Reading the Mind’s Eye: Detecting Trauma-Vulnerability in Individuals by Analyzing Attention Through Eye-Tracking

In this work, we propose a novel approach for detecting and predicting trauma vulnerability in individuals by monitoring their attention to threat images using eye-tracking. We utilized a combination of statistical analysis, machine learning models and cognitive psychological theories. To do so, initially, an online survey was conducted with 183 participants to gather information about individuals’ previous experiences with trauma. Based on the survey results, an image set was prepared, consisting of four categories: positive, neutral, general threat and trauma-relevant threat. 39 volunteers, categorized into trauma-exposed and non-trauma-exposed groups based on clinical diagnosis and previous treatment, then participated in an eye-tracking experiment, where they were presented with 10 slides containing 4 images from each category; each slide lasting for 6 s. Eye gaze patterns were recorded and processed during this period, which was used to extract different features that were later used for the machine learning models to train on. Our findings revealed that trauma-exposed group showed more attentional bias towards the specific trauma-relevant threat images than the non-trauma-exposed group. Results from training the models suggest that it is highly likely that predicting trauma vulnerability in individuals is possible and this demands further research.