Incremental Value of EEG Beyond EMG for Participant-Independent Forecasting of Bilateral Lower-Limb Joint Angles: Reproducibility Package

<p dir="ltr"><b>Reproducibility package for “Incremental Value of EEG Beyond EMG for Participant-Independent Forecasting of Bilateral Lower-Limb Joint Angles.”</b></p><p dir="ltr">This repository contains the code, frozen derived inputs, saved model parameters, out-of-fold predictions, full-precision result tables, figure-generation materials, and verification files supporting a participant-independent analysis of multimodal gait data.</p><p dir="ltr">The study evaluates whether scalp electroencephalography (EEG) provides incremental predictive value beyond electromyography (EMG) for forecasting six bilateral lower-limb joint angles.

The analysis includes 57 participants, 163 treadmill trials, and 145,885 matched prediction rows derived from the public PhysioNet Multimodal Gait Dataset. Model evaluation uses participant-disjoint nested cross-validation and participant-level statistical inference.</p><p dir="ltr">The reproducibility package corresponds to analysis/software version 2.0.0 (causal-native). Its primary reproduction entry point is the frozen derived feature and target data.