This study investigates the use of Natural Language Processing (NLP) to predict recurrent chat contact in a 24/7 German crisis service for youth. Using approximately 800,000 messages from 19,000 users, an XGBoost classifier achieved an AUROC of 0.68 in predicting recontact within six months. Explainability analysis revealed that younger age, female gender, and mentions of self-harm and suicidal thoughts were associated with a higher likelihood of recontact. The findings suggest NLP's potential for personalized care in chat-based hotlines.
Publisher
npj Digital Medicine
Published On
May 18, 2024
Authors
Silvan Hornstein, Jonas Scharfenberger, Ulrike Lueken, Richard Wundrack, Kevin Hilbert
Tags
Natural Language Processing
NLP
crisis service
youth care
predictive analysis
self-harm
mental health
Related Publications
Explore these studies to deepen your understanding of the subject.
