EEG-Based Pain Detection and Personalized VR Therapy

Modern medicine today faces several challenges in improving patient care, particularly when it comes down to pain management. A lot of specialty areas, including anesthesia, oncology, palliative care, and intensive care, struggle in addressing the patient’s pain. Pain is a complex and subjective experience to all, making it inherently difficult to assess through conventional means.

Thus, this capstone project introduces a novel system that coalesces the use of EEG, machine learning, and VR to recognize the severity of a patient’s pain and provide a personalized VR experience for relief. An opensource dataset with labeled pain events was used to train LightGBM and XGBoost models, where preprocessing and feature extraction were followed. The training of these two models resulted in a high accuracy, reaching up to 97%, and were integrated with the VR system as a modular, real-time system.

When a pain level is detected by the machine learning model, the system triggers a VR environment, where a personalized avatar guides the user through calming visuals, interactive exercises, and supportive statements–all to alleviate the discomfort the patient is facing. Overall, this project offers a promising step toward adaptive and non-invasive solution for pain recognition and relief, with strong potential to impact current applications in medical therapies, hospitals, and at-home care for patients.