Adaptive Virtual Reality Exposure Therapy Using AI-Driven Anxiety Prediction from Physiological and Behavioral Data

Abstract

All mental health disorders; anxiety disorders rank as the most common ones. They affect the cognitive functions of an individual as well as their ability to socialize or live in a quality manner. Exposure Therapy happens to be one of the most effective methods aimed at treating fears, phobias, and traumatic memories; however, the traditional approach to exposure therapy and VRET is unable to modify the exposure intensity in accordance with the situation that is occurring with the patient, thus making this method ineffective. This article suggests a solution to the problem by creating AVRET, or Adaptive Virtual Reality Therapy, which will provide the opportunity to predict the anxiety of the patient in real-time through their physiological (HR, HRV, GSR) and behavioral (eye-tracking, blink rate, pupil dilation) indicators. The suggested system uses an LSTM deep learning model to evaluate the amount of anxiety and change the scenario, distance to objects, sounds, or level of difficulties based on the current anxiety evaluation.