based hard-threshold denoising is widely used for suppressing ocular artifacts in EEG recordings, it may also alter signal complexity characteristics that contribute to entropy estimation. Because the present study did not explicitly quantify the effect of denoising on the extracted entropy features, it cannot be excluded that some feature distortion occurred. Future work should systematically evaluate the robustness of entropy-based flow biomarkers under different artifact-removal strategies and compare feature stability before and after preprocessing.

— Practical flow state detection: Entropy-based EEG classification from portable EEG headbands  (2609.19737 - Beiramvand et al., 17 Sep 2026) in Section 5, Discussions and conclusions