Answers are grounded in this paper and cite their sources.
Background: Psychiatric comorbidities, particularly anxiety and depression, significantly impair quality of life and psychosocial functioning in patients with drug-resistant epilepsy (DRE). This study aimed to develop and evaluate interpretable machine learning (ML) models to predict post-treatment depressive and anxiety symptoms using exclusively pre-treatment baseline data. Methods: Baseline demographic, clinical, and psychological data from a cohort of DRE patients were analyzed. Eight ML algorithms were trained and evaluated using repeated nested cross-validation to predict post-treatment depressive and anxiety symptoms as separate outcomes. Model performance was assessed using the area under the receiver operating characteristic curve (AUC), Brier score, precision, recall, and F1-score. SHAP (SHapley Additive Explanations) analysis was performed for the Random Forest models to identify the key predictors for each outcome. Results: Five of the eight trained models achieved a mean AUC ≥ 0.80 for depressive symptoms, and seven for anxiety symptoms. Random Forest achieved the highest mean AUC for both depressive symptoms (0.838 ± 0.008) and anxiety symptoms (0.862 ± 0.011). SHAP analysis identified that baseline mood severity, treatment modality, and seizure-related variables as the primary predictors across both models. Conclusions: The findings indicate that post-treatment depressive and anxiety symptoms in DRE patients can be predicted before treatment using routine baseline data, highlighting the potential of baseline clinical information for early risk stratification in this high-risk population.
No takes yet. Share an insight, caveat, or question.
Xu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: