How ihgumilar/hyperscanning2-redesign works
• EEG data pre-processing and analysis
• Eye Tracker data pre-processing (with ongoing analysis development)
• Questionnaire data processing and subsequent statistical analysis (such as ANCOVA)
Main components
• Data Sources & Ingestion:
– Raw EEG data for both baseline and experimental sessions (coming in separate folders).
– Raw Eye Tracker files (to be processed to separate baseline and experimental data).
– Questionnaire responses.
• Pre-processing Modules:
– For EEG:
• Module to separate baseline/experimental data.
• Module to combine pre and post data for each eye condition.
• Cleaning module with functions for epoching and removing artifacts.
– For Eye Tracker:
• Extraction, combining, and cleaning modules (note: one extraction script is deprecated).
– For Questionnaire:
• Data parsing and score calculation for SPGQ, Co-Presence, and other sub-scales.
• Analysis & Statistical Processing:
– EEG: Analysis modules that identify significant connections, apply permutation statistical tests, and calculate average scores.
– Eye Tracker: (Still under construction, but would eventually include analysis modules).
– Questionnaire: ANCOVA tests on questionnaire scores.
• Data Storage & HPC Integration:
– Specific folders leading to HPC storage paths for raw, pre-processed, and analyzed data.
• (Additional) VR/Unity Component:
– Mentioned in the README for questionnaire integration which represents further input integration via improvement in UNITY.
Relationships and data flow
• Raw data (from EEG, Eye Tracker, Questionnaire) flows into individual pre-processing modules.
• Pre-processed data then feeds into analysis modules where, for example, EEG data is partitioned in baseline versus experimental conditions, and subsequently processed for statistical significance.
• The outputs from these analysis pipelines then lead to further computations (e.g., ANCOVA) or final aggregated results.
• Notice that within EEG there are separate branches for each eye condition (averted, direct, natural) and for different phases (pre and post). These should be depicted as sub-components or branches within the EEG module.
Architectural patterns and design principles
• Modular Design: Each experimental dataset (EEG, Eye Tracker, Questionnaire) is handled by its own set of processing modules.
• Pipeline/Workflow Structure: The sequence of pre-processing steps and analyses is clearly defined (data separation, combination, cleaning, statistical analysis).
• Emphasis on reproducibility and traceability (noting commit IDs in the README suggests version control integration and code traceability).
• Scalability: The modules are designed to work with large raw datasets stored on HPC systems.
Relevant technologies and aspects
• Python (as the main programming language, indicated by .py files).
• Sphinx documentation (as seen in the docs directory within the “phd_codes” folder).
• HPC storage integration (shown by the mentioned storage paths).
• Specific functions like add_leading_zero, total_significant_connections, and others that symbolize utility modules.
• UNITY integration improvements for questionnaire inside VR (this might be represented as an external or interfaced module).