EEG-Based Brain Age Prediction

This study examines the potential of combining electroencephalography (EEG) with machinelearning to estimate brain age as a scalable, cost-effective alternative to MRI-based approaches.Using the Resting-state EEG data before and after cognitive activity across the adult lifespan and a5-year follow-up dataset, collected as part of the Dortmund Vital Study and publicly available viaOpenNeuro, we extracted biologically relevant EEG features, including frequency bandpower(delta, theta, alpha, beta, gamma), spectral entropy, Hjorth parameters, and relative power metrics.We compared two machine learning models: XGBoost for structured tabular learning and aMulti-Layer Perceptron (MLP) for non-linear feature interaction. Model performance wasevaluated using 5-fold cross-validation, with metrics including mean absolute error (MAE) andcoefficient of determination (R²). Results demonstrate the feasibility of EEG-based brain ageprediction, offering a path toward accessible cognitive health screening. The findings alsocontribute insight into which EEG feature types and modeling strategies hold the most promise forbiological age estimation.