Prediction of Recovery from Traumatic Brain Injury with EEG Power Spectrum in Combination of Independent Component Analysis and RUSBoost Model

The computational electroencephalogram (EEG) is recently garnering significant attention in examining whether the quantitative EEG (qEEG) features can be used as new predictors for prediction of recovery moderate traumatic brain injury (TBI). However, brain’s recorded electrical activity has always been contaminated with artifacts, which turn further impede subsequent processing steps. As a result, it crucial to devise strategy meticulously flagging and extracting clean data retrieve high-quality discriminative successful model development. This work proposed use multiple artifact rejection algorithms (MARA), an independent component analysis (ICA)-based algorithm, eliminate artifacts automatically, explored their effects on predictive performance random undersampling boosting (RUSBoost) model. Continuous were acquired using 64 electrodes from 27 TBI patients at four weeks one-year post-accident. MARA incorporates removal stage based ICA prior RUSBoost, SVM, DT, k-NN classification. area under curve (AUC) RUSBoost was higher absolute power spectral density (PSD) AUCδ = 0.75, AUC α 0.73 AUCθ 0.71 bands than k-NN. provided good generalization