Predicting psychosis using social media and natural language processing: Potentials and pitfalls

Resumen

Of all people that develop a mental disorder, most are recognized late, and many not at all. The insufficiency of mental health infrastructure, which is driving these problems, is even more pronounced in developing countries. This underlines the pressing need for global and scalable solutions to the diagnosis and prediction of individuals’ mental health. In this perspective, we argue for a possible utility of social media (SM) derived data combined with natural language processing (NLP) for the monitoring of individual-level mental health on a global scale. Using psychosis-spectrum disorders (PSD) as an example, we demonstrate how freely expressed speech can be processed automatically to predict and classify PSD with high levels of accuracy. We further outline the current state of PSD classification from SM-derived data and discuss methodological considerations that are hampering progress in this field. Finally, we delve into potential pitfalls of such systems and provide insight into how these may be navigated.

Áreas y palabras clave

Cómo citar (APA 7)

Plank, L., Dere, E., & Zlomuzica, A. (2024). Predicting psychosis using social media and natural language processing: Potentials and pitfalls. https://doi.org/10.31234/osf.io/ghw53