What's in a trauma? Using machine learning to unpack what makes an event traumatic

Abstract

What differentiates a trauma from an event that is merely upsetting? Wildly different definitions of trauma have been used across various settings. Yet there is a dearth of empirical work examining the features of events that individuals use to define an event as a 'trauma'. First, a group of qualitative coders classified features (e.g., actual physical injury, loss of possessions) of 600 event descriptions (e.g., "was verbally harassed by a boss", "watched a video of an adult being shot and killed"). Next, across two studies, machine learning was used to predict whether individuals rated event descriptions as 'trauma' or 'traumatic' in over 100,000 judgment tasks. In Study 1, examining continuous ratings, a cross-validated LASSO regression with interaction terms provided the best out-of-sample predictions (r 2 = 0.76), outperforming ridge regression, support vector regression, and linear regression. In Study 2, using binary judgments, a random forest model accurately predicted out-of-sample individual responses (AUC = 0.96), outperforming a neural network and an AdaBoost ensemble classifier. The most important event features across the two studies were actual death, threat of death, and the presence of a human perpetrator. The most important human features in predicting judgments were political orientation and gender.

The Contributions of Pandemic Severity, Government Stringency, Cultural Values and Internet Usage to Post-traumatic Stress Disorder During the COVID-19 Pandemic: An Analysis of Data From 35 Countries

“…While some authors have argued for a rigorous definition which would include only severe events (involving personal or occupational exposure to death or the risk of death) as “traumatic stress”, others have suggested considering “pandemic exposure”, or events such as being placed in quarantine or subjected to movement restrictions, as traumatic events per se (Norrholm et al, 2021 ). This debate is an extension of ongoing discussions of whether the spectrum of “traumatic events” is being unduly broadened (Jones, 2021 ). The current results, though subject to certain important limitations, suggest that the “narrower” definition of traumatic stress may be accurate, as the case-fatality ratio is a reflection of the risk of death in an infected individual, as opposed to measures such as prevalence (which includes mild and asymptomatic cases).…”

Understanding trauma as contextualized adverse life events that threaten the individual: Commentary in response to Marx et al. (2024)

“…Here, I want to highlight a crucial feature that makes an event traumatic for an individual: threat. In a study investigating 600 events, the most important features found to distinguish upsetting versus traumatic events were actual death, the threat of death, and the presence of a human perpetrator (Jones, 2021). Particularly, perceived threat is a key feature responsible for driving an individual's intrusion symptoms—a unique PTSD symptom—following trauma (Kube et al., 2020), emphasizing a subjective component in the perception, appraisal, and response to life events.…”

Psychological Response Data on the Traumatic Nature of 600 Written Events

“…Researchers could examine whether ratings differ across different demographic groups or prior experiences with trauma and mental health. Researchers could use text analysis or qualitative coding to separate the items into different types of trauma (e.g., "violence", "sexual abuse", "natural disaster") or different types of exposure (e.g., "experienced directly", "witnessed", "learned about") and examine whether differences emerge between ratings on these categories [8].…”