This study developed a machine learning model to classify suicidal thoughts in patients with major depressive disorder using unstructured psychiatric charts and brain MRI. The XGBoost model combined data from clinical notes (topic probabilities from 5 psychiatric symptom topics) and brain T1-weighted MRI (independent components-map weightings from 5 brain networks). The combined model had the highest area under the ROC curve (0.794) compared to clinical notes only (0.748) and brain MRI only (0.738) models. Results were consistent across performance metrics and external validation, suggesting that integrating neuroimaging and NLP variables improves suicidal thought evaluation.
Publisher
Translational Psychiatry
Published On
Jul 04, 2024
Authors
Dong Yun Lee, Gihwan Byeon, Narae Kim, Sang Joon Son, Rae Woong Park, Bumhee Park
Tags
machine learning
suicidal thoughts
depression
neuroimaging
psychiatry
XGBoost
clinical notes
Related Publications
Explore these studies to deepen your understanding of the subject.
