Predicting Anxiety Among Young Adults Using Machine Learning Algorithms

This study describes the process, procedure, and analysis used to quantify the amount of anxiety experienced by engineering students during the set of activities. The paper presents a comprehensive literature review, collection methods for anxiety detection, and emphasizes the advantages of portable EEG devices. The core methodology involves inducing specific anxiety types through structured activities, coupled with EEG data analysis followed by use of a machine learning algorithm towards classification of anxiety. The experimentation included data from nine participants aged 18-21 years. EMOTIV EPOC+ portable EEG machine was used to record the EEG from the participants during anxiety-triggering activities. The authors used statistical methods such as the Wilcoxon signed-rank test and Mann-Whitney U test to identify significant changes in EEG signals during anxiety-triggering activities compared to the relaxed state. The Self-Assessment Manikin(SAM)method is employed to categorize data into light, moderate, severe anxiety, and normal states. Machine learning algorithms, including Random Forest and XGBoost, demonstrate notable efficacy, achieving an accuracy of 72.27%. This research contributes valuable insights to mental health research and treatment, emphasizing the potential for personalized interventions based on neural and behavioral markers.