EEG-Driven Age Prediction: Advancements in Machine Learning Models

This research article proposes a novel method for predicting human age using EEG signals and machine learning (ML) models. EEG signals provide valuable insights into brain activity and have the potential to serve as biomarkers for age prediction. The study collected a diverse dataset of EEG recordings from individuals of different ages from 30–50 years. With the random selection of signals from people of different ages in between 30–50 years from those diverse datasets, a novel dataset was created for this study. Preprocessing techniques such as ‘Max-min’ scaling and the ‘SelectKBest’ method of feature selection were applied to enhance the quality as well as the relevance of the novel dataset. Multiple ML algorithms, including Regressor models like Support Vector, XGBoost, K-Nearest Neighbors, Voting Regressor, Stacking Regressor, Ridge Regressor, Random Forest, and Lasso Regressor, were trained and evaluated on the preprocessed EEG dataset. The model's performance was assessed using various evaluation metrics that include Mean Absolute Error (MAE), Root Mean Square Error (RMSE), Mean Absolute Relative Error (MARE), coefficient of determination (R2), Nash-Sutcliffe Efficiency (NSE), and Mean Absolute Percentage Error (MAPE). The selected EEG features revealed specific spectral bands and temporal patterns strongly correlated with age, shedding light on the biological and cognitive factors associated with aging. The proposed framework offers a reliable and non-invasive approach for estimating human age using EEG signals. This research proves the efficacy of ML-based age prediction models and emphasizes the practical use of EEG-based age prediction models in advancing healthcare and biometric systems.