Shiba K, Daoud A, Kino S, Nishi D, Kondo K, and Kawachi I. (2022), Uncovering heterogeneous associations of disaster-related traumatic experiences with subsequent mental health problems: A machine learning approach. Psychiatry Clin. Neurosci., 76: 97-105. https://doi-org.dartmouth.idm.oclc.org/10.1111/pcn.13322
Researchers investigated the heterogeneous effects of disaster-related traumatic experiences on post-disaster mental health problems, using a new machine learning approach. Data was derived from a prospective cohort study of Japanese older adults (65 and older) in an area severely affected by the Great East Japan Earthquake of 2011. Baseline data were from 7 months before the earthquake (N=4,957 participants) and two follow-ups were conducted 2.5 and 5.5 years after the earthquake (n=3,567 and n=2,781 respectively). Disaster-related traumatic experiences were defined as home loss and loss of loved ones due to the disaster. Depressive symptoms and posttraumatic stress symptoms were assessed at the two follow-up time points. Researchers applied a novel machine learning approach called the generalized random forest algorithm to estimate the conditional average treatment effects of the disaster damages on mental health outcomes. Results showed significant heterogeneity in the impact of disaster damages across individuals, with unique patterns in characteristics of individuals who were more severely impacted. As an example, the most vulnerable group tended to be from lower socioeconomic status with preexisting depressive symptoms. The study demonstrates that this machine learning method can identify heterogeneity in mental health problems among respondents following a disaster event. Analyzing such heterogeneity may be beneficial in designing future post-disaster mental health interventions.