Post-Traumatic Stress Disorder (PTSD) is a mental health condition triggered by experiencing or witnessing traumatic events. Symptoms of PTSD include intrusive thoughts, avoidance behaviors, negative alterations in cognition and mood, and heightened arousal and reactivity. These symptoms can severely impact an individual's daily functioning and quality of life. Refugees, who often face extreme stress and traumatic experiences, are particularly susceptible to PTSD. The high prevalence of PTSD among refugee populations demands effective screening and early intervention to mitigate long-term mental health consequences. Therefore, the primary objective of this study is to leverage machine learning algorithms to predict PTSD in individuals using data derived from the PTSD Checklist for DSM-5 (PCL-5) and sociodemographic information. By developing a predictive model, in R and using the Python and random forests, we aim to identify individuals at high risk of developing PTSD, according to specific factors interacting in the context, as well allowing for timely and targeted interventions.

Sociodemographic variables and symptoms (features) were collected in 77 survivors with refugee status admitted in Portugal, through sociodemographic questionnaire and the PCL-5, respectively. The predictive model based in each set of factors indicated area under the curve (Receiver Operating Characteristics) with moderate to high values (between >50 and <93) for validations trials with pool sensitivity variation between 33% and 70%. Specificity showed unbalanced scoring (false positives approx. 80%) for some clusters from PCL-5 considering certain variables introduced in model as potential predictors. Intrusive memories and cognitive and mood alterations were the clusters with highest predictive value to determine the ML model by integrating the following sociodemographic factors: date of entry in host country, academic background, household and monthly income. This model may inform and discern future interventions in refugees following trauma exposure with different features.