Validation-Aware Retrospective EEG Treatment-Response Modelling Using Chaotic Pattern of Prime Numbers Features:...

Area of Science:

• Neuroscience
• Computational Psychiatry
• Biomedical Engineering

Background:

• Predicting depression treatment response remains challenging.
• Electroencephalography (EEG) offers potential biomarkers for treatment outcomes.
• Existing EEG feature extraction methods may lack robustness and interpretability.

Purpose of the Study:

• To develop and validate a novel EEG framework for depression treatment-response modeling.
• To introduce Chaotic Pattern of Prime Numbers (CPPN) as a new EEG feature representation.
• To establish a rigorous validation hierarchy for assessing feature performance and model generalizability.

Main Methods:

• A validation-aware framework was developed using CPPN features.
• EEG data from SSRI and rTMS cohorts were analyzed.
• A seven-protocol validation hierarchy was employed, including segment-level and subject-level cross-validation.
• Feature selection methods like Normalization and NCA ranking were integrated.

Main Results:

• CPPN features demonstrated strong discriminative power, achieving high accuracies (e.g., 99.42%) in segment-level cross-validation.
• CPPN features significantly outperformed conventional EEG features (by up to 29.53 percentage points).
• Subject-wise validation yielded more conservative but still promising results (best LOSO accuracy 80.00%).

Conclusions:

• CPPN offers a compact, inspectable, and computationally accessible EEG feature representation.
• The proposed validation hierarchy provides transparent evaluation of performance across different validation depths.
• This methodological contribution offers a rigorous foundation for future EEG-based treatment-response studies.