The researchers address the challenge of predicting psychiatric hospitalizations through social media posts. They propose a framework that extracts time spans of self-reported psychiatric hospitalizations from social media data to build predictive models of psychiatric hospitalization. The study compares various feature sets and classification methods to obtain the best predictive model. The results show that social media data can be leveraged to predict acute psychiatric crises before they occur, with the best model achieving an F1 of .718 using 7 days of posts. This framework is crucial in collecting hospitalization data and can potentially save lives and improve outcomes for individuals with mental illness.
