Optimal Inter-Session Intervals in Neurofeedback Training: A Randomized Trial of Retention and Individual Response Patterns in Elite Judo Athletes

This repository contains de-identified data and analysis code from a randomized controlled trial in elite judo athletes comparing two neurofeedback (NFB) session spacings (48 h vs 72 h). The study evaluates (i) EEG dynamics focused on the Frontal Alpha Index (FAI) and individual alpha frequency (IAF), (ii) lower-limb strength performance across multiple relative loads (%1RM), and (iii) short-term retention of training effects. The dataset supports analyses of group-level effects, heterogeneity of individual responses (responder phenotypes), and growth/retention modeling. Keywords: neurofeedback, EEG, alpha, FAI, IAF, motor performance, retention, training spacing, elite athletes, randomized trial. Contents /data/ EEG_timeseries_F3_F4_IAF.csv — session-wise EEG summaries (F3/F4 power around IAF), FAI (log-ratio), session indices, and pre/post/retention flags. Strength_Squat_35_55_70_85_100.csv — strength outcomes (e.g., repetitions or derived performance metrics) at 35–100% 1RM measured pre/post within sessions. Participants_Metadata.csv — pseudonymous IDs, group allocation (48 h / 72 h / control), training history, and basic anthropometrics. Retention_Assessments.csv — retention measurements collected ~48 h or ~72 h after the last NFB exposure (EEG and strength). /code/ 01_preprocessing_EEG.m — EEGLAB preprocessing pipeline. 02_PSD_FAI_calc.m — power spectral density estimation and FAI computation. 03_stats_LMM.R — mixed-effects models (group × session/time), planned contrasts, multiplicity control. 04_growth_models.R — nonlinear growth/learning trajectories with model diagnostics. 05_retention_models.R — decay/retention modeling and sensitivity analyses. 06_figures.R — figure generation scripts (publication-grade output). /docs/ README_variables.pdf — variable definitions, coding, and units. DataDictionary.xlsx — machine-readable data dictionary. CONSORT_flow.pdf — participant flow diagram. Supplementary_Tables.xlsx — supplementary descriptive and model outputs. Methods Participants. Elite judo athletes meeting high-performance training criteria. All data are de-identified and pseudonymized. Design. Randomized allocation to two NFB spacing conditions (48 h vs 72 h) with an active or usual-practice control. Up to 15 sessions per participant; pre/post assessments embedded within sessions; terminal retention assessment after the final exposure. EEG Protocol. Eyes-open/eyes-closed blocks per protocol; F3/F4 channels summarized around IAF; FAI computed as log-ratio of right/left frontal alpha power. Signal quality checks include artifact rejection and standardized PSD procedures. Strength Testing. Lower-limb strength/performance assessed at multiple relative loads (%1RM). Standardized warm-up and familiarization; consistent testing order across sessions. Statistical Analysis. Linear mixed-effects models for repeated measures; growth curves for within-participant learning; retention/decay models post-training; responder phenotyping; multiple-comparison adjustments (e.g., Holm/FDR). Model assumptions and diagnostics reported in scripts. Variables Identifiers: participant_id (pseudonym), group (48h / 72h / control), session (1–15), timepoint (pre / post / retention). EEG: IAF (Hz), alpha_power_F3, alpha_power_F4 (µV² within IAF±band), FAI = log(alpha_F4) − log(alpha_F3). Strength: squat_metric_%1RM for 35/55/70/85/100 (see DataDictionary for exact units/definitions). Compliance & Safety: adherence (%), adverse_event (0/1) with optional narrative field. A complete, canonical specification is provided in DataDictionary.xlsx. Quality Control EEG preprocessing relies on reproducible EEGLAB/Matlab routines with fixed parameter seeds where applicable. Outlier handling and exclusion rules (e.g., excessive artifacts, protocol deviations) are documented in the analysis scripts. All inferential outputs include model diagnostics; sensitivity analyses are provided for key assumptions.