Research Article
EEG-Based PTSD Detection using Machine Learning Approaches
@INPROCEEDINGS{10.4108/eai.28-4-2025.2358071, author={Rekha Ravi and Janani Selvam and Yamunarani T and R. Prabu}, title={EEG-Based PTSD Detection using Machine Learning Approaches}, proceedings={Proceedings of the 4th International Conference on Information Technology, Civil Innovation, Science, and Management, ICITSM 2025, 28-29 April 2025, Tiruchengode, Tamil Nadu, India, Part II}, publisher={EAI}, proceedings_a={ICITSM PART II}, year={2025}, month={10}, keywords={post-traumatic stress disorder (ptsd) electroencephalography (eeg) machine learning (ml) support vector machine (svm) convolutional neural network (cnn) random forest (rf) least angle regression (lars) feature engineering}, doi={10.4108/eai.28-4-2025.2358071} }- Rekha Ravi
Janani Selvam
Yamunarani T
R. Prabu
Year: 2025
EEG-Based PTSD Detection using Machine Learning Approaches
ICITSM PART II
EAI
DOI: 10.4108/eai.28-4-2025.2358071
Abstract
Post-Traumatic Stress Disorder (PTSD) is a multifaceted mental health condition characterized by prolonged emotional distress, cognitive disturbances, and heightened stress responses following exposure to traumatic experiences. Traditional diagnostic approaches often rely on subjective clinical evaluations, which can sometimes lead to inconsistent or inaccurate diagnoses. To address these limitations, this study proposes leveraging electroencephalography (EEG) in conjunction with machine learning (ML) algorithms to enable a more objective and automated method for detecting PTSD. EEG, being a non-invasive and relatively affordable technique, provides real-time insights into brain function and has shown potential in revealing distinct neural patterns associated with PTSD—such as elevated delta and theta activity, reduced alpha power, and disruptions in frontal brain asymmetry. This research utilizes a variety of ML classifiers, including Support Vector Machine (SVM), Random Forest (RF), Convolutional Neural Network (CNN), and Least Angle Regression (LARS), to interpret EEG-derived features. Feature analysis incorporates elements from the spectral profile, time-domain patterns, and neural network connectivity metrics. To ensure robustness, the models undergo evaluation using cross-validation techniques like 10-fold and Leave-One-Out Cross-Validation (LOOCV).
